NTIA Report TR-06-444

Effects of RF Interference
on Radar Receivers

Frank H. Sanders
Robert L. Sole
Brent L. Bedford
David Franc
Timothy Pawlowitz

U.S. DEPARTMENT OF COMMERCE

Carlos M. Gutierrez, Secretary
John M. R. Kneuer, Acting Assistant Secretary
for Communications and Information

September 2006
 ACKNOWLEDGEMENT

The body of work presented in this report is the result of a long-term, dedicated, and cooperative
work effort. It would have been impossible to accomplish without critical contributions from the
following governments, organizations, and individuals: the Administration of the United
Kingdom (including critically important guidance and assistance from Mr. Peter Griffith; Mr.
Kim Fisher of Maritime and Coast Guard Agency; and the staff of the Fraser Range facility at
Portsmouth); the Sperry Marine radar company; the Kelvin Hughes radar company; the Federal
Aviation Administration (including critical assistance from Mr. Greg Barker and Mr. Mitch
Hughes and the rest of the staff of the long range radars of the FAA Technical Center in
Oklahoma City, OK; the staff of the airport surveillance radars at that facility; Mr. Michael
Biggs; and the FAA radar staff at Kirksville, MO); the U.S. Coast Guard radar staff at Curtis
Bay, MD, who assisted us well beyond ordinary limits in their already-busy schedule; the
National Weather Service (including Mr. Dan Friedman and the radar staff at Norman, OK); the
US Air Force, including especially Darrell McFarland and Ricardo Medilavilla of the 84th Radar
Evaluation Squadron (RADES) from Hill AFB, UT; staff of the NTIA Office of Spectrum
Management (OSM), especially Mr. Robert Hinkle and Mr. Mike Doolan, and also Matt Barlow
and Larry Brunson; and finally Ms. Kristen Novik of the NTIA/ITS laboratory, who drafted the
technical drawings in this report. All photographic images in this report were taken by one of the
authors (Sanders).

DISCLAIMER

Certain commercial equipment, instruments, or materials are identified in this report to specify
the technical aspects of the reported results. In no case does such identification imply
recommendation or endorsement by the National Telecommunications and Information
Administration, nor does it imply that the material or equipment identified is necessarily the best
available for the purpose.

iii
 CONTENTS
Page

FIGURES ................................................................................................................................. viii

TABLES................................................................................................................................... xiv

ABBREVIATIONS/ACRONYMS.......................................................................................... xvi

EXECUTIVE SUMMARY...................................................................................................... xix

1 RADAR FUNDAMENTALS RELATED TO INTERFERENCE TESTING AND

MEASUREMENTS ........................................................................................................ 1
1.1 Introduction ............................................................................................................ 1
1.2 Noise Versus Interference ...................................................................................... 2
1.3 Detection of Radar Targets .................................................................................... 4
1.4 Radar Receiver Inherent Noise .............................................................................. 6
1.5 Minimum Detectable Signal .................................................................................. 8
1.6 Increase in Receiver Noise Figure as a Function of Interference Power Level ..... 8
1.7 Detection of Radar Targets When Their Levels Fluctuate..................................... 9
1.8 Criteria for Radar Receiver Interference Thresholds ............................................. 10
1.9 Radar Detector Characteristics............................................................................... 11
1.10 Radar Features for Detection of Targets in Clutter and Interference..................... 14

2 SETTING CONDITIONS FOR INTERFERENCE MEASUREMENTS ON RADAR

RECEIVERS ....................................................................................................................... 22
2.1 Interference Coupling Technique........................................................................... 22
2.2 Calibration of Interference Levels ......................................................................... 24
2.3 Injection of Desired Targets................................................................................... 29
2.4 Use of Fluctuating Versus Non-Fluctuating Targets in Radar Testing.................. 32
2.5 Target Identification and Counting During Tests and Measurements ................... 36
2.6 Radar Receiver Parameter Settings........................................................................ 37

3 INTERFERENCE MEASUREMENTS ON LONG-RANGE AIR SEARCH

RADARS ....................................................................................................................... 39
3.1 Introduction ............................................................................................................ 39
3.2 Description of Long Range Radar 1....................................................................... 39
3.3 Test Approach for Long Range Radar 1 ................................................................ 41
3.4 Undesired Signals in Long Range Radar 1 ............................................................ 43
3.5 Test Procedures on Long Range Radar 1............................................................... 44
3.6 Results from Interference Tests on Long Range Radar 1 ...................................... 45
3.7 Description of Long Range Radar 2....................................................................... 49
3.8 Test Approach for Long Range Radar 2 ................................................................ 51
3.9 Undesired Signals in Long Range Radar 2 ............................................................ 53

v
 3.10 Test Procedures on Long Range Radar 2............................................................... 54
3.11 Results of Tests on Long Range Radar 2 ............................................................... 55
3.12 Summary of Interference Effects on Long Range Radars ..................................... 56

4 INTERFERENCE MEASUREMENTS ON A SHORT-RANGE AIR SEARCH

RADAR ........................................................................................................................... 59
4.1 Introduction ............................................................................................................ 59
4.2 Short Range Air Search Radar Technical Characteristics...................................... 59
4.3 Interference Procedures and Methods for the Short Range Air Search Radar....... 64
4.4 Test Results for the Short Range Air Search Radar............................................... 66
4.5 Summary of Interference Effects on a Short-Range Air Search Radar ................. 66

5 INTERFERENCE MEASUREMENTS ON A FIXED GROUND-BASED

METEOROLOGICAL RADAR...................................................................................... 68
5.1 Introduction ............................................................................................................ 68
5.2 Theoretical Calculation of Necessary Protection Criteria...................................... 68
5.3 System Operation, Output Products and Interference Sensitivity.......................... 72
5.4 Data Analysis Methodology and Results ............................................................... 75
5.5 Summary of Measurement Results ........................................................................ 80
5.6 Meteorological Radar Improvements..................................................................... 81
5.7 Summary of Interference Effects on a Weather Radar .......................................... 82

6 INTERFERENCE MEASUREMENTS ON MARITIME RADIONAVIGATION

RADARS .......................................................................................................................... 84
6.1 Introduction ............................................................................................................ 84
6.2 Description of Maritime Radars Tested ................................................................. 86
6.3 Interference Signal Characteristics ........................................................................ 91
6.4 Target Generation................................................................................................... 95
6.5 Maritime Radar Test Conditions............................................................................ 98
6.6 Maritime Radar Test Procedures (Non-Fluctuating Targets)................................. 100
6.7 Test Results for Maritime Radionavigation Radars ............................................... 101
6.8 UWB Interference Tests on a Maritime Radar ...................................................... 112
6.9 Summary of Interference Effects on Maritime Radars .......................................... 114

7 INTERFERENCE MEASUREMENTS ON AN AIRBORNE METEOROLOGICAL

RADAR ........................................................................................................................... 125
7.1 Introduction ............................................................................................................ 125
7.2 Characteristics of the Airborne Weather Radar ..................................................... 125
7.3 Interference Measurement Protocol for the Airborne Weather Radar ................... 127
7.4 Results of Interference Measurements on the Airborne Weather Radar................ 132
7.5 Summary of Interference Effects for the Airborne Weather Radar ....................... 135

8 SUMMARY OF INTERFERENCE EFFECTS ON RADARS ...................................... 136

8.1 Radars are Vulnerable to the Effects of Communication Signal Interference ....... 136
8.2 Radars Perform Robustly in the Presence of Interference from Other Radars ...... 137
8.3 Low-Level Interference Effects in Radar Receivers are Insidious ........................ 137

vi
 8.4 Low-Level Interference Can Cause Loss of Radar Targets at Any Range ............ 137
8.5 Radar Interference Waveforms and Test Reporting Should be Standardized........ 138

9 REFERENCES ................................................................................................................ 139

APPENDIX A: EXAMPLE INTERFERENCE REJECTION (IR) RESPONSES OF A

MARITIME RADIONAVIGATION RADAR........................................................................ 141

APPENDIX B: SELECTED INTERFERENCE EMISSION SPECTRA ............................... 147

APPENDIX C: CALIBRATION OF UNDESIRED SIGNALS AND EXAMPLES OF

RADAR IF SELECTIVITY CURVES.................................................................................... 153

APPENDIX D: TEST RESULTS ILLUSTRATING THE EFFECTIVE DUTY CYCLE OF

FM-PULSED WAVEFORMS IN A MARINE RADAR RECEIVER.................................... 156

vii
 FIGURES

Page

Figure 1. Probability of detection versus number of radar pulses. .................................... 6

Figure 2. Effective increase in receiver noise figure, (I+N)/N, as a function of I/N. ......... 9

Figure 3. Block diagram of a logarithmic amplifier detector. ............................................ 12

Figure 4. An example of clutter on a 9-GHZ maritime radionavigation radar PPI display. 15

Figure 5. Example STC curve for an air surveillance radar. .............................................. 16

Figure 6. An example of continuous severe co-channel radar interference on a PPI display. 19

Figure 7. Pulse repetition interval interference rejection circuit block diagram. ............... 20

Figure 8. Radar display with IR turned “off” versus “on” in the presence of interference. 21

Figure 9. Typical conditions inside an air-search radar station.......................................... 22

Figure 10. Block diagram of a typical test configuration in this interference study. ........... 23

Figure 11. NTIA and FAA engineers determining the locations on a radar circuit card. .... 24

Figure 12. Example of a 0-dB I/N calibration noise spectrum. ............................................ 25

Figure 13. Calibration technique for pulsed interference. .................................................... 26

Figure 14. Typical arrangement of NTIA interference and target generators at an air
search radar station.............................................................................................. 30

Figure 15. Block diagram of the radar target-generation hardware. .................................... 31

Figure 16. Example of a maritime radar PPI display during interference testing. ............... 31

Figure 17. Sector-wedge distribution of desired targets on a PPI display during testing..... 32

Figure 18. Statistical distributions for Swerling Case 1 fluctuating radar target power. ..... 35

Figure 19. Long Range Radar 1 single channel operation Pd with BPSK interference. ...... 46

Figure 20. Long Range Radar 1 dual channel operation Pd with BPSK interference. ......... 46

Figure 21. Cumulative PPI display of Long Range Radar 1 during interference................. 47

viii
Figure 22. Details of Long Range Radar 1 PPI display during live target interference
tests...................................................................................................................... 48

Figure 23. Test results with injected targets at a second installation of Long Range
Radar 1. ............................................................................................................... 49

Figure 24. Sequence of desired-target pulses in the IF stage of Long Range Radar 2......... 52

Figure 25. Long Range Radar 2 single channel operation variation in Pd with
interference............................................................................................................. 55

Figure 26. Long Range Radar 2 dual channel operation variation in Pd with interference.. 56

Figure 27. The effects of strong interference in two radar receivers.................................... 57

Figure 28. Example of target losses due to low-level interference effects........................... 58

Figure 29. Beam coverage patterns for the short-range air surveillance radar..................... 61

Figure 30. Short-range air surveillance radar input/output gain curve................................. 62

Figure 31. Test set-up for interference injection into short-range air search radar. ............. 64

Figure 32. Interference response curves for the short-range air search radar....................... 67

Figure 33. Meteorological radar test set-up block diagram.................................................. 75

Figure 34. Injected versus detected interference level in the meteorological radar receiver. 76

Figure 35. Detected minus injected interference level in the meteorological radar receiver. 77

Figure 36. Reflectivity regression for interference in the meteorological radar. ................. 78

Figure 37. Spectrum width regression for the meteorological radar. ................................... 79

Figure 38. Impact of near-term processing improvements on the weather radar interference
threshold.............................................................................................................. 83

Figure 39. Example of synthetic and raw video targets on a PPI display. ........................... 91

Figure 40. Target generator instrumentation for maritime radionavigation radar tests........ 95

Figure 41. Target generator timing diagram for maritime radionavigation radar tests. ....... 96

Figure 42. Radar A baseline state with video targets. .......................................................... 102

ix
Figure 43. Radar A with QPSK interference at I/N = -8 dB................................................. 102

Figure 44. Radar A with QPSK interference at I/N = +2 dB................................................ 103

Figure 45. Radar B Pd curves. .............................................................................................. 104

Figure 46. Radar D Pd curves. .............................................................................................. 107

Figure 47. 1-µs pulsed interference at 7.5% duty cycle and I/N= +40 dB. .......................... 108

Figure 48. 1-µs pulsed interference at 7.5% duty cycle and I/N= +12 dB. .......................... 109

Figure 49. OFDM interference effect at I/N= +3 dB............................................................ 110

Figure 50. OFDM interference effect at I/N= +6 dB............................................................ 110

Figure 51. Radar A PPI display, 100 kHz gated UWB interference = -85 dBm/MHz......... 115

Figure 52. Radar A PPI display, 100 kHz gated UWB interference = -95 dBm/MHz......... 116

Figure 53. Radar A PPI display, 100 kHz gated UWB interference = -105 dBm/MHz....... 116

Figure 54. Radar A PPI display, 100 kHz gated UWB interference = -110 dBm/MHz....... 117

Figure 55. Radar A PPI display, 1 MHz gated UWB interference = -115 dBm/MHz. ........ 117

Figure 56. Radar A PPI display, 1 MHz gated UWB interference = -110 dBm/MHz. ........ 118

Figure 57. Radar A PPI display, 1 MHz gated UWB interference = -105 dBm/MHz. ........ 118

Figure 58. Radar A PPI display, 1 MHz gated UWB interference = -95 dBm/MHz. .......... 119

Figure 59. Radar A PPI display, 1 MHz gated UWB interference = -85 dBm/MHz. .......... 119

Figure 60. Radar A PPI display, 1 MHz gated UWB interference = -75 dBm/MHz. .......... 120

Figure 61. Radar A PPI display, 10 MHz gated UWB interference = -116 dBm/MHz. ...... 120

Figure 62. Radar A PPI display, 10 MHz gated UWB interference = -111 dBm/MHz. ...... 121

Figure 63. Radar A PPI display, 10 MHz gated UWB interference = -106 dBm/MHz. ...... 121

Figure 64. Radar A PPI display, 10 MHz gated UWB interference = -86 dBm/MHz. ........ 122

Figure 65. Radar A PPI display, 10 MHz gated UWB interference = -66 dBm/MHz. ........ 122

x
Figure 66. Radar A video and synthetic targets without interference. ................................. 123

Figure 67. Radar A video and synthetic targets with 1 MHz UWB interference................. 123

Figure 68. Radar A loss of video targets and generation of false synthetic targets due to
1-MHz UWB interference................................................................................... 124

Figure 69. Block diagram of part of the airborne weather radar IF stage. ........................... 126

Figure 70. Example of the airborne weather radar display in the weather surveillance
mode.................................................................................................................... 127

Figure 71. Block diagram of the airborne weather radar interference test setup.................. 129

Figure 72. Interference burst as observed in the airborne weather radar IF stage................ 130

Figure 73. Example of a strongly visible interference strobe............................................... 131

Figure 74. Example of a marginal interference strobe ......................................................... 131

Figure 75. Example of barely visible interference ............................................................... 132

Figure 76. Strobes caused by high duty cycle (OFDM) interference................................... 134

Figure 77. Strobes caused by low duty cycle interference. .................................................. 134

Figure A-1. +50 dB I/N, ungated interference, 10 µs pw, 0.1% duty cycle, IR off................ 142

Figure A-2. +50 dB I/N, ungated interference, 10 µs pw, 0.1% duty cycle, IR on. ............... 142

Figure A-3. +50 dB I/N, ungated interference, 10 µs pw, 1% duty cycle, IR off................... 143

Figure A-4. +50 dB I/N, ungated interference, 10 µs pw, 1% duty cycle, IR on. .................. 143

Figure A-5. +80 dB I/N, ungated interference, 10 µs pw, 1% duty cycle, IR off................... 144

Figure A-6. +80 dB I/N, ungated interference, 10 µs pw, 1% duty cycle, IR on. .................. 144

Figure A-7. +10 dB I/N, ungated interference, 10 µs pw, 5% duty cycle, IR off................... 145

Figure A-8. +10 dB I/N, ungated interference, 10 µs pw, 5% duty cycle, IR on. .................. 145

Figure A-9. +15 dB I/N, ungated interference, 10 µs pw, 5% duty cycle, IR off................... 146

Figure A-10. +15 dB I/N, ungated interference, 10 µs pw, 5% duty cycle, IR on. .................. 146

xi
Figure B-1. 1.023 MBit/s BPSK interference signal spectrum. ............................................. 147

Figure B-2. 10 MBit/s BPSK interference signal spectrum. .................................................. 147

Figure B-3. 5 MBit/s BPSK interference signal spectrum. .................................................... 148

Figure B-4. 0.5 MBit/s BPSK interference signal spectrum. ................................................. 148

Figure B-5. 2 MBit/s QPSK interference signal spectrum. .................................................... 149

Figure B-6. W-CDMA interference signal spectrum. ............................................................ 149

Figure B-7. CDMA-3X interference signal spectrum. ........................................................... 150

Figure B-8. QAM interference signal spectra for maritime radar tests. ................................. 150

Figure B-9. CDMA interference spectra for maritime radar tests.......................................... 151

Figure B-10. UWB interference spectra as a function of receiver bandwidth. ........................ 151

Figure B-11. Chirped-pulse interference spectrum. ................................................................. 152

Figure C-1. Channel A IF response of Long Range L-Band Radar 1. ................................... 153

Figure C-2. Channel B IF response of Long Range L-Band Radar 1. ................................... 154

Figure C-3. IF response of one channel of Long Range Radar 2. .......................................... 154

Figure C-4. IF response curve of the short-range air search radar. ........................................ 155

Figure C-5. IF response curve of a typical maritime radionavigation radar........................... 155

Figure D-1. Frequency response of a marine radar IF stage to a CW input. .......................... 157

Figure D-2. 1-µs unmodulated pulse in a marine Radar F receiver........................................ 159

Figure D-3. Chirped waveform 1 in a marine Radar F receiver. ............................................ 159

Figure D-4. Chirped waveform 2 in a marine Radar F receiver. ............................................ 160

Figure D-5. Chirped waveform 3 in a marine Radar F receiver. ............................................ 160

Figure D-6. Chirped waveform 4 in a marine Radar F receiver. ............................................ 161

Figure D-7. Chirped waveform 5 in a marine Radar F receiver. ............................................ 161

xii
Figure D-8. Chirped waveform 6 in a marine Radar F receiver. ............................................ 162

Figure D-9. Chirped waveform 7 in a marine Radar F receiver. ............................................ 162

xiii
 TABLES

Page

Table 1. Example of Interference Power Levels When Interference Bandwidth

Exceeded Radar IF Bandwidth .............................................................................. 28

Table 2. Example of Interference Power Levels When Interference Bandwidth was

Less Than or Equal to Radar IF Bandwidth........................................................... 29

Table 3. Fluctuating Target Power Levels Derived from the Curves of Figure 18.............. 36

Table 4. Technical Characteristics of Long Range Radar 1................................................. 41

Table 5. Technical Characteristics of Long Range Radar 2................................................. 50

Table 6. Technical Characteristics of the Short-Range Air Surveillance Radar.................. 60

Table 7. Control Settings for the Short Range Air Surveillance Radar. .............................. 65

Table 8. Technical Characteristics of the Meteorological Radar ......................................... 72

Table 9. Sensitivity of Meteorological Products to Interference Induced Error .................. 73

Table 10. Reflectivity Results for Example Analysis. ........................................................... 79

Table 11. Error Reduction Values.......................................................................................... 80

Table 12. Spectrum Width Results for Example Analysis..................................................... 80

Table 13. Measured I/N Thresholds Necessary for Protection of the Meteorological Radar 80

Table 14. Technical Characteristics of Maritime Radionavigation Radar A ......................... 86

Table 15. Technical Characteristics of Maritime Radionavigation Radars B and D ............. 88

Table 16. Technical Characteristics of Maritime Radionavigation Radars C and E.............. 89

Table 17. Technical Characteristics of Maritime Radionavigation Radar F.......................... 90

Table 18. Chirped-Pulse Interference Waveform Characteristics.......................................... 92

Table 19. Phase Coded Pulsed Interference Waveform Characteristics ................................ 94

Table 20. Unmodulated Pulsed Interference Waveform Characteristics ............................... 94

xiv
Table 21. Swerling Case 1 Target Power Levels (Relative to Nominal Value of -88 dBm) . 98

Table 22. Maritime Radar Test Control Settings ................................................................... 98

Table 23. Target Power Levels (Non-Fluctuating) Required to Achieve a Pd of 0.90. ......... 100

Table 24. Radar A Responses to QPSK Interference............................................................. 103

Table 25. Radar C Responses to Continuous QAM Interference .......................................... 105

Table 26. Radar C Responses to Gated CDMA Interference................................................. 106

Table 27. Radar E Responses to Gated CDMA Interference................................................. 107

Table 28. Radar F Responses to Interference......................................................................... 111

Table 29. Characteristics of the Airborne Weather Radar ..................................................... 125

Table 30. Characteristics of Interference Waveforms for Airborne Weather Radar Tests .... 128

Table 31. Results of Interference Tests on the Airborne Weather Radar .............................. 133

Table 32. I/N Levels of Communication Signal Modulations at which Performance

Decreased for All Radars Tested............................................................................ 136

Table D-1. Characteristics of Chirp-Pulse Waveforms Injected into Marine Radar F ............ 158

xv
 ABBREVIATIONS/ACRONYMS

A-D analog-to-digital (information conversion) in a radar receiver

AGC automatic gain control radar receiver processing
AIS automatic identification systems (for ships)
AWG arbitrary waveform generator
BPSK binary phase-shift keyed signal modulation
CDMA code division multiple access signal modulation
CFAR constant false alarm rate radar receiver processing
CRT cathode ray tube
COHO coherent oscillator (for MTI, or Doppler, processing)
CW carrier wave (sine wave) signal modulation
DTE digital target extractor
DVB digital video broadcast
EESS earth exploration satellite service
EIRP effective isotropic radiated power
ENG-OB electronic news gathering-outdoor broadcast (video data) signal modulation
FAA Federal Aviation Administration (of the United States of America)
FTC fast time constant (or logarithmic FTC, log-FTC) radar receiver processing
HF high frequency
I interference power level (in a bandwidth in a radar receiver)
I/N interference-to-noise power ratio (in a radar receiver)
IF intermediate frequency (of a radar receiver) stage
IAGC instantaneous automatic gain control (also simply AGC)
IMO International Maritime Organization
IMT-2000 International mobile telecommunications (year 2000) signal modulation
defined by ITU-R (also known as wireless 2.5 G, wireless 3G, next-
generation (NG) wireless mobile, and IMT-Advanced)
I-Q in-phase and quadrature components of a signal, differing by a phase shift
of π/2
IR interference rejection (feature in a radar receiver used against pulsed
interference)
IS-95 interim standard 95, also known as TIA-EIA-95 and by a trade name,
cdmaOne
ISM industrial, scientific, and medical (spectrum bands)
ITS Institute for Telecommunication Sciences (of NTIA, U.S. Dept. of
Commerce)
ITU-R International Telecommunication Union, Radiocommunication Sector
LNA low noise amplifier
Log FTC logarithmic fast time constant (also simply FTC) radar receiver processing
MCA Maritime and Coast Guard Agency (of the United Kingdom)
MDS minimum detectable signal level (in a radar receiver)
MTI moving target indicator (radar target processing feature)
N noise power level (in a bandwidth in a radar receiver)
NTIA National Telecommunications and Information Administration (of the
U.S. Department of Commerce)

xvi
NWS National Weather Service (of the United States of America)
OFDM orthogonal frequency division multiplexing signal modulation
OS CFAR ordered statistic constant false alarm rate
OSM Office of Spectrum Management, NTIA, U.S. Department of Commerce
OTR on-tuned rejection factor (for bandwidth mismatches)
Pd probability of detection (of radar target(s))
PPI plan position indicator radar display
prf pulse repetition frequency (of radar pulses)
pri pulse repetition interval (between radar pulses)
prt pulse repetition train
pw pulse width (of a radar)
QAM quadrature amplitude modulation (phase-coded signals, with a numeric
prefix indicating the available number of phase states, e.g. 64 QAM is
QAM with 64 possible phase states)
QPSK quadrature phase-shift keyed signal modulation
Rmax maximum range of a radar
Radar radio detection and ranging, paired receiver and transmitter
RBW resolution bandwidth (or IF bandwidth) of a spectrum analyzer
RCS radar cross section (often simply called cross section in radar-specific
contexts)
RF radio frequency
RMS root mean square (average power detection)
S signal power in a radar receiver
Smin minimum detectable signal level
SOLAS safety of life at sea (international regulations)
STC sensitivity time control (or swept gain) radar receiver processing
TBM threshold bias map
TDMA time division multiple access signal modulation
Tx/Rx transmitter-receiver combination
UHF ultrahigh frequency
USCG United States Coast Guard
UWB ultrawideband
VHF very high frequency
W-CDMA wideband code division multiple access signal modulation

xvii
 EXECUTIVE SUMMARY

Radars play critical roles in national security, air traffic control, weather observation and
warning, scientific applications, mapping, search and rescue operations, and other safety-of-life
missions. Radar transmitter and receiver characteristics are engineered to successfully
accomplish their missions in these areas. The technical characteristics of radars (such as high
transmitter peak power levels and very sensitive designs for the receivers) have typically resulted
in exclusive or primary spectrum allocations for radar operations in selected radio bands.

In recent years, spectrum crowding has led to proposals for reduction of available spectrum for
exclusive or primary radar operations, as well as for co-channel (or nearly co-channel) spectrum
sharing between radars and non-radar (communication-type) signals. It has been proposed in
various forums, for example, that communication signals can (and should) share spectrum bands
with radar systems. Such proposals typically presume that radar receivers will not suffer undue
loss of performance due to such sharing as long as the interference levels are relatively low.
Some sharing analyses assume that radar receivers are relatively robust against radio frequency
(RF) interference effects from communication signals at low levels.

Such proposals raise critical questions regarding the robustness of radar receiver performance in
the presence of RF interference, especially at low levels. These include: At what power levels do
interfering signals cause adverse effects on the performance of radar receivers? How are
interference effects manifested in radar receivers? What are the interference levels at which radar
receivers lose desired targets? What if any low-level interference effects occur on radar receiver
displays? Are radar targets only lost at the edge of radar coverage?

To answer these questions this report’s authors have performed extensive tests and measurements
on many radar receivers performing a variety of critically important missions in several spectrum
bands. Radar types that have been examined include: short-range and long-range air traffic
control; fixed ground-based weather surveillance; airborne weather surveillance; and maritime
navigation and surface search. The radars that have been tested operate in the spectrum bands:
1215-1400 MHz; 2700-2900 MHz; 2900-3700 MHz; and 8500-10500 MHz.

The most important results of this work can be summarized in four major findings:

1) Interference with communication-type signals degrades radar performance at interference-to-

noise (I/N) levels between -9 dB and -2 dB (well below the internal noise of the radar receivers).
One radar lost targets at an I/N level of -10 dB, and I/N = -6 dB caused all but one of the radars to
lose targets. Future improvements to weather radars are predicted to render them vulnerable at
I/N levels as low as -14 dB.

2) In contrast to the effects of interference from communication signal modulations, pulsed

interference at low duty cycles (less than about 1-3%) is tolerated at I/N ratios as high as +30 dB
to +63 dB. In other words, the radar receivers that were tested performed very robustly in the
presence of the type of signals that are transmitted by other radars. Radars tend to share spectrum
well with other radars.

xix
3) When radar targets are lost due to low-level interference, those losses are insidious. There are
no overt indications to radar operators or even to sophisticated radar software that losses are
occurring. No dramatic indications such as flashing strobes on radar screens are observed. This
insidiousness can make low-level interference more dangerous than higher levels that will
generate strobes and other obvious warning indications for operators or processing software.

Even when radars experience serious performance degradation due to low-level interference, it is
very unlikely that such interference will be identifiable as such. It is therefore unlikely that such
interference will ever result in reports to spectrum management authorities even when it causes
loss of desired targets. Since low-level interference is not expected to be identified or to generate
reports when it occurs, lack of such reports cannot be taken to mean that such interference does
not occur.

4) Interference can (and will) cause loss of targets at any distance from any radar station; loss of
targets due to radio interference is not directly related to distance of targets from radar stations.
When radar performance is reduced by some number of decibels, X, then all targets that were
within X decibels of disappearing from coverage will be lost. Range from the radar is not a factor
in this relationship. Interference can cause loss of desired, large targets (such as commercial
airliners, oil tankers, and cargo ships) at long distances, as well as small targets at close distances.
Low-observable targets that could be lost include, for example, light aircraft; business jets;
incoming missiles; missile warheads; floating debris including partially submerged (and therefore
dangerous) shipping containers; life boats; kayaks, canoes, dinghies; periscopes; and swimmers
in life jackets.

Because any marginal radar target can potentially be lost at any distance in the presence of radio
interference, radio interference does not translate directly into an equivalent radar range
reduction. While “range reduction” due to interference can be used as a metric, its use must be
qualified with the (unrealistic) condition of a fixed, constant cross section for all desired targets.

Taken together, the technical results in this report show that radar receivers are not generally robust
against low-level interference from non-radar (communication-type) radio signals. The low thresholds at
which interference effects occur, the insidious characteristics of such interference, and the wide range of
distances over which radar receivers can experience interference effects are critical technical indications
that, to avoid impairment of radar operations, care must be taken to ensure that radar receivers are not
subjected to unnecessary interference from non-radar signals above identified thresholds. Conversely,
radars have now been shown to operate robustly when they share spectrum with other radar signals
having low duty cycles (i.e., when the interference duty cycles are less than about 1-3%).

xx
 EFFECTS OF RF INTERFERENCE ON RADAR RECEIVERS

Frank H. Sanders,1 Robert Sole,2 Brent L. Bedford,1 David Franc,3 and Timothy Pawlowitz4

This report describes the results of interference tests and measurements that have
been performed on radar receivers that have various missions in several spectrum
bands. Radar target losses have been measured under controlled conditions in the
presence of radio frequency (RF) interference. Radar types that have been examined
include short range and long range air traffic control; weather surveillance; and
maritime navigation and surface search. Radar receivers experience loss of desired
targets when interference from high duty cycle (more than about 1-3%)
communication-type signals is as low as -10 dB to -6 dB relative to radar receiver
inherent noise levels. Conversely, radars perform robustly in the presence of low
duty cycle (less than 1-3%) signals such as those emitted by other radars. Target
losses at low levels are insidious because they do not cause overt indications such as
strobes on displays. Therefore operators are usually unaware that they are losing
targets due to low-level interference. Interference can cause the loss of targets at any
range. Low interference thresholds for communication-type signals, insidious
behavior of target losses, and potential loss of targets at any range all combine to
make low-level interference to radar receivers a very serious problem.

Key words: radar interference; radar interference vulnerability; radar performance

degradation; radar target loss; radar target detection; RF interference; UWB
interference effects

1 RADAR FUNDAMENTALS RELATED TO INTERFERENCE TESTING

AND MEASUREMENTS

1.1 Introduction

Radars play critical roles in national security, air traffic control, weather observation and
warning, scientific applications, mapping, search and rescue operations, and other safety-of-life
missions. Radar transmitter and receiver characteristics are engineered to successfully
accomplish their missions in these areas. The technical characteristics of radars (such as high
peak power levels emitted by the transmitters and very sensitive designs for the receivers) have
usually resulted in exclusive, or at least primary, spectrum allocations for their operations.

In recent years, spectrum crowding has led to proposals for reduction of available spectrum for
exclusive or primary radar operations, as well as for co-channel (or nearly co-channel) spectrum

1
Institute for Telecommunication Sciences, National Telecommunications and Information Administration (NTIA),
U.S. Department of Commerce, Boulder, CO.
2
Office of Spectrum Management (OSM), NTIA, Washington, DC.
3
National Oceanic and Atmospheric Administration (NOAA), formerly with the National Weather Service (NWS),
Washington, DC.
4
Federal Aviation Administration (FAA), Washington, DC.
sharing between radars and non-radar radio signals. It has been proposed in various forums, for
example, that communication signals can (and should) sometimes share spectrum bands with
radar systems. Such proposals typically presume that radar receivers will not suffer undue loss of
technical performance due to such sharing so long as the interference levels are relatively low.
Some sharing analyses assume that radar receivers are relatively robust against radio frequency
(RF) interference effects from low levels of non-radar signals.

These proposals raise critical technical questions. These include: At what power levels do
interfering signals cause adverse effects on the performance of these receivers? How are
interference effects manifested in these radar receivers? Specifically, what are the interference
levels at which radar receivers lose desired targets, and do low-level interference effects create
observable manifestations on radar receiver displays?

NTIA, in cooperation with other US Government agencies and the United Kingdom (UK),5 has
performed a series of tests and measurements of the response of microwave radars to low levels
of RF interference from a variety of communication and non-communication signals. This report
describes the results of interference tests and measurements on several representative radar
receivers. Interference data have been collected on radars performing various missions in several
different spectrum bands. Radar types that have been examined include short range air traffic
control; long range air traffic control; fixed weather surveillance; airborne weather surveillance;
and maritime navigation and surface search. These radars have been operational in the spectrum
bands 1200-1400 MHz; 2700-2900 MHz; 2900-3700 MHz; and 8500-10500 MHz.

1.2 Noise Versus Interference

Radars operate by directing transmitted radio energy through space toward remote objects that
need to be observed and subsequently receiving and processing reflections of small portions of
the transmitted energy returned as echoes from the desired objects (called targets). The time
delay between transmission and reception of the energy is used to determine the distances
between the radar and the targets, while the direction in which the energy is initially directed and
then returned as reflections informs the direction of the targets from the radar. Ultimately, the
key to successful radar design is the solution of two technical problems: generation and direction
of an adequate amount of radio energy from the transmitter in sufficiently localized directions
through space; and the reception, detection, and discrimination of the small fraction of energy
that has been reflected from both desired targets and undesired reflecting objects (collectively
called clutter) in the presence of natural (and sometimes manmade) noise in the radar receiver.

This report addresses the general question of how, and to what extent, RF interference causes
degradation to the performance of radar receivers. Before interference effects are discussed, it is
necessary to define interference, describe how it differs from other environmental degradation
effects, and explain why it is considered separately from those effects. It is also necessary to
describe the normal processes of target reception, detection, and discrimination in radar
receivers. Those topics are addressed in the following parts of this report section.

5
Including the National Oceanic and Atmospheric Administration (NOAA), U.S. Coast Guard (USCG), Federal
Aviation Administration (FAA), and the United Kingdom’s (UK) Fraser Test Range at Portsmouth.

2
1.2.1 Definition and Distinction of Noise Versus Interference

Unwanted energy coupled into radar receivers from natural sources and from manmade devices
that are not intentional radio transmitters is called noise. Noise degrades radar receiver
performance and is generated by many sources. A natural source is the energy produced by
thermal electrons within radar receiver circuitry; it is one of the most fundamental limiting
factors in radar receiver performance. Other noise is coupled into radar receivers from naturally
occurring, distant sources. In the high frequency (HF) spectrum these prominently include noise
from lightning strokes reflected from our planet’s ionosphere and (as a smaller effect) Jupiter,
while at higher frequencies (very high frequency (VHF), ultrahigh frequency (UHF), and lower
microwave), important noise sources are the Sun and to a lesser extent our galactic center and a
variety of other astronomical objects. Manmade noise across the spectrum from HF to lower
microwave frequencies (around 1.5 GHz) is coupled into radar receivers from unintentional
emission sources such as electric motors, above-ground power lines, and automotive ignition
systems. Spectrum survey data [1-4] indicate that manmade noise levels tend to decrease below
receiver thermal noise at higher microwave frequencies (above about 1.5 GHz), although some
sources (e.g., microwave ovens) can produce significant background microwave emissions in
industrial, scientific, and medical (ISM) radio bands such as at 2.5 GHz [5-8].

Randomly distributed radio energy from all sources, both natural and manmade, combines to
generate overall environmental background noise against which radar receivers must always
perform. Noise sources are unavoidable and generally uncontrollable. However, they do produce
known background levels, their emissions obey well-understood patterns in time and space, and
they are generally predictable in both their location and their timing (e.g., the diurnal location and
movement of astronomical sources across the celestial sphere and the diurnal power levels,
cycles, and spectrum ranges of manmade noise near cities, highways, and industrial zones).

In contrast to noise, interference is unwanted radio energy6 coupled into radar receivers from
manmade, intentional radio transmitter sources such as communication devices. Like noise,
interference degrades radar receiver performance. Interference, however, generally has statistical
and spectrum characteristics that are different from noise. Another important contrast between
noise and interference is that, while noise may be unavoidable and largely uncontrollable,
interference is both avoidable and controllable through sound spectrum engineering and
management procedures.

Although both noise and interference cause degradation in radar receiver performance, the
existence of noise does not logically imply that radar receivers should therefore be expected to
operate in the presence of interference. If any connection is to be drawn between the two
phenomena, it is that the existence of noise as a source of uncontrollable and unavoidable
degradation in radar environments implies that interference (which in contrast, and by definition,
is both controllable and avoidable) must be restricted all the more carefully in order not to add to
the inevitable performance degradation caused by environmental noise.

Noise phenomena and their effects on radar receiver performance have been treated extensively
in existing technical literature [9-12, for example]. Noise effects are typically taken into account
6
Interference is sometimes defined as the effect of unwanted energy in a receiver, rather than the energy itself.

3
in radar receiver design and are distinct and separable in both their origins and their behavior
from interference effects. For these reasons, and because interference is a controllable
phenomenon, this report does not address noise effects in radar receivers other than as an
inherent limiting factor generated by thermal electron activity in receiver circuitry. Only
interference degradation effects on radar receivers are addressed in this report.

1.3 Detection of Radar Targets

Radar targets are detected by transmitting radio energy into space, receiving a portion of the
transmitted energy that is reflected as echoes from targets, and processing it inside receiver
circuitry to separate the echo energy from whatever noise and interference are present in the
receiver. A radar will always have a maximum range that is a function of the transmitter power,
the degree to which the transmitted energy is confined directionally in space, the effective area of
the receiving antenna, the efficiency with which targets reflect radio energy at the radar
frequency, and the minimum threshold within the receiver for detection of echo energy.

1.3.1 The Radar Equation

Assuming theoretically perfect free-space radio signal propagation conditions, the maximum
range of a radar, Rmax, is given by the so-called radar equation (modified from [9, pg. 15]):
1
⎡ PG A σ ⎤ 4

Rmax = ⎢ t 2t e 0 ⎥ (1)
⎢⎣(4 π ) LsSmin ⎥⎦

where:

Pt = transmitted power, watts;

Gt = gain of the transmitter antenna, linear term;
Ae = effective aperture of the receiving antenna, m2;
σ0 = nominal (specified) target cross section, m2;
Ls = system loss between transmitter/receiver and antenna, linear dimensionless quantity;
Smin = minimum detectable signal, watts.

With the exception of the target cross section term, σ0, all of the variables in (Eq. 1) are in
principle controllable by radar designers (although there are practical limits on all of the
variables). Rmax is a theoretical maximum that is never realized due to practical engineering and
design limitations of field-deployed equipment and unanticipated losses within radar systems.
Actual values of Rmax are typically half of the theoretical value [9]. A variety of environmental
factors tend to decrease radar range. These include propagation losses that exceed the free-space
assumption of the radar equation as well as fluctuations in the radar cross sections of targets.

4
1.3.2 Radar Target Detection Probability

The peak values of transmitted power, Pt, that are required for effective observation and tracking
of radar targets are so high that they often can only be generated for short intervals. Furthermore,
most radar designs require that the radar transmitters be periodically turned off so that the
receivers can listen for the target echoes. Therefore most radars transmit radio energy in pulses.7

Detection of radar targets is a probabilistic process; for any given target, there is a probability
that it will be detected as an echo from any individual pulse of transmitted energy. But the
probability of detection on a per-pulse basis is usually too low for effective radar operations. The
per-pulse detection probability can be increased by raising the values of the parameters in
(Eq. 1), but in well-engineered radars the values of the (Eq. 1) parameters have already reached
the limits of available technology and practicality. The only way to obtain high probabilities of
target detection is to illuminate a target with more than one pulse, so that the cumulative
probability of detection becomes adequate for effective operations. In the most simplistic
approach, it is established that the probability of detection per pulse is x (where x<1). The
probability of not detecting a target with a single pulse is thus (1-x). Assuming independence of
detection probability from pulse to pulse, the probability of not detecting a target with any of n
pulses is (1-x)n. The probability of detection of a target, Pd, with at least one of n pulses is unity
minus that value:

Pd (x,n ) = 1− (1− x ) .
n
(2)

Figure 1 illustrates the principle of variation of Pd as a function of n for a variety of values of x.

For typical operational radar designs, n is often between 10 and 20. When fewer than 10 pulses
illuminate a target, detection probability is low and detection probability increases rapidly with
the use of additional pulses. But detection probability only increases slowly after more than about
20 pulses are used on a target; diminishing returns in radar performance relative to longer dwell
times on each target tend to preclude the use of many more pulses for illumination.8 The need for
targets to be illuminated with 10 to 20 pulses leads in turn to the selection of a variety of
additional radar parameter values, such as the maximum rate that the radar beam may be scanned
through space.9

Explained another way, the low per-pulse probability of detection must be somehow overcome to
make radars work effectively. Since the parameters in (Eq. 1) (such as transmitter power and
antenna gain) have already been optimized in good radar designs, the only remaining method for
reliably observing targets is through receiver processing of multiple pulses reflected from each
target. This is receiver processing gain. Although processing gain enhances target detection, and
may tend to mitigate some special types of interference (such as from pulsed sources that

7
Some radars, especially those used for certain types of target tracking, do transmit continuously.
8
There are some exceptions. For example, some meteorological radars examine the very weakly reflecting
phenomena of high-altitude winds. These radars, called wind profilers, may need to integrate echo energy from tens
of thousands of pulses over a period of a few minutes in order to generate a single data point.
9
For example, if a radar must have a 200 nm range and at least 15 pulses must illuminate each target within a 1-
degree wide beamwidth, then the radar must scan at a rate that is not faster than one revolution every 13.4 sec. This
of course becomes the lower bound for the time interval between updates on target locations.

5
transmit energy at a low (less than 1 per cent) duty cycle), it is not an interference rejection
technique per se.

Figure 1. Probability of detection versus number of radar pulses illuminating a target for three
probabilities per pulse, from (Eq. 2): x=0.05, x=0.10, and x=0.20. Many radars balance the need
to detect targets with high probability against the need to update scans in a timely manner by
using between 10 to 20 pulses per target (indicated with solid vertical lines).

1.4 Radar Receiver Inherent Noise

As noted in Section 1.2, one of the most fundamental limitations on microwave radar receiver
performance is inherent, internal noise generated by thermal electrons in receiver circuitry. The
average power level in watts, P, of this noise is:

P = kTB (3)

6
where:

k = Boltzmann’s constant (1.38⋅10-23 J/(K⋅Hz));

T = temperature of the circuitry (K);
B = bandwidth of the circuitry (Hz).

For example, a circuit at room temperature (about 290 K) operating in a bandwidth of 1 MHz
(106 Hz), will have an inherent noise level of 4⋅10-15 W, or -144 dBW, or -114 dBm.

The noise factor, fn, of a receiver is defined as:

N out ⎛ ∆N ⎞
fn = = ⎜1+ ⎟ (4)
kTB ⎝ kTB ⎠

where:

Nout = total available output noise power from the circuit;

∆N = noise power in addition to kTB generated by the circuit.

The noise figure, Fn, of the circuit, a decibel quantity, is defined as:

Fn = 10log( f n ). (5)

Ideally, Nout = kTB, or, equivalently, ∆N = 0. If this performance could be achieved, then the
noise factor of the receiver would be unity and the noise figure of the receiver would be zero
decibels. The sensitivity of such a receiver would be theoretically optimal.

Realistically, such performance is impossible; the values of Nout and ∆N will be somewhat larger
than kTB and zero, respectively. In a well-designed receiver in which the noise figure and gain
characteristics of the first low-noise amplifier (LNA) are properly cascaded with all of the
circuitry after the LNA, the LNA noise figure will nearly determine the overall noise figure of
the entire receiver. A high-performance LNA noise figure in a radar receiver is a few decibels
higher than kTB. Insertion loss of components in the radar receiver ahead of the LNA (such as RF
band-limiting filtering) will add directly to the overall noise figure of the receiver. Such losses
can add up to another decibel or more.

In practice, an overall noise figure of 10 dB is an easily achievable value for radar receivers,
while a value of 5 dB or less is usually considered to be nominal. To continue with the example
given above, if kTB is -114 dBm and the noise figure of the receiver is 5 dB, then the radar
receiver overall noise level will be effectively -109 dBm. Ultimately, all desired targets must
somehow be detected against receiver noise.

7
 1.5 Minimum Detectable Signal

Figure 1 illustrates the importance of obtaining the highest possible probability of detection on a
per-pulse basis for the overall problem of detecting targets with n pulses. Each pulse is detected
relative to radar receiver noise. The total signal power available to the receiver from all pulse
echoes is S. There is a minimum S value, Smin (Eq. 1), which determines the maximum radar
range, Rmax, for a given target cross section. (It can also be used to determine the minimum target
cross section that may be detected at any range.) This critical (Smin) itself is limited by the radar
receiver inherent noise.

If a radar is barely able to detect a target at maximum range, Rmax, then the corresponding
minimum signal-to-noise ratio, (S/N)min, must be related to Smin as follows:

⎛S⎞
Smin = kTBFn ⎜ ⎟ . (6)
⎝ N ⎠ min

For a false-alarm probability of 10-6, a detection probability of 0.8, and a standard echo-power
variation model (Swerling case 1, see Section 1.7 below), the required signal-to-noise ratio is
about 17 dB [12, pg. 24], which is a linear factor of 50:1. With a nominal noise figure of 5 dB (a
linear factor of 3.16), the value of (Eq. 6) in a (nominal) 1-MHz bandwidth is Smin=6.3⋅10-13 W.

The value of Smin is further reduced by the effect of integrating n pulses, where typically n=20.
This effect results in a value of Smin that is about 10-13 W [12, pg. 25].

Although (Eq. 6) is obtained by inverting (Eq. 1) for the maximum detection range of a target
with a nominal cross section σ0, it is important to note that the value of Smin of 10-13 W is itself
independent of range. Any target, at any range, that returns less than this much energy to the
radar receiver will probably not be observed on a radar display. If all targets had the nominal
cross section σ0 of (Eq. 1), then Smin would be returned only by targets at the edge of the coverage
range and all targets within that range would be detected. But in reality, cross sections vary
widely and many targets will have cross sections that are less than σ0. Such targets will be lost at
distances less than Rmax.

1.6 Increase in Receiver Noise Figure as a Function of Interference Power Level

If interference is introduced into a radar receiver, the average power contribution from the
interference, I, will sum with the radar inherent noise power, N, and that summed power will tend
to mask the detection of desired targets. The ratio of the summed noise-plus-interference to
inherent noise is (I+N)/N, and its behavior relative to the ratio of I/N is portrayed graphically in
Figure 2.

As shown in Figure 2, the receiver noise figure is increased by 0.5 dB when the average
interference power is 9.5 dB below the nominal receiver noise level, and the receiver noise figure
is increased by 1 dB when the average interference power is 6 dB below the nominal receiver
noise level. These effective noise figure increases would represent equal increases in the
minimum detectable signal level of radar receivers that are subjected to interference.

8
 1.7 Detection of Radar Targets When Their Levels Fluctuate

The power level of echoes from radar targets will normally vary in time. Such variations may
occur due to time-dependent changes in the range, the aspect angle of the target, and the
propagation and multi-path characteristics between the radar and the target. The result is that the
variable of cross section in (Eq. 1), σ0, is really a time-varying effective cross section, σeff(t).

Figure 2. Receiver noise figure increases as the sum of interference power, I, and inherent
internal noise, N, divided by the same noise power: (I+N)/N. This is the relation between
(I+N)/N and I/N.

How do radar designers take these fluctuations in effective target cross sections into account
when they design the systems? A naïve approach would be to increase radar transmitter power to
accommodate the deeper power dips in such fluctuations. But this is not practical because the
capacity for high peak power generation in radar transmitters is very expensive. Furthermore,
most radar transmitters are designed to produce the highest possible peak power levels that
technology can create and that budgets can afford. Thus it is not generally possible to increase
the transmitter peak power, and another way must be found to compensate for target power
fluctuations.

The compensation method that is actually implemented is to transmit more pulses at fluctuating
targets than would have been needed if the targets had constant echo amplitudes, and to integrate
the resulting multiple echoes to detect targets even when they are weaker than the nominal cross
section. The approach is explained in standard radar design references [9-15, for example]. In
this approach, standard mathematical models are used to describe the statistical characteristics of
radar echo amplitudes. But however useful such models are when applied to radar design, they

9
ironically do not reproduce the behavior of real radar targets [9, pg. 52]. “The uncertainty
regarding fluctuating target models makes the use of the constant (non-fluctuating) cross section
in the radar equation an attractive alternative when a priori information about the target is
minimal,” [9, pg. 52].

1.8 Criteria for Radar Receiver Interference Thresholds

A recent, comprehensive compilation of currently accepted interference criteria from the existing
literature indicates that currently accepted limits on I/N for carrier wave (CW) and noise-like
interference in radar receivers are currently between -6 dB to -10 dB [16, pp. 5-2 and 5-3]. No
standards currently exist for any other categories of interference (e.g., digital signal modulations)
in radar receivers.

Based on the existing literature on this subject [16], limits on I/N in radar receivers are
understood to mean the maximum allowable ratio of interference power to radar receiver noise
power in the IF stages of radar receivers. The protection criteria that have been identified do not
ensure that interference effects will not occur at (or below) those I/N levels; rather, it has simply
been generally understood that these values should not be exceeded unless additional
compatibility studies are performed which show that higher I/N levels would be acceptable under
any given set of circumstances for any given type of radar receiver. The state of the existing
literature on this subject begs several questions:

1) Is a limit on the I/N ratio in a radar receiver IF stage necessarily the best (or most appropriate)
criterion for interference protection?

2) How reliable is the range of existing I/N protection criteria of -6 dB to -10 dB for protection of
radar receivers from CW and noise-like interference?

3) For types of interference other than CW and noise-like (e.g., digital signal modulations), what
should be the I/N protection criteria for radar receivers?

4) Can tighter bounds be placed on interference protection criteria for radar receivers?

In answer to the first question, alternative proposals have been variously problematic because
they do not address some types of radar, or because they invoke protection criteria that cannot be
analyzed or verified in any practical sense for operational, deployed systems. For example, it has
been proposed that interference might be allowed to occur in some types of radar receiver in
some spectrum bands on a statistical basis, with more interference allowed for shorter amounts of
time and lower interference allowed for longer periods, according to some sort of predetermined
‘interference budget.’ But one of the problems with this approach is that no practical method has
been identified for monitoring such interference in radar receivers, determining the sources,
computing the current state of the interference budget from the interfering sources, and then
somehow communicating from the radars to the interference sources that those sources must
modify (in some unspecified way) their operations so as to mitigate the interference they are
generating. In contrast, the approach of analyzing worst-case interference levels in radar receiver
IF stages from the known characteristics of the radars and potential interference sources, while
conservative, is nevertheless practical and verifiable (both theoretically and through

10
measurements), while providing assurance that critical radar missions (including safety-of-life)
are not adversely impacted by interference sources. As documented in this report, degradation of
radar receiver performance can be closely correlated with I/N levels in radar receivers for high-
duty-cycle interference such as CW, noise, and digital signal modulations.

As for the remaining questions above, until now very little information has been available in the
technical literature regarding the effects of low-level interference in radar receivers. It has
therefore been debatable as to whether I/N interference protection limits of -6 dB to -10 dB I/N
are necessarily appropriate for radar receivers in general, and whether those bounds might be
determined more closely for various types of radar systems. These questions have been the
impetus for the tests and measurements that are documented in this report.

1.9 Radar Detector Characteristics

The detector in a radar receiver is the device that extracts modulation information from the
received carrier signal of the target echoes. The “detector” usually comprises the parts of the
radar receiver between the output of the last IF amplifier and the input of the data processor (or
other data display indicator). Detectors operate on both signal and noise inputs. The most widely
used types are described below.

1.9.1 Envelope (Square-Law, Optimum-Law, or Quadratic) Detectors10

An envelope detector produces an output that is a function of the amplitude of the input
waveform modulation envelope; phase information in the waveform is lost. The output of the
detector is usually either a linear function of the input envelope11 or, more commonly, is
proportional to the square of the input envelope level. The response of an associated radar video
integrator, if any exists, is normally included as part of the detector response for purposes of
analysis. Empirical work [15, pp. 58-62] revealed that the optimum response function for a radar
detector (when the signal-to-noise ratio of targets is small) is:

ψ (v ) = Av 2 = y (7)

where:

ψ = the output of the detector;

v = the amplitude of the modulation envelope of the input;
A = a scaling constant, often 1/2;
Y = the value of a single, discrete output from the detector.

The scaling constant, A, is only a multiplier of a bias level; it does not ultimately affect the Pd for
a radar target. The square-law detector output usually feeds a linear-law video integrator. The
integrator accumulates the results of n samples (such as from n pulses) [15, pg. 58]:

10
The terms square-law, optimum-law, and quadratic detector are all used interchangeably.
11
Noting that even so-called linear detectors are non-linear devices, by definition of detectors.

11
 n n
Y = ∑ y i = A∑ v i2 (8)
i=1 i=1

where:

Y = the output of the integrator;

yi = the amplitude of the i-th sample from the square-law part of the detector, from (Eq. 7).

The square-law detector-integrator is a simple and widely used device in radar receiver designs;
as a default it is usually assumed to be the type of detector that is used in radar receivers. Linear-
law detectors are not unknown, but their performance is generally similar to square-law detectors
[15, pg. 62].

1.9.2 Logarithmic Amplifier Detectors

Some types of radar, such as maritime navigation units, experience a very wide range of echo
amplitudes. This wide dynamic range of target echoes may span orders of magnitude, and will
tend to exceed the response range of a single, linear receiver IF amplifier. To cope with this
problem, logarithmic amplifier detectors are used. In logarithmic amplifier detectors, a cascaded
chain of amplifier stages operates in parallel with a chain of envelope detectors to generate a
receiver output that is proportional to the logarithm of the input envelope. Each stage in the
cascade must have fixed, identical gain. A conceptual block diagram of a logarithmic amplifier
detector is shown in Figure 3.

Figure 3. Block diagram of a logarithmic amplifier detector. The gain value of 10 is for
example only.

The key principle of the design is that each detector is able to operate on the output of a
graduating accumulation of gain. Thus the first detector operates on a small amount of gain that
is suitable for very strong target echoes. Conversely, the last detector operates on perhaps 60 dB
of gain, which will have been totally saturated for strong echoes but which is still adequate to
provide a linear response to extremely weak target echoes. Thus strong target echoes produce
outputs from all detector stages, extremely weak echoes produce an output from only the last
detector stage, and all other echoes produce intermediate numbers of detector responses. The fact
that the detector responses are finally summed means that a logarithmically varying range of
output levels occurs across a very wide dynamic range of target inputs. Along with avoiding

12
saturation problems, logarithmic detectors also reduce the effects of unwanted clutter targets in
radar receivers that do not have moving target indicator (MTI, or Doppler) processing (see
below).

Logarithmic detectors cause a penalty in the probability of target detection that is manifested as a
receiver gain reduction factor [9]. The reduction in the Pd as a function of the number of
integrated pulse-echoes can be characterized as follows [9]: For 10 pulses integrated, logarithmic
receiver loss is about 0.5 dB; for 100 pulses integrated, the loss is about 1.0 dB; and for an
indefinitely large number of pulses integrated, effective reduction due to the use of a logarithmic
detector maximizes at about 1.1 dB.

1.9.3 Zero-Crossings Detectors

Zero-crossings detectors work on the principle that signals can be identified in noise by
extracting information about the zero-crossings of the desired energy. The operative principle is
that the average number of zero-crossings will decrease as signal-to-noise ratio increases. Put
another way, the principle is that signal-plus-noise voltage will tend to exceed the voltage of
noise alone. In a zero-crossings detector, a device is constructed that counts the average number
of zero-crossings per unit time, nave, compares nave to a predetermined threshold value, T, and
then declares that a target is present if (nave < T). Whereas envelope detectors discard phase
information in target echoes, zero-crossings detectors eliminate amplitude information about
target echoes.

1.9.4 Coherent Detectors

A coherent detector is a balanced mixer with a low-pass filtered output. One of the balanced
mixer inputs is a reference signal, generated by an internal reference oscillator, which provides a
signal with some specified frequency and phase. The other input of the mixer is from the last
radar receiver IF amplifier. The detector works on the assumption that the frequency and phase of
the target echo from the IF amplifier output matches the reference signal. The low-pass output
filter allows only the dc and low-frequency modulation components to pass, rejecting higher
frequencies near the carrier. In effect, the carrier frequency is translated to direct current.

Coherent detectors preserve both amplitude and phase information in target echoes, and are
theoretically more efficient than envelope detectors (by about 1-3 dB), especially at low signal-
to-noise ratios [9]. Coherent detectors are seldom used in radar receivers, however, because the
phase of the received signal usually cannot be known or controlled relative to the phase of the
transmitted pulse.

1.9.5 Moving Target Indicator (MTI) or Doppler Processing

In many radar applications it is desirable and even necessary to distinguish moving targets from
all other echo returns. This feature, called moving target indicator (MTI) or Doppler return
processing, is ubiquitously implemented in air search radars. Although MTI processing is not

13
strictly a type of detection, its close integration with radar detection functions makes it
worthwhile to mention in this context.

MTI works on the principle that electromagnetic echo energy emitted by a target that is moving
with respect to the receiver platform will experience a Doppler shift in wavelength (and,
consequently, frequency) compared to the wavelength of the radiation emitted by the radar
transmitter. If echo energy is blocked in the receiver if it has the same frequency as the
transmitted pulses, but is passed through the receiver otherwise, then only targets that are moving
relative to the receiver will be displayed.

One low-cost method for implementing MTI is to lock a coherent oscillator (COHO) to the
frequency of a set of individual transmitted pulses, use the COHO to beat against frequency of
the echo returns from the same pulses, and filter the output with a DC block. This allows only
non-zero beat frequency outputs to pass through the filter and be displayed. More costly MTI
implementations may rely on high-precision frequency references for the generation of
transmitted pulses.

Although MTI processing offers a number of advantages in terms of target discrimination,

especially against clutter, it also has drawbacks. Some radars cannot use it because their PPI
displays must show non-moving targets (e.g., floating debris and shorelines). Because MTI
processing will be impaired by targets that are moving at speeds that cause the Doppler shift of
the return echoes to be an integral multiple of the prf of the radar (e.g., a Doppler shift of 1 kHz
or 2 kHz or 3 kHz, etc., in a radar that is transmitting a 1-kHz prf), radars that are designed for
MTI must normally stagger the intervals between transmitted pulses by intervals that do not
divide evenly into one another. Typically three or more such intervals must be generated between
successive pulses to eliminate these so-called “blind speeds.” Staggering the transmitted pulses,
providing for a precision frequency reference and/or COHO-mixer hardware in the transmitter,
and incorporating MTI filters and software into the receiver adds to the cost and complexity of a
radar. Furthermore, like all target-return processing, implementation of MTI will cause an
effective increase in noise figure and the consequent loss of some desired targets from the PPI
display under various circumstances. MTI processing does not reject interference.

1.10 Radar Features for Detection of Targets in Clutter and Interference

Unwanted radar echoes are called clutter [9, pg. 470] because the output of the undesired echoes
tends to clutter the radar display. Sources of clutter include terrain features and structures, ocean-
surface features, precipitation, flocks of birds, and insect swarms. Small pieces of metal foil
(chaff) intentionally ejected in the atmosphere cause degradation of radar performance by
intentionally creating clutter effects. Regardless of the origin, clutter always tends to degrade
radar performance. An example of clutter is shown in Figure 4.

14
Figure 4. An example of clutter on a 9-GHz maritime radionavigation radar plan position
indicator (PPI) display. The clutter includes local buildings, terrain, and vegetation. No
intentional interference or test targets are present on this display.

Clutter effects may be mitigated by some radar design features. Since clutter sources tend to be
either stationary or slow-moving, they may be somewhat reduced by the use of Doppler filtering
(MTI) that eliminates the effects of echo energy with less than a minimum amount of frequency-
shift relative to the transmitted frequency. However, motion characteristics of birds and insects
do not always allow their echoes to be eliminated by such processing.

If fixed, ground-based clutter due to terrain is severe in the vicinity of a fixed-sited radar (an
example being large hills or mountain ranges that protrude into the radar beam coverage), a
clutter fence that is electromagnetically opaque may be built around part of the radar site. The
fence attenuation may reduce the clutter effects, but will tend to adversely affect the radar
performance in other respects [9, pg. 498].

Otherwise, clutter effects may be reduced generally by the use of the feature of sensitivity time
control, as described below. None of the techniques that reduce clutter will mitigate the effects of
radio interference in radar receivers.

1.10.1 Sensitivity Time Control (STC), or Swept Gain

Sensitivity time control (STC) suppresses the effects of clutter-producing objects that are located
near the radar. STC reduces the dynamic range requirements of radar receivers by reducing
receiver gain just after each pulse is transmitted. This sensitivity reduction, also called swept
gain, is gradually recovered in a controlled manner, often as a fourth-power function of time [12,

15
pg. 13]. The receiver gain, in other words, is reduced for echoes that originate near the radar, and
is increased to a nominal level for echoes that originate at large distances from the radar. Figure 5
shows an example of an STC curve for an air surveillance radar. Although some interference
energy might fall below the STC noise threshold for close-in targets, STC is not, per se, a radio
interference suppression mechanism.

Figure 5. Example STC curve for an air surveillance radar.

1.10.2 Log Fast Time Constant (Log-FTC or FTC) and Comparison with STC

Another important method for reducing the effect of clutter echoes is log fast time constant (log-
FTC, or simply FTC). FTC log-compresses the dynamic range of clutter energy [12, pg. 77] and
therefore reduces the variation in clutter levels at the receiver output. This effect, in turn, results
in a relatively constant false alarm rate (CFAR) with a somewhat inexpensive implementation in
the radar design. CFAR is described in more detail below.

FTC works if the input clutter (or noise) is described by a Rayleigh probability density function.
Sea clutter and land clutter are non-Rayleigh, and thus log-FTC is not so effective against them.
But FTC works well against precipitation clutter, and is sometimes called a weather fix [9,
pg. 506-507].

The effect of FTC is somewhat like automatic STC, but unlike STC it does not suppress clutter
entering the radar receiver through the antenna sidelobes. (Because STC reduces gain when the
transmitted radar pulse energy is still at close distances, it does work against echoes entering
through the sidelobes.) If clutter is strong enough to be received through the sidelobes it may not
be adequately reduced by the log effect of FTC. Therefore, STC and logarithmic receivers are
sometimes used together, with the STC working in the RF part of the receiver and the
logarithmic amplifier working in the radar IF section.

16
STC is range-dependent, while log-FTC is range-independent. Neither technique improves the
ratio of target-to-clutter power. The benefits of FTC are traded against some loss in the
probability of detection of targets. FTC is not a radio interference-suppression feature.

1.10.3 Instantaneous Automatic Gain Control (IAGC, or AGC)

Instantaneous automatic gain control (IAGC, or simply AGC) is yet another technique for
avoiding overload of radar displays by strong clutter blocks. AGC works through a negative-
feedback loop between the IF amplifier output and the gain control of that circuit [12, pg. 78].
The time constant of the AGC is set to a long interval, so that extended clutter effects are
reduced, while the short-response time responses due to echoes from discrete targets are not
affected. AGC is commonly implemented in aircraft-tracking radars that implement conical
beam-scanning and monopulse tracking techniques [9, 11]. The AGC filter loop is a low-pass
design that passes frequencies from dc to just below the antenna beam-scan rate, thus assisting
tracking capabilities.

AGC causes reduction in receiver sensitivity in the presence of strong signals and thus mitigates
radar display overload, but it also decreases the probability of detection of undesired targets. It is
therefore not an interference mitigation technique.

1.10.4 Constant False Alarm Rate (CFAR)

A simple form of analog constant false alarm rate (CFAR) processing was discussed in Section
1.10.2 in connection with log-FTC circuitry. CFAR is often an optional feature for many simple,
analog radar displays, but its implementation is often essential for digitally processed radar
signals [12, pg. 98]. The reason for the discrepancy is the intelligence that human operators bring
to the problem of interpreting radar plan position indicator (PPI) displays, versus the inherent
limitations of digital processing algorithms relative to human intelligence. Human operators can
become skilful at distinguishing true targets from noise and clutter artifacts on such displays, and
if too many false targets (false alarms) are occurring, then the operator can reduce the gain
setting of the PPI display—in effect providing an approximation of automatic CFAR, albeit with
a relatively slow response time [12, pg. 99].

But in automated detection and tracking systems, where the design goal is to automatically
designate targets and display them, the processing algorithm can be overwhelmed by too many
false targets. (A change of only 1–dB in the threshold-to-noise or threshold-to-clutter ratio can
change the false alarm rate by 2 orders of magnitude [12, pg. 99]. Thus an automatically adaptive
CFAR technique is therefore essential for such radars.

In a typical CFAR implementation, a test cell is designated. Around that cell, a group of tapped
digital delay lines are used to estimate the average level of noise or clutter. The tap outputs
(usually 16 to 20) are summed, the sum is multiplied by an appropriate constant, and the result is
used to set the detection threshold for the test cell [12, pg 99, and 9, pp. 392-395]. The output of
the test cell is the radar output. This approach is called adaptive video thresholding.

17
Additional variations of CFAR exist [9, pg. 394]. One of these is a post-detection integrator (a
low-pass filter) that is used to estimate average noise, the output then being fed forward to
control the detection threshold. Another technique, called the Dicke fix, is to use a broadband IF
filter followed by a hard limiter and a narrow matched filter. The hard limiter is set low enough
to limit receiver noise. But signal energy causes the output of the matched filter to increase by a
factor that is equal to the ratio of the bandwidths of the wideband and narrowband filters. A
variant of this approach uses limiters to discriminate between desired target echoes and
interference by observing the change in time of the phase pattern at the limiter output [17, pg.
3.49]. The limiter output is correlated with phase-coded information in the transmitted waveform;
the better the correlation, the more likely the energy is a target echo. This approach, called phase-
discrimination CFAR or coded-pulse anticlutter system (CPACS), destroys all amplitude
information in the echo returns.

CFAR processing has a number of drawbacks. It causes losses relative to optimum detection, and
the number of pulses that are required for processing must be large to minimize this loss. CFAR
inevitably reduces the overall probability of detection of desired targets, and thus will cause some
targets to be lost. These losses can be insidious because operators will often be unaware that
CFAR-controlled detection thresholds are gradually creeping upward as noise, clutter, or
interference increase in some environment [9, pg 395; 12, pg. 99]. In CFAR radars, anti-
interference measures should therefore be implemented (if any are available) ahead of the stage
where CFAR processing occurs.

In summary, CFAR processing is a necessary evil that suppresses clutter and noise effects at the
cost of losing some desired targets. CFAR will suppress interference from low duty-cycle (less
than a few percent), asynchronously pulsed sources (i.e., other radar transmitters). But CFAR
does not function as an anti-interference technique for higher duty-cycle (i.e., non-radar) signals.
In fact CFAR can make the effects of high duty-cycle interference worse by insidiously
suppressing desired targets in the presence of such interference without any awareness of the
problem on the part of radar operators.

1.10.5 Interference Rejection (IR)

As described in [18], many radar designs include an operator-selectable feature called

interference rejection (IR). The purpose of IR is to reject or suppress interference into a radar
receiver from co-channel transmissions from other radars. For reasons that will presently
become clear, IR is not effective against non-radar (communication-type) signals. IR is especially
useful in radar bands in which large numbers of radars are tuned to the same frequency; two of
these cases are 3050 MHz, a popular frequency for S-band maritime radionavigation radars, and
9400 to 9410 MHz, a popular range for the tuning of X-band maritime radionavigation radars.
Without the IR feature activated, the effect of interference from multiple, co-channel radars is
shown in Figure 6.

18
Figure 6. An example of continuous severe co-channel radar interference on a PPI display. The
average I/N level of the interference is 0 dB and the IR and STC features are switched off.
IR circuit designs are often protected as trade secrets, but there are at least two broad approaches
that work on two different principles of pulse-to-pulse correlation: Pulse repetition interval (pri)
discrimination, and pulse width (pw) discrimination [18].

A block-diagram for the pri discrimination IR design is shown in Figure 7. In this approach, echo
pulses in the radar video section are passed through a threshold comparator and then through two
channels via a splitter. In one channel the pulses are delayed by the pulse repetition interval and
in the other channel there is no net delay relative to the first channel. The two resulting outputs
are then ANDed. If the echo intervals match the radar transmitter pri, then the AND gate output
goes high; otherwise the AND output goes low; the AND gate condition is used to pass video
data. The result is that radar pulse sequences that do not match the transmitter pri are suppressed
at the video output. Note that one of the input pulses is lost, and that low-amplitude echoes are
discriminated against. Also, this technique does not enhance desired signals, but only suppresses
undesired signals.

19
Figure 7. Pulse repetition interval (pri) interference rejection circuit block diagram.

A somewhat similar conceptual approach is used for pw discrimination IR. In this design, the
input pulses are differentiated and split into two channels. In one channel the differentiated
pulses are delayed an interval corresponding to desired pulse width, τ, while in the other channel
the differentiated pulses are inverted. If an input pulse is of width τ, the differentiated trailing
edge inverted pulse will coincide in time with the leading edge pulse delayed in time τ. The
coincidence circuit passes signals in the two channels only if they are in time coincidence; if an
input pulse is not of width τ, the two spikes will not be coincident in time and the pulse is
rejected. This technique results in reduced receiver sensitivity and probability of target detection.
Figure 8 shows a radar display (with radar interference present) when IR is deactivated versus
activated.

In addition to pulse-to-pulse IR techniques, some radars also incorporate IR that works on a scan-
to-scan basis. This feature operates by using a memory of apparent targets that is either
reinforced (or not) from one scan to the next. Since false targets generated by stray pulses from
other radar systems will not be expected to occur at the same location on the PPI display from
one scan to the next, such targets may be dropped from the display if no reinforcement occurs on
a scan-to-scan basis. Targets that do continue to occur at the same location on the PPI display (or
at nearly the same location, within some specified bounds of translational movement) are
remembered from previous scans and continue to be displayed.

The IR feature in radar receivers only works against low duty cycle, pulsed, asynchronous
interference. The use of IR is a necessary evil for radar operations, because it causes some
targets, especially ones that are weak, to be lost. For pulse-to-pulse IR techniques, losses occur
due to the comparator threshold setting (in the pri IR design) and because of the reduced number
of pulses that are available for data processing and overall reduced receiver sensitivity with these
circuits. For scan-to-scan IR processing, weak targets that tend to fade out normally in a high
percentage of scans will be lost because of their intermittent behavior. IR is not effective against
communication-signal interference because such interference is normally of much higher duty
cycle than radars. Appendix A expands on the results of IR performance studies that were
performed in conjunction with the test effort described in this report.

20
Figure 8. Radar display with IR turned ‘off’ versus ‘on’ in the presence of interference from
other, gated radar-like pulses. A set of ten desired targets has been injected in both images, at
30 degrees azimuth at the left and at 340 degrees azimuth on the right. The targets are obscured
by coincident radar interference when the IR is turned off (left), but are easily visible through the
same interference when the IR feature is activated (right). IR has been found to be effective only
for low duty-cycle (less than about 1-3 percent) interference, such as generated by radar
transmitters.

21
 2 SETTING CONDITIONS FOR INTERFERENCE MEASUREMENTS
ON RADAR RECEIVERS

For tests and measurements of radar receiver performance in the presence of RF interference, it
has proven crucial to establish baseline conditions for the operation of the radars and the
injection of interference into the receivers. The basic baseline conditions and parameters are
described here. Additional and more detailed descriptions of the test conditions are provided in
later sections of this report, in connection with the discussion of tests and measurements on the
individual radars in this study. Figure 9 shows the typical appearance and conditions of a radar
station control console and PPI display.

Figure 9. Typical conditions inside an air-search radar station, showing the main console for
display and control.

2.1 Interference Coupling Technique

It was essential that the levels of interference in each radar receiver be quantified as precisely as
possible. Such quantification is difficult or impossible if interference is coupled radiatively into
the radar receivers. Radiative coupling also introduces problems with spectrum coordination with
other agencies. There is also always a chance that other signals in the vicinity of the radar might
cause some sort of additional interference in the receiver. Taking all of these factors into

22
consideration, interference signals were injected into all of the radars in this study via hardline
couplings, with desired targets likewise injected at the RF front end via a combiner. The typical
test configuration is shown in Figure 10.

Figure 10. Block diagram of a typical test configuration in this interference study. Interference
and desired targets were generated at radio frequencies and were injected at the radar receiver RF
front-end ahead of the first RF filter and LNA. The spectrum analyzer is an external, diagnostic
test and measurement accessory that is tapped from the IF via a directional coupler.

The hardline connections for interference were always made at the radar RF front end, as close as
possible to the radar antenna, with the radar running in a receive-only mode. In particular, the RF
interference was fed into the radars ahead of the first RF low noise amplifier (LNA). Thus the
receivers were subjected to RF interference that replicated energy that would have otherwise
been fed into the receivers via the radar antennas. This approach ensured that all sections within
the radars behaved normally, the same way that they would have if the interference had been
radiated normally into the radar units.

For the second model of long range air route surveillance radar, as described in Section 3, the
first RF low-noise amplification section consisted of a distributed array of LNAs built on a feed
platform in front of the antenna reflector; it was mechanically impossible to couple the RF
interference into this array in such a way as to feed all of the LNAs. Instead, a hardline
connection was made to the input of the first RF stage ahead of a selected LNA on a selected
element row of the array.

23
 2.2 Calibration of Interference Levels

Whenever possible, the relative levels of interference, I, and receiver noise, N, were directly
measured as I/N ratios within the IF section of each radar. Figure 11 shows NTIA and FAA
engineers assessing such an IF tap point. When no access was available to a radar’s IF section,
then the radar receiver noise figure was measured and the I/N levels were computed. Procedures
for both the direct observation method and the indirect computation method are provided here.

Figure 11. NTIA and FAA engineers determining the locations on a radar IF-stage circuit card
where the I/N level could be monitored.

In all of these I/N calibration procedures, it is critical to distinguish between cases in which the
interference bandwidth exceeds the radar IF bandwidth (wideband) versus cases in which the
interference bandwidth is less than or equal to the radar IF bandwidth (narrowband). The critical
difference between these two cases is that all of the power generated for narrowband interference
will be coupled into the radar IF, whereas only a fraction of wideband interference power will
contribute to the I/N ratio, and account must be taken of that offset (called on-tuned rejection, or
OTR) in the effective value of I in the receiver IF stage.

24
2.2.1 I/N Calibration with Access to the Radar IF for Non-Pulsed Signals

In cases where access to the radar IF was available, the technique for determining I/N levels was
to first connect a spectrum analyzer with average detection to the radar IF output and observe the
radar noise power level in a resolution bandwidth (RBW) matched as closely as possible to the
radar IF bandwidth. Then the interference generator output was fed into the radar RF front end
input and the interference power was adjusted until it caused an increase of 3 dB on the spectrum
analyzer display. Since the spectrum analyzer display showed the quantity [I/(I+N)], this meant
that [I/(I+N)]=3 dB, implying that I=N and therefore I/N=0 dB. With this method, all I/N levels
were referenced to the indicated I=N level. Figure 12 shows an example of the IF output for this
approach.

Because the interference level was directly monitored in the radar IF bandwidth, no special
account needed to be taken of the I power level if the interference bandwidth exceeded the
bandwidth of the radar IF. Whatever the amount of OTR that might occur due to the bandwidth
difference, the direct observation of a 3-dB increase in the IF at some interference power
generator setting provided assurance that the corresponding setting of the generator output panel
was in fact creating a 0-dB value for I/N in the radar IF bandwidth.

Figure 12. Example of a 0-dB I/N calibration noise spectrum (57.55 MHz IF frequency).
Interference is wideband CDMA (W-CDMA). The noise-plus-interference curve is 3-dB higher
than noise-only curve.

2.2.2 I/N Calibration with Access to the Radar IF for Pulsed Signals

For tests in which pulsed interference was injected, the level of the pulsed interference was
adjusted until the peak power from the pulses was observed 20 dB above the receiver IF level. In
that case, [I/(I+N)]=20 dB, implying that I/N was virtually 20 dB. (Care was always taken to be
sure that the pulse peak power level was measured in a bandwidth that was wider than (1/pulse
width) to ensure that all pulse power was convolved into the measurement.) This approach is
depicted in Figure 13. With this method, all interference I/N levels were simply referenced
directly to the level that gave an indicated I/N of 20 dB in the receiver IF stage.

25
Figure 13. Calibration technique for pulsed interference. Peak pulse power is measured in a
bandwidth exceeding (1/pulse width), and peak power for all subsequent testing is referenced to
the injected level where [(I+N)/N]=20 dB, which differs from I/N=20 dB by only 0.04 dB, well
within the uncertainty of the measurement itself.

Although pulsed signals would generally produce emission bandwidths that would be wider than
the radar IF bandwidth, the OTR factor due to the bandwidth offset was physically incorporated
into the observation of effective interference power in the radar IF stage. Thus the direct
observation of the effective I/N level in the radar IF stage obviated any need to compute an OTR
factor for such interference; the signal generator setting that produced a 20-dB increase relative
to the radar IF noise floor was in fact the setting that produced a true I/N level of 20 dB.

The analysis above only applies if it is possible to match the spectrum analyzer resolution
bandwidth (RBW) to the radar IF bandwidth. If it is impossible to make the spectrum analyzer IF
bandwidth match the radar IF bandwidth, then the I/N at the radar IF output is determined by
taking into account the OTR:

⎛I⎞ ⎛I⎞
⎜ ⎟ = ⎜ ⎟ − OTR (9)
⎝ N ⎠ IF ⎝ N ⎠T

where:

(I/N)IF = interference to noise ratio in the bandwidth of the radar IF stage;

(I/N)T = interference to noise ratio that would occur if the radar IF bandwidth were equal to or
wider than the bandwidth of the RF interference produced by the transmitter;
OTR = on-tuned rejection factor.

26
For CW-like signals, the OTR factor is given by:

OTR = 0 for BIF ≥ BT (10a)

OTR = 10log(BT/BIF) for BIF < BT (10b)

where:

BIF = radar receiver IF 3-dB bandwidth (Hz);

BT = transmitter 3-dB bandwidth of the interference signal (Hz).

During the measurements described in this report, the I/N ratio calculated above (taken into
consideration with the OTR) was noted where applicable, along with any associated attenuation
settings. The interference generator power output level and associated attenuation settings that
produced an I/N ratio of 0 dB at the radar IF output was recorded, and other I/N values were
obtained by adjusting the interference attenuator setting.

2.2.3 I/N Calibration without Access to the Radar IF

Almost all radars used in this study were accessible at their IF stages. But in the rare cases in
which no access to the radar IF stage was available, an indirect I/N calibration method based on
the specified radar noise figure and 3 dB IF bandwidth was used; radar noise figure and radar IF
bandwidth were taken from documentation.12

Thermal noise at room temperature (about 290 K) is kTB = -114 dBm/MHz. The receiver
inherent noise level (in dBm) N, at the receiver IF output referred to the receiver RF input is
given by:

N = -114 dBm/MHz + 10log(BIF ) + NF (11a)

or
N = -168.6 dBm + 10log(BIF) + 10log(T) (11b)
where:

BIF = receiver IF bandwidth (MHz);

NF = radar receiver noise figure (dB);
T = radar receiver effective noise temperature (K).

For example, suppose that a radar operated with a 100-ns pulse width. Then a typical IF
bandwidth matching that pulse would be (1/100 ns) = 10 MHz. A good radar receiver noise
figure would be 5 dB. In this case the radar noise power in the IF would be (-114 dBm +

12
Radar IF bandwidth would normally be nearly equal to (1/pulse width) of the radar, and this knowledge was used
to help verify that the specified IF bandwidth was sensible.

27
10log(10 MHz) + 5 dB) = -99 dBm (that would be measured in a 10 MHz RBW). In other values
of RBW, the measured noise power in the radar IF would vary as 10log of this value.

The OTR complication (already described above) arises for interference signals with bandwidths
that are wider than the radar IF bandwidth. Extending the example above, suppose that an
interference signal that has noise-like characteristics (such as code division multiple access
(CDMA)), and which thus varies in power as 10log(RBW) is to be injected into the radar RF
front end. The complication arises because the radar IF is too narrow to convolve all of the power
in the interference signal; in effect, the radar sees only a fraction of the interference power, that
fraction being given by an OTR ratio 10log(Binterference/ BIF).

Suppose, for example, that the CDMA signal bandwidth were 20 MHz and the goal is to inject
this signal into the radar with a 10 MHz IF bandwidth at an I/N ratio of 0 dB. The nominal 0–dB
signal generator power output level would in this case have to be increased by the OTR amount
of 10log(20 MHz/10 MHz) = +3 dB. Account would also have to be taken of all losses between
the signal generator and the RF injection input point, which would typically be on the order of
15 dB loss. The resulting 0-dB I/N is calculated by the equation:

Pgen = Pnoise + OTR + PL (12)

where:

Pgen = power output (panel display level) from the interference generator, dBm;
Pnoise = radar IF noise power level in the bandwidth of the radar IF (dB);
OTR = on-tune rejection factor 10log(Binterference/BIF) dB;
PL = loss between the signal generator output and the radar RF front end input, dB.

Suppose for example that Pnoise = -99 dBm in 10 MHz; OTR = 10log(20 MHz/10 MHz) = +3 dB;
and PL = 15 dB. Then Pgen = -99 + 3 + 15 dB = -81 dBm displayed power output from the
interference generator for an I/N level of 0 dB. All other values would follow from this, as shown
in Table 1. It should be noted that, in this method, all losses and gains between the interference
generator output and the radar RF front end input must be carefully measured.

Table 1. Example of Interference Power Levels When Interference Bandwidth Exceeded Radar
IF Bandwidth
I/N (dB) Path Loss Radar IF Interference OTR Interference Signal
(dB) Bandwidth Signal (dB) Generator Panel
(MHz) Bandwidth Display (dBm)
-6 15 10 20 +3 -87
-3 15 10 20 +3 -84
0 15 10 20 +3 -81
+3 15 10 20 +3 -78
+6 15 10 20 +3 -75

28
2.2.4 Setting I/N Levels from Calibration Data

For all calibration techniques, the various levels of interference were subsequently adjusted to
particular values by attenuating the input relative to the calibrated level. For example, suppose
that the panel display power value that gives (I+N)/N = 3 dB is -84 dBm. This becomes the 0-dB
I/N point. Then a 3-dB reduction in the interference level relative to that point would be an I/N
level of -3 dB, which would occur at -87 dBm output power from the generator. An example of a
complete conversion table as used for typical radar interference measurements is shown in
Table 2.

Table 2. Example of Interference Power Levels When Interference Bandwidth was Less Than
or Equal to Radar IF Bandwidth
I/N (dB) Interference Signal Generator Panel Display (dBm)
-10 -94
-9 -93
-6 -90
-3 -87
0 -84
+3 -81
+6 -78

2.3 Injection of Desired Targets

All of the radars that were tested were designed to detect and display discrete targets, with the
exception of a meteorological (weather surveillance) radar. Degradation effects needed to be
assessed based on target detection effects (loss of targets versus false targets). Therefore it was
critical to ensure that a useful set of baseline targets were provided at all stages of testing on each
radar receiver. For most testing, the need for a good set of baseline targets was met by injecting a
hardline-coupled set of ntarget fixed-amplitude pulses, the number of pulses being:

⎛ θ ⎞⎛ Tscan ⎞
n t arg et = ⎜ ⎟⎜ ⎟ (13)
⎝ 360 ⎠⎝ pri ⎠

where:

ntarget = number of pulses injected for each target;

θ = 3-dB radar antenna azimuthal beamwidth (degrees);
Tscan = time required for a 360-degree radar antenna scan;
Pri = pulse repetition interval (sec).

For example, if a radar with a 1-degree beamwidth generated 1000 pulses per second (for a pri of
0.001 sec) and had a scan interval of 4.75 sec, then ntarget = 13 (rounded to the nearest integer).

29
The injection was done at the radar IF frequency. Usually a set of ten targets were injected in this
way. Figure 14 shows the target and interference injection equipment typically transported by
NTIA engineers to a radar station. Figure 15 shows a block diagram of the target-generation
hardware developed for the study; additional hardware was used to synchronize target injection
with radar PPI scan triggering.

Figure 14. Typical arrangement of NTIA interference and target generators at an air search
radar station. The NTIA equipment is spread across the table in the foreground and the cart in the
background, with the radar PPI display at right.

The set of desired targets was injected on every radar scan.13 In most tests the targets were
injected synchronously with a scan trigger from the radar set, thus keeping the targets at fixed
locations on the radar PPI displays from scan to scan.14 In tests in which scan synchronicity was
not possible to achieve, target positions shifted a few degrees in azimuth from one scan to the
next.

On each scan, at least ten equally spaced test targets were usually injected at equal power levels
in a sequence that placed them radially on the radar PPI display, as shown in Figure 16.15 For real
targets, this radial distribution would imply a change in distance between the radar site and the

13
In this report, “scan” means a single, 360-degree rotation of the radar beam.
14
For a few radars, no such trigger was available and operator-adjusted synchronous timing was used
instead.
15
In some tests, a set of four such radials were injected during each scan at angular intervals of ninety degrees,
providing forty targets per scan.

30
targets, with a concomitantly increasing cross section with distance. But for the purpose of the
NTIA test and measurement series, this arrangement was not meant to replicate effects due to
changes in distance. Instead, it was simply an expedient to ease the difficult process of counting
targets during interference runs; radially distributed targets were found to be easier to count than
equal-radius, sector-distributed targets.

Figure 15. Block diagram of the radar target-generation hardware developed and used for the
study.

Figure 16. Example of a maritime radar PPI display during interference testing with ten
injected targets on a single radial. Clutter condition was used diagnostically to verify proper radar
operation.

One radar, a long-range air surveillance system, incorporated a built-in target generator and an
automated target counter; NTIA engineers took advantage of those features during the testing.
Those targets were displayed throughout sector wedges on the PPI display as shown in Figure 17,
but again the target cross sections were constant; no distance effects were incorporated. As was
the case for the NTIA-generated targets, the radar PPI display arrangement in sector wedges was
simply a convenient way for the system software to arrange the targets on the screen.

31
Figure 17. Sector-wedge distribution of desired targets on a PPI display during testing on a
long-range air search radar. Targets are shown as “+” signs, and are generated at the radar RF
stage.

Some live-target testing was done within range of a busy airway with a long-range air
surveillance radar in the Midwestern area of the US. Although a test phase at that radar included
the use of standard NTIA-generated desired targets, the live-target phase of the testing allowed
the observation of interference effects against actual aircraft in flight, ranging from commercial
airliners to small airframes.

For meteorological radar tests and measurements, discrete targets could not be used because
those radars are designed to monitor extended atmospheric phenomena. For those tests, actual
weather echoes and radar internal noise were used in the course of the test program, as described
more fully in Sections 5 and 7.

2.4 Use of Fluctuating Versus Non-Fluctuating Targets in Radar Testing

The question naturally arises whether tests of the effects of low-level interference in radar
receivers should be performed with constant amplitudes for the desired targets, versus against
desired targets with fluctuating levels. It can be argued that fluctuating targets might resemble
real-world targets most closely, and thus should be used. But this argument has been found to be
incorrect. The reasons are as follows:

32
 1) Real targets do not fluctuate in accordance with the mathematical models (called
Swerling models or cases [9, pp. 46-52]) that are used to design microwave radars. There
is no general-purpose mathematical description of target fluctuation that works for a
general class of radar targets under all operational conditions. The Swerling models are
used because they are analytically tractable and provide some resemblance to the range
and rate of variation that might be seen in some real targets.

2) Although radars are designed to use a sufficient number of pulses per target per scan to
accommodate some amount of target-power fluctuation, this does not necessitate the use
of power fluctuations during testing of radar performance. Rather, testing can be just as
well performed against a fixed, nominal target power level. Such a level would represent
the behavior of many real targets, which have been observed by the authors to fluctuate
little or not at all from one scan to the next in many cases.

3) During testing, radar receiver performance must be compared in two conditions:

interference being present versus no interference being present. A meaningful comparison
of these two conditions means that a baseline target detection performance level must be
established when no interference is present. Experience of NTIA engineers has shown
that such a baseline is easy to establish if target power level in the receiver is held
constant, but is difficult to achieve if target power levels fluctuate about that nominal
value in accord with the Swerling models. The problem is that, if the target power
fluctuates according to, for example, Swerling case 1 [9, pp. 46-47], then (roughly) half
of the targets nearly burn the phosphor on the PPI display, while the other half (roughly)
are simply absent. Therefore the baseline level of detection becomes approximately 50%
for such targets. But meaningful performance tests need to be performed with a target
probability of detection that is closer to 90%. To achieve 90% performance with Swerling
case 1 fluctuating target distributions, as shown in Figure 18, the mean value of the
fluctuating targets must be increased by perhaps 10 dB or more. This approach would be
problematic because it would be equivalent to compensating for target fluctuations by
turning up the radar transmitter power by 10 dB, which is exactly what operational radars
cannot do to enhance target returns. Furthermore, if this step is taken with a target
generator in actual testing, then most of the target returns nearly burn the phosphor on the
PPI, which again is not a realistic scenario.

4) Observation of genuine targets, both aircraft and boats, by NTIA engineers indicates that
non-fluctuating test targets more nearly resemble the behavior of real targets on radar PPI
displays. Furthermore, non-fluctuating test targets can be adjusted to a 90% probability of
detection and that level can be used as a reliable baseline through all testing on a radar
unit.

For all the reasons above (maintenance of a constant non-interference performance baseline,
replication of the behavior of genuine airborne and maritime radar targets, and overall
practicality of testing), NTIA testing has mostly used constant-amplitude test targets that are held
at a power level at which they are detected 90% of the time on the radar PPI display when no
interference is present. An exception has been made for meteorological radars, which do not
detect discrete targets, and in which actual atmospheric echo returns have been used.

33
It is worth noting that by setting the probability of detection of the test targets to 90%, NTIA
engineers provided substantially stronger target levels for the radar receivers than might have
been justified by the 80% probability of detection criteria that many reference sources [9-14 and
19] quote for purposes of adequate radar design. The two probabilities might not be comparable
because setting a design criterion is not the same as setting an interference-test performance
criterion. The purpose of the NTIA tests and measurements was to determine upper limits for
interference levels that produce deleterious effects in radar receivers. In other words, the purpose
was to test interference effects against targets that were undoubtedly high-quality, but still within
the range of credibility for real radar echoes. If an interference level with some modulation was
found to affect the detection of targets that had a 90% probability of detection in the absence of
interference, then the detection of almost any real targets by the same radar receiver would be
expected to be adversely affected by the same interference level of that modulation. Put another
way, with 90% probability of target detection in the absence of interference, the tests were
structured in the direction of accepting higher rather than lower levels of interference before
radar receiver performance was adversely affected.

2.4.1 Overview of Fluctuating Target Level Generation

Even considering all of the arguments above, some interference tests were performed on a
maritime radar with fluctuating targets. Those test procedures and their results are specifically
described in Section 6, but an overview of the fluctuating target approach is as follows.

As seen in Figure 18, the behavior of logarithms causes the values in each distribution to
naturally be below 0 dB (and further below 0 dB) more often than they are above 0 dB. This is
consistent with the physical reality that reflected power levels from radar targets may increase
only a finite amount above an average, but may be reduced by an infinite amount below an
average.

A baseline Pd curve needs to be determined without any interference being present and with the
target power levels fluctuating. To assess the effects of interfering signals when the targets are
fluctuating, the Pd is determined when the I/N level is set to an array of values such as: -9 dB,
-6 dB, -3 dB, 0 dB, +3 dB, +6 dB, and so on. This approach allows the fluctuating target baseline
Pd to be compared with the steady target power Pd and just as importantly allows comparison
between the Pd when the targets are at a fixed power level at some I/N level to the Pd when
targets are fluctuating under the same I/N condition. A key element is to use the same fluctuating
target power levels with and without the presence of interference signals.

In choosing target fluctuation values for the tests, the lower part of Figure 18 can be used to
select the decibel power values relative to the median target power level, which is 1.6 dB below
the mean value. The rationale for this offset can be gleaned from standard texts [e.g., 17, pg. 46-
52]. The set of test values are then selected from the curve. For example, the -5.5 dB value has a
cumulative probability of about 0.175. So if the target power that gave Pd=0.90 was -75 dBm
from the testing signal generator, then the value for that cumulative probability would be
-80.5 dBm. Twenty fluctuating target values from the curve will be selected from the figure, an
example group being listed in Table 3. One value will be selected for each scan. On each scan, all
of the injected targets (usually ten) will thus have a common value.

34
Figure 18. Statistical distributions for Swerling case 1 fluctuating radar target power. If the
average level is set to the same value as required for 90% probability of detection of non-
fluctuating targets, then most of the targets are not visible on the radar display.

35
Table 3. Fluctuating Target Power Levels Derived from the Curves of Figure 18
Signal levels (dB)
Relative to median power level
-14.4 0.30
-9.5 0.91
-7.2 1.5
-5.6 2.1
-4.3 2.7
-3.3 3.3
-2.5 4.0
-1.7 4.8
-0.90 5.7
-0.30 7.3

At the end of a test run at a given I/N level, (usually twenty scans altogether), the total number of
observed targets would be divided by the total number of injected targets (say 200) to obtain a
single data point for the Pd of fluctuating targets under the interference condition being tested.
Then the interference I/N level would be changed (its value would be either raised or lowered, or
the modulation would be changed), and the entire sequence of fluctuating target levels would be
used again for another set of (usually twenty) scans at the new interference level.

2.4.2 Test Methods with Fluctuating Targets

Specifically, the procedure for interference testing with Swerling Case 1 targets was as follows:

First, the zero mean power level for the targets was set to the value that produced a target Pd of
0.90 without interference present and steady target power levels. Then the variation of the target
power level from the value that gave the 90 percent Pd was chosen from Table 3. The target
power was set to that level and the targets were counted. The power of the groups of pulses that
defined a target was set to cycle through the twenty entries in the list of power values (relative to
the mean value) contained within Table 3. A Pd count was made by observing the PPI through
twenty scans.

Next, the I/N level was set to one of the following values (-9, -6, -3, 0, 3, 6, 9, 12, 20, or 40 dB)
using one of the undesired waveforms and the target generator was cycled through the same
varying set of power levels for the chosen sigma; the Pd was computed for those twenty scans,
and so forth.

2.5 Target Identification and Counting During Tests and Measurements

Although many radars, including those used in the NTIA test and measurement series, provide
automatic target identification and tracking features, NTIA engineers determined during the tests
that a human operator (one of the authors, Sanders) was more skilled at identifying and
discriminating radar targets than the automated routines built into the radar processors. Many

36
marginal targets that would be rejected by automated radar processing algorithms were easily
identifiable by the human operator, and false alarms were more easily rejected. Most of the tests
were therefore run without automated target detection, and the targets that were observed were
counted, scan by scan, by the human operator. An additional benefit of using the human operator
to identify targets and reject false alarms was that the operator brought the same acceptance
criteria to bear on all of the target displays of all the radars in the tests. Figure 16 shows the
typical appearance of radar PPI displays under test conditions.

The result was that uniform criteria were applied to target identification on all of the radars
(except meteorological, where discrete targets were not used), and the effects of interference
were more flexibly allowed to occur before targets were declared to be degraded than would have
been the case if automated target identification features had been used. That is, thresholds for the
effects of interference were documented at higher interference levels than would have probably
been allowed if automated processing of targets had been performed. As was the case with 90%
probability of detection for targets in the absence of interference, this structured the test results in
favor of the occurrence of higher, rather than lower, interference levels before performance
degradation was documented.

2.6 Radar Receiver Parameter Settings

Every radar used in the NTIA tests and measurements had to be configured for some sort of
nominal operation throughout every test series. The most critical settings that had to be
considered for all of the discrete-target radars were STC, AGC, and CFAR. (For the
meteorological radar, these features were not applicable, and the configuration was adjusted by
site personnel who were intimately familiar with the radar behavior, and who adjusted the radar
to function nominally in the presence of the local atmospheric and weather conditions.)

Since STC is used to suppress close-in clutter, and since no such clutter could occur due to the
hardline-coupled nature of both the RF interference and the injected test targets, the STC feature
was normally disabled during radar interference measurements unless otherwise noted in the
descriptions of individual tests.

As for AGC, this feature was normally activated during tests in which interference was gated on
momentarily during each radar scan, but AGC was deactivated during tests in which interference
was continually present during every radar scan. (The first case corresponds to desired target
displays on one radial or a few radials, and the latter case corresponds to situations in which the
desired targets were displayed in broad sectors throughout radar PPI displays.) AGC was
activated for the momentary-interference testing because such interference behavior is expected
to replicate ordinary coupling into a radar antenna mainbeam, and AGC could be expected to
respond normally (which is to say minimally) to such momentary interference. But in the case of
continually-present interference (which would not be expected in the real world except in
extreme cases, where the interference would couple strongly into the radar antenna sidelobes),
the AGC had to be deactivated because otherwise the AGC feature would decrease the radar gain
to a very low value, thereby effectively desensitizing the radar and resulting in the loss of most or
all of the desired targets. Such a condition would become a test of AGC behavior rather than of
radar receiver performance; the goal of the tests was to determine more inherent system

37
responses, rather than the response of the AGC feature to interference. (Note, too, that AGC, like
STC, is not an interference-suppression feature.)

Regarding CFAR, this feature is most useful when automated target identification routines are
invoked. It was usually activated during testing because it is used so ubiquitously in the
operational modes of most radar receivers. In some cases, CFAR processing can be controlled by
manipulating the number of guard cells, but the settings for this feature are usually not changed
much during the operational life of most radars.

Considering IR functionality, this feature was sometimes toggled on and off during testing (for
radars that incorporated this feature, such as the maritime units) to demonstrate the effectiveness
of the IR features in the radar system. Because testing revealed that IR was not effective at high
duty cycles, some tests were performed to determine the threshold at which IR functionality was
impaired. The behavior of IR in radar receivers is further described in Appendix A.

38
 3 INTERFERENCE MEASUREMENTS ON LONG-RANGE AIR SEARCH RADARS

3.1 Introduction

Long-range air surveillance radars routinely search for, and track, airborne targets at distances
exceeding 200 nmi. They sometimes incorporate meteorological (weather) surveillance
capabilities as well. These radars perform critical missions for air traffic control and national
security. Although their designs vary, they tend to be similar in some vital receiver characteristics
such as noise figure. For this study, NTIA engineers worked closely with personnel from other
US Government agencies16 to determine the susceptibility of two differently designed types of
long-range air surveillance radars. Both of these radars operated in the so-called L-band, on
individual frequencies assigned between 1250-1350 MHz. In this report these radars are referred
to as Long Range Radars 1 and 2.17

3.2 Description of Long Range Radar 1

3.2.1 General Description

Long Range Radar 1 detects weather and aircraft within a radius of about 200 nautical miles (370
km) for the FAA and other Federal agencies. It provides data concerning the location and
strength of weather as well as range, azimuth, and altitude of aircraft within its search volume. A
separate beacon interrogator system that is incorporated into the radar design provides additional
information on aircraft beacon codes, aircraft-encoded altitudes, and emergency status18 of
aircraft.

The radar has two channels, denoted as F1 and F2, which require at least 25 MHz of frequency
separation.19 Two frequencies are provided for the purpose of compensating for atmospheric
fading, distortion, and other effects on any one frequency; effects that degrade one frequency are
not expected to affect the other channel. Two-frequency capability is not provided for the
purpose of using one frequency to compensate for interference effects on the other frequency.
This study did not address radar performance on two channels when one is affected by
interference.

Two receiver processing channels, designated A and B and corresponding to each of the two
operational frequencies, are provided. Each channel consists of a synchronizer, a frequency
generator, a transmitter, weather and target receivers, a receiver processor, and a digital target
extractor (DTE). The radar uses frequency separation and orthogonal polarization for the two
channels so that they can transmit and receive via the same antenna for the diplex mode of
operation. The radar transmits RF energy on a low beam, and receives reflected RF energy on

16
Including the Federal Aviation Administration and the US Air Force.
17
Long Range Radar 1 is the air route surveillance radar model 3 (ARSR-3), and Long Range Radar 2 is the
ARSR-4.
18
Beacon code, altitude, and emergency status are provided via an associated secondary surveillance radar operating
at 1030 MHz and 1090 MHz. Interference to that out-of-band system was not considered in this study.
19
The tested radar installation operates at 1292.08 and 1330.20 MHz.

39
both the low beam and a high beam. The high beam is used to detect aircraft at short ranges and
the low beam is used to detect aircraft at longer ranges and also weather.

3.2.2 Receiver Processing

Each channel in the radar receiver has both normal and moving target indicator (MTI) 20 video
available. The normal and MTI video both have identical constant false alarm rate (CFAR)
processing. Both channels contain independent CFAR circuitry. The CFAR samples the input
signal at 1/8 mile intervals, from 5/8 mile before the target to 5/8 mile after the target, to
determine the mean noise level from the appropriately delayed main target signal.

Long Range Radar 1 also contains integrator-adder circuitry that functions on the normal and
MTI video. The integrator-adder contains a digital video integrator. The integrator digitally sums,
into a single output, all of the target-hit video occurring within the 1.1-degree (3 dB points)
azimuth beamwidth of the radar antenna. Because there are about 12 transmitted pulses within
that beamwidth, there can be approximately 12 consecutive target hits or echo returns at the same
range and azimuth. Also, since the integrator operates synchronously with the radar’s transmitter,
asynchronous pulses (from other nearby radars) are automatically eliminated from the video
output.

Independent receiver processors in channels A and B are electrically identical. Each receiver
processor contains a target processor function. After detection the target information is digitally
processed into suitable form for identification by the DTE. When the radar is operating in a two-
frequency (called diplex) mode, processed target video data in each radar channel are digitally
summed with the adjacent channel target video data before distribution to the DTE. This cross-
channel target video summation feature enhances target detection. Cross-channel video
summation during dual channel operation improves the probability of target detection since a
fluctuating target fade in one channel should not be experienced in the opposite channel due to
frequency-polarization diversity between the two channels.

3.2.3 Antenna Characteristics

Long Range Radar 1 uses a dual horn fed parabolic reflector enclosed in a radome and mounted
on a dual drive pedestal. The antenna forms two cosecant-squared beams shaped for additional
high elevation gain. The two beams are almost identical. The azimuth beamwidth of both beams
is 1.1 degrees at the 3 dB points. The upper beam has coverage from 3.6 to 44 degrees in
elevation while the lower beam covers 2 to 42 degrees in elevation. The antenna rotates at 5
revolutions per minute (rpm). A complete list of the relevant radar technical parameters is
contained in Table 4.

20
Moving target indicator is a Doppler processing output with a velocity threshold. It reduces clutter by eliminating
the display of target returns that are moving at less than a critical threshold velocity.

40
 Table 4. Technical Characteristics of Long Range Radar 1

Parameter Value
Frequency range 1250-1350 MHz
Range 200 nmi
Altitude coverage Up to 18,300 m (60,000 ft)
Distance resolution 0.25 nmi
Azimuth resolution 2 degrees
Receiver IF bandwidth 420 kHz
Pulse width 2 µs
Noise figure 2 dB
Pulse repetition frequency 310 – 364 (8 discrete prf values)
Receiver noise (calculated) level in receiver bandwidth -115 dBm
Sensitivity with normal processing -115 dBm
Sensitivity with moving target indicator -112 dBm
False alarm rate 10-4 for a 2-m2 cross section
Antenna type Dual horn reflector
Antenna mainbeam gain Low beam: 34.5 dBi
High beam: 33.5 dBi
Antenna elevation beamwidth cosecant2 3.6º to 44º degrees
Antenna azimuth beamwidth 1.1 degrees
Antenna scan rate 5 rpm (12 sec/rev)
Antenna polarization Linear and circular available

3.3 Test Approach for Long Range Radar 1

3.3.1 General

The radar’s performance was monitored by observing targets on the radar’s PPI display and
through the use of the radar’s built-in target counting software. Desired targets were generated
using RF signal generators and additional testing was accomplished using live traffic. In addition
to the desired target signals, signal generators were used to inject CW and binary phase-shift
keyed (BPSK) interference into the radar receiver. Radar Pd performance was evaluated as a
function of the calibrated I/N ratio at the IF output of the radar receiver.

As noted above, the radar could operate on either of channels A and B singly, or on both
channels simultaneously. Target power was adjusted to achieve a Pd value of 0.9 in the absence
of interference on a single channel. The same target power was maintained during both single-
channel and dual-channel testing on the radar. As a result, the Pd value was close to 1 when the
two channels were used simultaneously. Interference tests were performed in both single-channel
and dual-channel operational modes.

41
3.3.2 Injected Targets

In order to speed testing, a total of forty injected test targets (ten targets on each of four separate
radials) were separated in range to allow for easy determination of missed targets. They were
monitored for five consecutive antenna rotations for a total of 200 injected, desired targets
examined for each data point. Each target, regardless of range, was set to the same power at the
receiver input, making each target equivalent in terms of receiver Pd. The target power was
adjusted to a level that provided an average target Pd in the absence of interference of about 90
percent.

The target Pd, based upon 200 attempted target scans, was based on manual counting from the
radar PPI display by one of the authors (Sanders). Although information about target Pd on a per-
pulse basis was collected by the radar’s internal data collection software, that information was
not used in this study because the radar’s overall performance in the presence of interference was
the topic of study. (Any one target or “blip” on the radar screen also has an intrinsic Pd for each
individual pulse within the group of pulses that defines that target.) The target generator
produced about 12-13 pulses per target. Whenever the radar integrated enough of those pulses to
define a target and produce a “blip” on the display, it was counted as a good target. As a result,
the actual Pd per pulse in the group of pulses that defines the target could have been lower than
90 percent.

In order to determine the equivalent radar cross section (RCS) represented by injected test
targets, a measurement was made at the end of the test cable connecting the signal generator and
the waveguide. Accounting for the waveguide coupler and other losses between the injection
point and the receiver, the target signal level into the receiver IF was -107.2dBm.

The radar requirement21 is for detection at main beam gain of a 2 m2 target out to at least 195
nmi. One form of the basic radar equation (Eq. 1) is:
1
⎡P G A σ ⎤ 4
R = 0.282 ⋅ ⎢ t t e 0 ⎥ (14)
⎣ Pr ⎦

where:

Pt = peak transmit power, watts;

Gt = transmitter low beam gain, linear term;
σo = test target cross section, m2;
Ae = effective antenna aperture, m2;
Pr = received target signal level, watts;
R = target range, m.

In addition, the Long Range Radar 1 specification allows for a total of 4.1 dB loss in the receiver-
to-antenna path. This same loss is also present on the transmitter-to-antenna signal path. For the
tested configuration, then:

21
FAA-E-2483b, paragraph 3.4, of the radar System Performance Requirements document.

42
Pt = 5 MW – 4.1 dB loss = 1.945x106 W;
Gt = alog(34.5dB/10) = 2818.38;
Ae = Gtλ2/4π = 12.07;
Pr = -107.2 dBm + 4.1 dB = -103.1 dBm = 4.898x10-14 W;
R = 195 nmi = 361235 m.

Rearranging the equation to solve for the test target cross section for a received power of
-107.2 dBm, (Eq. 14) becomes:

⎛ R ⎞ 4 ⎛ Pr ⎞
σ0 = ⎜ ⎟ ⋅⎜ ⎟ (15)
⎝ 0.282 ⎠ ⎝ Pt Gt Ae ⎠

σ0 = (361235/0.282)4 /[(1.945x106)(2818.38)(12.07)/(4.898x10-14)]

σ0 = 1.99 m2.

3.3.3 Live Targets

A limited amount of data were also collected using live ‘targets of opportunity,’ For that testing,
the radar was placed into its normal operational mode and the PPI screen was monitored for
detected aircraft in the radar coverage area. Though the level of these targets was uncontrolled
(being dependent on the interaction geometry between the radar and the aircraft itself), the local
FAA air traffic control center was contacted to determine the aircraft type for specific targets of
interest on the PPI display.

3.4 Undesired Signals in Long Range Radar 1

3.4.1 General

RF signal generators and arbitrary waveform generators (AWGs) were used to generate CW
interference and also the following BPSK signals:

1) BPSK at 0.511 MBit/s,

2) BPSK at 1.023 MBit/s,
3) BPSK at 5.11 MBit/s,
4) BPSK at 10.23 MBit/s.

Spectra of the injected signals are shown in Appendix B. BPSK signals contain data bits that are
encoded into symbols that a spreading pseudorandom code further breaks into chips. However,
the radar receiver does not discriminate between phase changes representing either a chip or a bit
for this type of interfering waveform. The radar receiver processes the BPSK signals as band-
limited constant amplitude noise sources, which fall into all of the range cells. Since the noise-
like BPSK signal falls into all of the range bins, the CFAR processing cannot eliminate it and the
CFAR raises the target detection threshold.

43
The emission spectrum for each BPSK interference signal was measured and recorded, and tuned
to the frequency of the radar channel(s) under test. The BPSK signal generator had the capability
to generate I/N ratios of -12 dB to +30 dB at the receiver RF input; however testing was usually
accomplished only over the subset of that range where meaningful Pd data could be collected. For
each signal type, calibrations were performed to allow for conversion between signal generator
settings and the resultant I/N level using the process described in Appendix C.

3.4.2 Duration of Interference Signals with Injected Targets

For the injected desired signal tests, the undesired signals were also injected into the radar at the
RF input path to the receiver with durations equal to the main beam dwell time and overlaying
them on the desired targets at the same azimuth. As shown below, the dwell time for the main
beam of the radar’s antenna is 0.04 seconds through a stationary object. This dwell time was used
as the duration of the BPSK interference source for the simulated target tests. For any live target,
this interval may be different due to the motion of the target.

antenna _ beamwidth
dwell _ time = ⋅12sec = 0.04 sec . (16)
360deg

3.4.3 Live Target Tests

For the live target tests, as for the injected target tests, the undesired signals were injected into
the radar at the RF input path to the receiver (see Figure 10). Because of the mobile
characteristics of the live targets however, the injected targets maintained their positions for
longer intervals. In particular, the injected targets were controlled so as to cover approximately a
40-degree sector of the radar scan. In addition, for a given set of live targets only a single level of
I/N could be tested. These tests did not attempt to simulate a potential interference source’s exact
behavior as it might move through the radar’s antenna beam at any particular azimuth. The actual
dwell time of such an event (and its possible periodicity) would depend upon both the proper
motion of the source and the characteristics of the radar antenna beam scanning through space.

3.5 Test Procedures on Long Range Radar 1

3.5.1 Injected Target Tests

For the injected target tests, the desired signal targets were overlaid with BPSK interference
signal at a given I/N and observed for 5 complete scans. This resulted in a total of 200 possible
targets (5 scans x 10 targets/radial x 4 radials/scan), and probability of detection was calculated
as the number of observed targets divided by 200. The test was then repeated at a different I/N
level. For these tests the transmitter was turned off and though the antenna was still rotating and
receiving external signals, a spectrum analyzer measurement showed that none were present in
the IF stage (see Figure 10). When testing was completed for a given BPSK signal type, a
different BPSK interference modulation was used and the testing was repeated. For each I/N
level and BPSK modulation, testing was performed with:

44
 • interference and targets on one channel and the other channel disabled;
• interference and targets on both channels;
• baseline “no interference” measurements were performed before and after each
“interference” data set.

3.5.2 Live Target Tests

For these tests, since the target amplitude was largely uncontrolled, the procedure followed was
to operate the radar without injected interference for 15 scans, then turn on the injected BPSK
interference at a fixed I/N for 15 scans, and finally turn off the injected interference for another
15 scans. This allowed for examination of live targets within the BPSK-interfered sector to
determine whether there was any discernable impact to the target due to the BPSK interference.
The targets were recorded from the PPI display to a radar data file for all 45 scans. Upon replay
of the data file to a laptop computer, it was observed that the BPSK interference manifested itself
as a reduction of a ‘reinforced target’ (i.e., a primary, skin-track target that had a secondary
surveillance radar (SSR) beacon reply superimposed upon it). The skin track part of the
reinforced target would fade, but the SSR part, which operates at 1030 MHz/1090 MHz, was not
affected.

For targets of interest where a skin-track was lost but the beacon reply was still present, the type
of aircraft that was being tracked was determined by contacting the local FAA air traffic control
center (ATC) in Kansas City and providing them with the associated beacon code. The ATC was
then able to use the beacon code to provide the flight number and type of aircraft to the authors.

3.6 Results from Interference Tests on Long Range Radar 1

3.6.1 Injected Targets with Single Channel Operation

A plot of the target Pd versus the I/N ratio for the radar operating in single channel mode with
injected targets and BPSK interference is shown in Figure 19. The figure shows that as the I/N
ratio increases, the target Pd decreases. At an I/N level of -9 dB the Pd curves are beginning to
drop perceptibly. At -6 dB I/N the target Pd has dropped substantially below the baseline value of
0.9. Detection probability of simulated radar targets (of the minimal acceptable cross section) for
single channel operation was degraded by about 0.15 (from 0.90 to 0.75) at the I/N level of -6 dB.

3.6.2 Injected Targets with Dual Channel Operation

A plot of the target Pd versus the I/N ratio for the radar operating in dual channel mode with
injected targets and BPSK interference on both channels is shown in Figure 20. The figure shows
that as the I/N ratio increases, the target Pd drops. Figure 20 shows that though the baseline target
Pd per channel was set to the 90 percent value, the overall baseline target Pd is reinforced
(improved) with dual channel operation. But dual channel operation only makes the radar

45
perform better to a point; at I/N levels of -6 dB, the target Pd has dropped below the baseline
value. Detection of simulated radar targets (of the minimal acceptable cross section) for dual
channel operation was degraded by about 0.10 (from 0.9 to 0.80) from the improved dual-
channel baseline at I/N = -6 dB.

Figure 19. Long Range Radar 1 single channel operation Pd with BPSK interference.

Figure 20. Long Range Radar 1 dual channel operation Pd with BPSK interference.

46
3.6.3 Results of Live Sky Tests

The 10 MBit/s BPSK waveform was injected into the radar’s RF circuitry between the antenna
and the receiver (as shown in Figure 10, but with the waveguide connected to the antenna). The
I/N ratio was set to +20 dB for a 40 degree wide azimuth wedge between 160-200 degrees.
Figure 21 shows a recording of the aircraft traffic for a full 360 degree view of the PPI display
for 45 antenna rotations (scans). Figure 21 represents a 9-minute recording of both skin tracks
and beacon tracks from aircraft within the vicinity of the radar. Interference was present during
scans 16-30. The figure shows aircraft that were tracked with the radar and beacon. Tracks that
are dark represent targets that have both radar track and beacon data. Track segments that are
light represent targets with only beacon data.

Figure 21. Cumulative PPI display of Long Range Radar 1 for 45 scans during interference in
the indicated sector.

47
Figure 21 shows one aircraft (identified by flight number via the beacon data) within the 160-200
degree wedge of interest that had radar and beacon data for the initial 15 scans (no interference),
then only beacon data for the next 15 scans (10 MBit/s BPSK interference), and finally radar and
beacon data for the final 15 scans (no interference). The maximum range on the PPI display is
200 nmi. The figure shows that the simulated BPSK interference caused the radar to lose track of
this target. The radar also dropped other targets, but the aircraft flew outside of the ATC area
before the aircraft type could be identified. Figure 21 also shows that other parts of the PPI
output (without known interference) display some aircraft that have intermittent radar data along
with their beacon tracks. These are aircraft that may not be fully in the radar’s antenna
beamwidth due to their altitude and range.

Figure 22 shows the details for the three aircraft of interest. Flight 2211 was an MD-80 observed
before, during, and after the interference was injected. The aircraft’s initial track is dark, then it
goes light, and then it goes dark again. The points where the tracks transitioned from dark
represent the circumstance in which interference was turned on. There was a two-scan lag
between the moment that interference was switched on or off and the moment at which the
resulting effect was observed on the display. Flights 6530 (DC-8 Type 7) and 0574 (TLF-04)
entered the interference wedge when the interference was already on. Their tracks begin light
(which indicates beacon-only returns being received during interference) and then went dark
when the interference was removed (indicating that skin tracking had been restored at that point).
The aircraft in question tended to be flying toward or away from the radar station during these
episodes. Returns from small aircraft tended to be affected at shorter ranges from the radar than
those from larger aircraft, as shown in Figure 22.

Figure 22. Details of Long Range Radar 1 PPI display during live target interference tests.

48
3.6.4 Test Results from a Second Installation of Long Range Radar 1

Interference tests with injected, desired targets were performed on a second installation of the
Long Range Radar 1 model that was at a second location in the U.S. The test protocols were the
same as described above, although due to time constraints fewer interference levels were tested
on that radar. The results of those tests are shown in Figure 23. Performance was decreased
markedly at I/N=-6 dB. The individual channel Pd’s had been adjusted to 0.9, and therefore the
cumulative, dual-channel Pd values in Figure 23 (which are normalized to the single-channel Pd
values) are higher than 0.9.

Figure 23. Test results with injected targets at a second installation of the Long Range Radar 1
type.

3.7 Description of Long Range Radar 2

3.7.1 General Description

Long Range Radar 2 is a coherent, non-linear FM-pulsed system that uses a parabolic reflector, a
phased-array feed horn, and all solid-state electronics to meet the needs of both defense and air
traffic control. It detects targets at up to 250 nmi (400 km) at an altitude of 100,000 feet. The air
traffic control requirement for target detection is Pd=0.8 or more in clear air for a target cross
section of 2.2 m2 at a range of 200 nmi. To satisfy the defense requirement, the radar was
designed for very high reliability (e.g., designed for 1500 hours of mean time between critical
failures that would interrupt service) and high sensitivity (including the capability to detect
0.1 m2 targets at a distance of 92 nmi against a Sea State 5 clutter-background). The solid-state
transmitter power amplifiers are located below the antenna rotary joint to minimize out-of-

49
service time for maintenance, as the amplifiers can be maintained without halting antenna
rotation. The designs of the antenna, the transmitter, and the receiver signal processing hardware
and software have all been integrated to optimize the radar’s functionality for both defence and
air traffic control requirements. A separate beacon interrogator system that is incorporated into
the radar design provides additional information on aircraft beacon codes, aircraft-encoded
altitudes, and emergency status of aircraft. General characteristics are summarized in Table 5.

The radar is normally operated in a diplexed (two RF channels) mode, denoted as F1 and F2.
Unlike Long Range Radar 1, this radar’s CFAR is not independent between the two channels.
Two frequencies are provided for the purpose of compensating for atmospheric fading, distortion,
and other effects on any one frequency; effects that degrade one frequency are not expected to
affect the other channel. Two-frequency capability is not provided for the purpose of using one
frequency to compensate for interference effects on the other frequency.

Table 5. Technical Characteristics of Long Range Radar 2

Parameter Value
Tuning range 1215-1400 MHz
Channelization 44 frequency pairs
Uncompressed pulse widths 88.8 µs and 58.8 µs
Pulse repetition rates 291.5 or 312.5 pps
Pulse modulation Non-linear FM
3 dB RF bandwidth 58 MHz
3 dB IF bandwidth 690 kHz
Noise figure 3.6 dB
Receiver noise power (in 690 kHz bandwidth) -112 dBm
Antenna type Parabolic reflector with phased array feed array
Polarizations available Vertical, horizontal, RHCP, LHCP
Horizontal scan rate (scan interval) 5 rpm (12 sec/revolution)
Mainbeam gain 41.8 dBi
Beam number one 3-dB beamwidth Vertical 2.0 degrees, horizontal 1.4 degrees

3.7.2 Receiver Processing

Low-noise amplifiers (LNAs) for the RF front-end are located on the flat, phased-array feed
array of the antenna. LNAs are installed for row assemblies in the array. LNA outputs are fed
from the phased array to the receiver via the antenna rotary joint.

Two receiver processing channels, designated A and B and corresponding to each of the two
operational frequencies F1 and F2, are provided. Each channel has a receiver processor and a
modular DTE. A built-in tracker can process 800 aircraft plus 200 non-aircraft reports per scan.
The radar uses frequency separation and orthogonal polarization for the two channels so that they
can transmit and receive via the same antenna in the diplex mode of operation. Figure C-3 shows
the radar IF inherent noise spectrum. The receiver’s adaptive threshold algorithm establishes a
CFAR level by taking the average signal level over several range bins.

50
Processed target video data in each radar channel are digitally summed with the adjacent channel
target video data before distribution to the DTE. Cross-channel target video summation enhances
the probability of target detection since a fluctuating target fade in one channel should not be
experienced in the opposite channel due to frequency-polarization diversity between the two
channels.

3.7.3 Antenna Characteristics

The antenna is normally located inside a dome and mechanically rotates. The antenna design
provides some three-dimensional search capabilities using nine vertically stacked beams that
operate up to 30 degrees above the horizon. (The total range of elevation coverage is -7 to
+30 degrees.) To obtain accurate data for targets at long range (which will occur at low elevation
angles) the low beams are narrower in the vertical dimension than the higher-elevation beams.
An interleaving arrangement enables the upper stack of vertical beams to receive nearby target
echoes during the inter-pulse intervals of more widely spaced pulses from the lower-elevation
beams.

The phased array antenna feed has a flat configuration, is located at the base level of the
reflector, and contains approximately 600 radiators that can operate with either linear or circular
polarization. Linear polarization is normally used, but circular polarization provides better
penetration in rain conditions. As azimuth scans advance, the radar automatically selects the
optimum polarization for each of 32 azimuth sectors, depending upon whether each sector
contains rain. A significant feature of the radar is side lobes with low gain, which help to reduce
clutter and jamming effects.

3.8 Test Approach for Long Range Radar 2

3.8.1 General

The radar’s performance was monitored primarily using the built-in target generation and target
counting software. Desired signal targets were generated using a built-in, diagnostic RF signal
generator. In addition to the desired targets, signal generators were used to inject interference into
the radar receive path. The interference was coupled into the radar’s RF front end Beam 2
coupler via a fitting that was below the rotary joint. The interference energy and the target energy
were both fed through the rotary joint to a built-in coupler on the phased array feed, so that both
the desired targets and the interference entered the radar system ahead of one of the antenna
assembly RF front-end LNAs. Radar Pd performance was evaluated as a function of the
calibrated I/N level at the IF output of the radar receiver.

3.8.2 Injected Targets

Because the built-in target generator was used to create desired targets for Long Range Radar 2,
the test team performed extensive observations of the desired targets to thoroughly understand
them and demonstrate their adequacy for the tests. The radar target generator produced simulated

51
echo pulses that matched the time characteristics of the radar’s stagger22 pattern and, to some
extent, the antenna radiation pattern. Figure 24 shows a time-domain scan in the radar receiver IF
stage of the pulses comprising a target. The varying amplitudes of successive pulses are a
replication of the radar antenna beam shape, and the pulse intervals have the stagger sequence of
the radar.

Figure 24. Sequence of desired-target pulses in the IF stage of Long Range Radar 2.

In order to speed testing, a total of 202 injected test targets (distributed in sector wedges on the
PPI display) were monitored for 17 consecutive antenna rotations for a total of 3434 injected,
desired targets examined for each data point. Each target, regardless of displayed range, was set
to the same power at the receiver input, making each target equivalent in terms of receiver Pd.
The target power was adjusted to a level that provided an average target Pd in the absence of
interference of about 90 percent.

The target Pd, based upon 3434 attempted target scans, was based on automated counting by
radar validation software. Whenever the radar integrated enough target pulses to validate a target,
it was counted. As was the case for Long Range Radar 1, the actual Pd per pulse in the group of
pulses that defined each target would have been lower than 90 percent.

22
Stagger is a non-constant pulse repetition interval. It eliminates Doppler-processing ‘blind-speeds’ that occur at
the pulse repetition frequency of fixed-pulse interval radars.

52
 3.9 Undesired Signals in Long Range Radar 2

3.9.1 General

RF signal generators and AWGs were used to simulate the following interference signals in Long
Range Radar 2:

1) BPSK at 0.511 MBit/s,

2) BPSK at 1.023 MBit/s,
3) BPSK at 5.11 MBit/s,
4) BPSK at 10.23 MBit/s,
5) Carrier wave (CW),
6) QPSK at 2 MBit/sec,
7) Ultrawideband (UWB) pulses at 10 MHz prf (dithered pulse sequence),
8) UWB pulses at 10 MHz prf (undithered pulse sequence),
9) BPSK 10 MHz, radiated from a remotely located transmitter.

Spectra of the injected signals are very similar to those shown in Appendix B; with the exception
of the CW interference, the interference spectra are lobed (sinc)2 forms. BPSK and QPSK signals
contain data bits that are encoded into symbols that a spreading pseudorandom code further
breaks into chips. However, the radar receiver does not discriminate between phase changes
representing either a chip or a bit for these types of interfering waveforms. The radar receiver
will process the BPSK, QPSK, and UWB signals as band-limited constant amplitude noise
sources, which fall into all of the range cells. Since the noise-like BPSK and QPSK signals fall
into all of the range bins, the CFAR processing cannot eliminate them and the CFAR raises the
target detection threshold.

UWB signals consisted of pulsed emissions of 107 pulses per second at a pulse width of 0.511 ns.
The BPSK signal was radiated from the van through an external antenna mounted on a mast
towards the radar.

The emission spectra for the interference signals were measured and recorded, and tuned to the
frequency of the radar channel(s) under test. (The tuned frequencies for Long Range Radar 2
were 1228 MHz and 1311 MHz.) The loss between the interference signal injection point and the
LNA input on the radar feed array was about 74 dB (depending upon the exact path that was
used), so that a signal level of about -38 dBm from the interference generator produced about
0 dB I/N in the radar receiver. The interference signal simulator had the capability to generate I/N
ratios of up to +30 dB at the receiver RF input, but testing was usually accomplished only over
the subset of that range where meaningful Pd data could be collected. For each signal type,
calibrations were performed to allow for conversion between signal generator settings and the
resultant I/N level using the process described in Appendix C.

3.9.2 Duration of Interference Signals with Injected Targets

Because the injected targets were produced in sector wedges, the interference signals were kept
on continuously for about three minutes at a time, which was the duration of each interference

53
test run at each level for each modulation. To prevent the radar AGC from responding, its value
was frozen (at about 80 (arbitrary units)) during the runs. Test results were not affected by the
STC state.

Some radiated interference tests were performed with 10 MBit/sec BPSK signals. In these tests,
the duration of the interference was only as long as the interval for the radar beam to sweep
across the location of a remotely located vehicle that was the interference source, on the order of
0.4 sec per radar scan (Eq. 16). Because the interference was intermittent, AGC was allowed to
float for those tests.

3.10 Test Procedures on Long Range Radar 2

3.10.1 Hardline Coupled Interference Testing

The desired signal targets were overlaid with any one of the interference signals at a given I/N
and observed for 17 complete scans. This resulted in a total of 3434 possible targets (17 scans x
202 targets/scan), and probability of detection was calculated as the number of observed targets
divided by 3434. (During UWB interference testing the nominal number of 3434 targets was
increased to 4579.)

The interference test runs normally began with a given modulation at a low level (an initial I/N
level of -18 dB or -12 dB). The level was then gradually increased through each run, with 3434
(or 4579 for UWB interference) targets tested at each intermediate I/N level in increments of
about 3 dB, up to I/N values that sometimes reached +21 dB, but that usually stopped at I/N=+5
or +15 dB. The same interference modulation was always maintained throughout an entire
amplitude run. Interference runs were normally performed with:

• interference and targets on one channel and the other channel disabled;
• interference and targets on both channels;
• baseline “no interference” measurements were performed before and after each
“interference” data set.

As noted above, AGC was frozen at a fixed level during these runs to prevent adverse effects due
to the continuous presence of the interference. This expedient would replicate the behavior of a
radar if interference were momentarily encountered in its beam pattern (for about 0.04 sec, as
shown in Eq. 16), because in that circumstance the radar would be exposed to interference for
such a short period that the AGC would not respond.

The condition of STC (on versus off) was tested to determine whether it would affect the test
results. It did not matter, as it turned out, whether STC was activated or not. This was
presumably because the targets were displayed at distances that exceeded the STC turn-on
interval (which confines the effect to areas that are very close to the radar site).

54
 3.11 Results of Tests on Long Range Radar 2

3.11.1 Single-Channel Interference Test Results

A plot of the target Pd versus the I/N ratio for Long Range Radar 2 operating in single channel
mode with injected targets and interference is shown in Figure 25, with a smoothed-curve fit
between the data points. The figure shows that as the I/N ratio increases, the target Pd decreases.
At -6 dB I/N the target Pd has dropped several percentage points below the baseline value of
0.95.

3.11.2 Dual-Channel Interference Test Results

A plot of the target Pd versus the I/N ratio for the radar operating in dual channel mode with
injected targets and interference on both channels is shown in Figure 26. The figure shows that as
the I/N ratio increases, the target Pd drops. The figure shows that although the baseline target Pd
per channel was set to 0.95, the overall baseline target Pd is reinforced (improved) with dual
channel operation to about 0.98. But dual channel operation only helps the radar perform better
to a point; at an I/N level of -6 dB, the target Pd has dropped several percent.

Figure 25. Long Range Radar 2 single channel operation variation in Pd with interference.

55
Figure 26. Long Range Radar 2 dual channel operation variation in Pd with interference.

3.12 Summary of Interference Effects on Long Range Radars

From the data obtained, the following general conclusions can be drawn regarding the
performance of Long Range Radars in the presence of low-level interference signals:

• The Pd of simulated radar targets (of the minimal acceptable cross section) for single
channel operation showed degradation at interference levels as low as I/N= -9 dB.

• The Pd of simulated radar targets (of the minimal acceptable cross section) for single
channel operation was always measurably degraded, by as much as 0.15 (with Pd
dropping to 0.75 from a nominal 0.90), at I/N= -6 dB.

• The Pd of simulated radar targets (of the minimal acceptable cross section) for dual
channel operation with interference on both channels was degraded by as much as 0.10
(from 0.90 to 0.80) at I/N= -6 dB.

• The beginning of degradation appears to be independent of the interference modulation

type. This may be because all modulations that were tested were wider than the radar IF
bandwidths. However, this is a realistic condition, in that it is expected that the bandwidth
of most interference signals will exceed the bandwidths of most radar receivers.

56
 • Radar dual frequency operation with interference on both channels did not significantly
improve performance in the presence of low-level interference, but did seem to keep the
Pd value at 0.8 at I/N levels of +5 dB and above.

• Live sky tests showed that with an I/N of +20 dB and a 10 MBit/s BPSK interfering
signal, the radar lost track of some targets near the edge of its coverage range.

• While the effects of strong interference in radar receivers can be very dramatic, as shown
in Figure 27, the damaging effects of low-level interference in radar receivers are
insidious because there is no overt indication that interference is occurring, as shown in
Figure 28. For low levels of interference (typically less than I/N=+3 dB), target losses
were not accompanied by any other indications that interference was occurring. Were
such losses to occur during normal radar operations, the operators would probably not be
aware that targets that should otherwise have been detected were missing, nor that any
interference was occurring. It is suggested that one reason that low-level interference
effects are rarely reported by radar operators is the subtle, insidious nature of such
interference in radar receivers.

• Loss of desired targets in the presence of interference is purely a function of the target
echo level relative to the radar receiver noise limit; target range from the radar is not a
factor, as illustrated, for example, in Figure 28. It is true that, all other factors being equal,
targets at long ranges will return weaker echoes than closer targets. But all other factors
are never equal; target cross sections vary. Therefore, any target at any distance from a
radar can be lost due to interference when its return echoes are weak. This means that
low-level interference will not just result in loss of targets at the edge of radar coverage.
Weak targets that are close to radar sites will be lost as well. Weak targets, at any range,
are just as critical in national defense and safety considerations as larger cross section
targets at the edge of radar coverage; small cross section targets at short ranges may be
more critical in some situations.

Figure 27. The effects of strong interference in two radar receivers, such as these examples
where I/N is on the order of +20 dB, will normally be obvious to operators. (When such effects
occur on single azimuths they are visible as bright lines called strobes.)

57
Figure 28. In contrast to the effects of strong interference in Figure 27, target losses due to low-
level interference effects (occurring here at an I/N level of about -3 dB in Long Range Radar 1)
are insidious; there is no indication on the PPI display that interference is occurring. The
insidiousness of the phenomenon of low-level interference in radar receivers is of considerable
concern because of its implications for adverse impacts on radar missions without even the
possibility of warning indications for operators.

58
4 INTERFERENCE MEASUREMENTS ON A SHORT-RANGE AIR SEARCH RADAR

4.1 Introduction

Interference measurements and tests were performed to determine the effects of emissions from
digital communication signals on a short-range air search (radionavigation) radar operating
within the primary allocation for airport surveillance radars (ASRs) in the 2700-2900 MHz band.
Such radars are operated at shorter ranges than the radars examined in the previous section,
typically on the order of 100 nmi.

The radar that was tested employs pulse widths, prfs, IF bandwidth, noise figure, and antenna
beamwidths typical of those identified in an ITU-R Recommendation for radars in this band [20].
Those radars typically employ interference mitigation techniques/processing methods identified
in another ITU-R Recommendation [18] to allow them to operate in the presence of other
radionavigation and radiolocation radars. As noted in Section 1, those techniques are effective in
reducing pulsed interference between radars but not high duty cycle interference from
communication signals.

The tests described here investigated the effectiveness of the radar’s interference suppression
circuitry and software to reduce or eliminate interference due to the emissions from
communication signals employing a digital modulation scheme.

4.1.1 Objectives

The objectives of the testing were to:

• quantify the capability of a short-range air search (radionavigation) radar’s interference-

rejection processing to mitigate unwanted emissions from digital communication systems
as a function of their power level;

• develop I/N protection criteria for unwanted digital communication systems emissions
received by such a radar;

• observe and quantify the effectiveness of the short-range air search radar’s interference
rejection techniques to reduce the number of false targets, bright radial lines on the PPI
display (strobes), and background noise.

4.2 Short Range Air Search Radar Technical Characteristics

4.2.1 General Characteristics

The short-range air search radar that was tested is used for monitoring air traffic near airports,
and is also referred to as a terminal radar. Nominal values for the principal parameters of this
radar are presented in Table 6.

59
Table 6. Technical Characteristics of the Short-Range Air Surveillance Radar

Parameter Value
Frequency 2700-2900 MHz
Maximum range 60 nmi
Pulse width 1.08 µs
Pulse repetition frequency 928 and 1193 Hz; 1027 and 1321 Hz
IF bandwidth 653 kHz
Spurious response rejection 60 dB
System noise figure 4 dB
RF bandwidth 10 MHz
Antenna scan rate (scan time) 12.5 rpm (4.8 sec/revolution)
Antenna horizontal beamwidth 1.4 degrees
Antenna vertical beamwidth 4.8 degrees
Polarization Right-hand circular or linear

The radar divides its 60-nmi operational range into 1/16-nmi intervals and azimuths, and each of
those by 256 intervals of approximately 1.4 degrees each, for a total of 245,760 range-azimuth
cells. In each 1.4-degree azimuth interval the transmitter sends ten pulses at a single prf and then
sends eight pulses at a lower prf. The receiver processes each set of 18 pulses to form 18 Doppler
filters. Alternating prfs within every 1.4-degree wedge helps to eliminate blind speeds, unmasks
moving targets hidden by weather, eliminates second-time-around clutter returns, and divides the
radar output into approximately 4,483,584 range-azimuth-Doppler cells.

4.2.2 Antenna Characteristics

The radar employs a parabolic reflector with two horns, for high and low beams, in the antenna
feed array. Target echo pulses received by the high and low beam horns in the antenna array are
amplified by RF front-end low noise amplifiers and are sent via toggle-switched paths to their
respective receivers. The high-beam horn receives returns from high elevation-angle targets close
to the antenna, while the low-beam horn receives returns from low elevation-angle targets at
greater distances. The high beam receiver reduces clutter strength at short ranges in order to
improve sub-clutter visibility.

For these tests, the low beam receiver was selected because the radar would be most likely to
receive interference from local ground-based emitters through this path. Τhe low beam is used
for observation of targets at ranges exceeding about 15-20 nmi. The beams are not used
simultaneously; the radar receiver toggles sequentially between them. The coverage patterns for
the high and low beams for a 1 m2 target cross section with Pd=0.80 are shown in Figure 29.

60
Figure 29. Beam coverage patterns for the short-range air surveillance radar.

4.2.3 Target Receiver

The target receiver/processor employs STC and Moving Target Detection (MTD), which
includes Doppler filtering and CFAR processing, to detect and separate target returns from noise,
ground clutter, and weather. The target receiver/processor sorts the target returns according to
range, detects their Doppler shifts, and sends them to the radar system post processor.

4.2.4 STC

The radar uses STC in its receiver to provide attenuation immediately after the start of the
transmitted pulse to decrease its sensitivity at close range, as described in Section 1 of this report.
The radar STC curve is shown in Figure 5. The STC was disabled during the tests, but the target
amplitudes were set at values that would not have been affected by STC even if it had been
active.

4.2.5 IF Circuitry

The IF receiver amplifies the outputs of the RF receiver and detects their phase shifts. The IF
circuitry consists of a three-stage logarithmic video detector/amplifier with a wide dynamic range

61
and I and Q phase detectors. The output of the IF amplifier receiver is at 31.07 MHz. A CW
signal swept at RF frequencies was applied as an input stimulus to the radar receiver to obtain the
receiver’s 3-dB bandwidth, which was measured to be about 680 kHz at the input to the phase
detectors. The receiver’s response to the swept CW signal is shown in Figure C-4. The dynamic
range of the radar receiver was measured by varying the power level of a fixed frequency CW
signal and monitoring the output of the IF circuitry at the same test point. Figure 30 shows the
gain characteristics of the radar receiver. Measurable compression occurs with an input signal
power level of about -43 dBm.

Figure 30. Short-range air surveillance radar input/output gain curve.

The phase detectors at the output of the IF amplifier determine the change in phase between the
returns and the transmit pulses that produced them, using the coherent oscillator (COHO) from
the frequency generator as a transmit-pulse phase reference. The phase detectors each have
sinusoidal responses, and produce in-phase (I) and quadrature (Q) outputs with a sine-cosine (90
degree) phase relationship to each other. Because the I and Q phase detector responses are sine
and cosine functions, their outputs can be vector-summed to determine the actual magnitude of
the target returns. Software-implemented servo loops set the DC offsets, gain balance, and phase
balance of the I and Q outputs based on the state of the phase detectors. They also set the AGC
level of the RF and IF amplifiers to limit the noise level within one quantum (the change in RF
level represented by the least significant bit (LSB) output of the analog-to-digital (A/D)
converter) of the radar internal noise level.

The I and Q outputs of the IF circuitry are sampled and digitized by A/D converters during each
0.77 µs (equal to 0.75 percent of the transmitted pulse width), covering a 1/16-nmi range cell, at
a clock rate of 2.6-MHz. The results are then interleaved. The A/D converter produces 12-bit
digital words that represent the samples of the I and Q signals to the filter and magnitude
processor.

62
4.2.6 Doppler Filtering

In each 1/16-nmi range cell, coherent processing intervals (CPIs), consisting of returns from
alternately 10 and 8 successive pulse repetition intervals, are formed. In the 10-pulse case, the
batches associated with each successive 1/16-nmi range increment are sequentially applied to the
same set of 10 Doppler filters. The radar random access memory (RAM) stores digital
representations of the returns over several pulse-repetition trains (PRTs) and the Doppler filters
process them together so that pulse-to-pulse changes in target-return amplitudes (representing
apparent Doppler frequencies) can be calculated. For the 10-pulse CPI, five of the filters are used
to detect targets moving towards the radar antenna and the other five are used to detect receding
targets. A similar process is used for the 8-pulse CPI, except that eight filters are used. The
Doppler filters improve the receiver’s signal-to-noise ratio because the Doppler filters add or
integrate a series of target returns at their frequency. This causes return signals to progressively
accumulate at the output of the filter, while random-frequency noise accumulates at the filter
outputs at a much slower rate.

4.2.7 CFAR Processing

The radar uses a 27-cell sliding-window averaging (or range averaging) CFAR technique to
calculate the mean level threshold (MLT). (CFAR processing automatically varies a detection
threshold to maintain false target declarations, based on the return signal plus noise outputs of the
Doppler filters at a constant rate, as described in Section 1 of this report.) Each Doppler filter
sums the energy contained in the stream of returns received as the antenna sweeps over a target.
The energy combines with the noise energy that accumulates in the filter during the same time
interval. If the integrated signal plus noise at the output of a filter exceeds the MLT, the detector
concludes that a target is present.

Thresholds for the non-zero velocity resolution cells are established by summing the detected
outputs of the signals in the same velocity filter in a 27-cell window centered about the cell of
interest. Thus, each filter output is averaged to establish the mean level of non-zero velocity
clutter. Filter thresholds are determined by multiplying the mean levels by an appropriate
constant to obtain the desired false-alarm probability.

Random noise will occasionally exceed the MLT and the detector will falsely indicate that a
target is present. The higher the detection threshold to the mean level of the noise energy the
lower the probability of a false alarm will be, and vice versa. If the detection threshold is too
high, valid targets may go undetected. The outputs of the Doppler filters are continuously
monitored to maintain an optimum threshold setting. The CFAR sets the detection thresholds to
maintain the false alarm rate for each Doppler filter at an optimum value. A QPSK-type
waveform covering the band of the radar receiver will appear simultaneously in all the Doppler
filters as noise and cause the CFAR to raise the detection threshold, causing all targets to have a
correspondingly lower probability of detection.

63
 4.3 Interference Procedures and Methods for the Short Range Air Search Radar

4.3.1 Interference Signals

Three types of interference signals were injected into the radar as unwanted emissions through a
20 dB coupler port in the receiver’s waveguide path, as shown in Figure 31. The signals were an
un-modulated continuous wave (CW), a continuous 2-MBit/s QPSK waveform, and a 2-MBit/s
QPSK waveform with a 1/8 time slot duty factor. All three signals were co-channel with the
radar’s operating frequency and occurred within the full 360 degrees of the antenna’s rotation.
The continuous and pulsed QPSK waveforms represent the type of signals that would be
expected to be used by digital communication systems such as IMT-2000 in CDMA and TDMA

Figure 31. Test set-up for interference injection into short-range air search radar.

64
systems, respectively. The measured emission spectrum of the continuous QPSK signal is shown
in Figure B-5. The QPSK signal was generated and injected into the short-range air search radar
using the test set-up shown in Figure 31.

The CW signal was simulated using an RF signal generator. For the CDMA type QPSK
waveform, an AWG was programmed to output a QPSK waveform at a data rate of 2 MBit/s.
The program simulated a uniform distribution of random bits containing no repetitions with a
code length long enough to ensure that no spectral lines occurred within the bandwidth of the
radar receiver. For the TDMA (time-sliced) type QPSK waveform, another AWG was used to
pulse the QPSK waveform for a 1/8 time slot duty factor. The on-time of the pulses was 577 µs
and the repetition period was 4.6 ms.

The output of the AWG was fed to a mixer whose other input was connected to an RF signal
generator. The RF signal generator functioned as a local oscillator and its frequency was adjusted
so that the carrier of the QPSK waveform was co-tuned with the radar receiver. A YIG bandpass
filter was used to suppress any spurious emissions that resulted from the mixing process. An RF
step attenuator immediately after the YIG filter was used to control the power level of the QPSK
emissions.

4.3.2 Target Generation and Counting

Ten simulated, equally-spaced targets were generated along a radial using the radar’s built in test
target generator (consisting of a combination of hardware and software). The targets on the radial
had a constant power envelope. Each target count Pd was made with 20 rotations of the radar; in
20 rotations, 200 targets were generated. The value of Pd was determined by dividing the number
of counted targets by 200. For example, if 180 targets were counted the Pd was 0.90. The targets
were counted manually, by observing the correlated video output on the radar’s PPI display.

4.3.3 Test conditions

The tests were performed with the parameters shown in Table 7 set on the short-range air search
radionavigation radar.

Table 7. Control Settings for the Short Range Air Surveillance Radar

Receiver Parameter Value

Sensitivity time control (STC) Off
Interference rejection (IR) On
Automatic gain control (AGC) On
PPI image selected for target counting Processed video
Radar range mode 60 nmi

65
4.3.4 Test Procedures

The RF power output of the target generator system was adjusted so that the target Pd was as
close as possible to 0.90 (given that the target levels could only be adjusted in 1-dB increments)
without interference being present (for correlated video targets). As noted above, targets were
counted in twenty scans to set the baseline Pd. The radar was observed to have a sort of memory
of target conditions, as the target Pd was slow to change when the target power level was adjusted
using the target control software. Due to the CFAR processing, the radar took 8-10 scans before
it would reach a new steady state after the target power was adjusted.

After the radar was set to its baseline condition, the CW and QPSK interference was injected into
the radar receiver. The power of the CW and QPSK signals that were injected into the radar
receiver was set to different levels while the power level of the targets was held constant.

The power levels of the CW and QPSK signals were set to values that produced I/N levels of
-12 dB, -10 dB, -9 dB, -6 dB, -3 dB, 0 dB, +3 dB, and +6 dB in the IF circuitry of the radar
receiver, as described in Appendix C. To account for the radar’s target memory, the targets were
not counted until 10 scans had occurred after the interference had been enabled. After 20 scans
with the interference enabled and the targets counted, the interference was disabled and an
additional 10 scans were allowed to occur before the next I/N level was tested. Waiting 10 scans
to occur between each change in target and interference conditions ensured that each
measurement was not affected by the previous one.

As the CW and QPSK power levels were varied, the display of the radar was observed for any
increase in the number of false targets, bright radial streaks or lines (strobes), or any increase in
background speckling.

4.4 Test Results for the Short Range Air Search Radar

Curves showing the target Pd versus the I/N levels were produced for the CW, CDMA-QPSK,
and TDMA-QPSK interference are shown in Figure 32. The data in Figure 32 show that all three
emissions caused a reduction in the target Pd ranging from 9 to 25 percent for I/N levels equal to
-6 dB. The TDMA type emission did not cause as much of a reduction in the target Pd at the
higher I/N levels as the other waveforms. This is believed to be because the radar’s CFAR was
able to treat it as pulsed interference and reduce its effects to some degree. However, the radar’s
operation was still adversely impacted. Furthermore, higher time-slot duty factors would increase
the effects from the TDMA waveform, until at some point they would equal the effects from the
CDMA type signal.

4.5 Summary of Interference Effects on a Short-Range Air Search Radar

The Pd interference curves show that the short range air search radar’s target detection capability
was adversely impacted at an I/N level of -6 dB. This result is consistent with the observations of
the behavior of the long range air search radars.

66
Figure 32. Interference response curves for the short-range air search radar. CDMA-2000
mobile terminal modulation was used for these tests.

67
 5 INTERFERENCE MEASUREMENTS ON A FIXED GROUND-BASED
METEOROLOGICAL RADAR

5.1 Introduction

The key objective of the work described in this section was to establish the maximum
interference level that a representative meteorological radar, deployed on a worldwide basis,
could withstand before its forecasting capability was compromised; the interference being based
on typical communication signal modulations. Based upon the radar’s technical specifications,
mathematical models were developed for its key outputs, or ‘products’ (called base reflectivity,
mean radial velocity, and spectrum width), that indicate what these expected levels should be. In
order to physically validate this analysis, a test and data analysis methodology was defined
through which data were collected and analyzed.

The data analysis supported the calculated value required for protection of the reflectivity
measurements. Limitations in the radar calibration and noise removal process performed by the
system’s low-level data processor at the time the tests were run tended to limit the measurement
accuracy of the necessary protection criteria for the spectrum width measurements. However,
correction of the data for the limitations of this processing did result in values that supported the
calculated protection values.

Field tests were run on the meteorological radar to determine the appropriate criteria necessary
for protection from CW and CDMA signals in the 2700-2900 MHz band. The tests were
performed by injecting a CW signal and six different CDMA modulation schemes into the radar
receiver while it was scanning the atmosphere. Low-level (called base) meteorological products
(base reflectivity, mean radial velocity, and spectrum width) were recorded while conducting a
series of radar antenna rotations at a single antenna elevation. Interference signals were injected
with I/N levels ranging from -15 dB to +6 dB.

5.2 Theoretical Calculation of Necessary Protection Criteria

The radar generated three base products that are used by the signal processing system to derive
meteorological products. These base products are:
• Volume reflectivity, Z, in mm6/m3, which for rain is a measure of total water in the radar
sample volume;
• Mean radial velocity, V, in m/s, which is the power-weighted mean radial motion of the
targets in the sample volume;
• Spectrum width, W, in m/s, which is a measure of the radial velocity dispersion of the
targets in the sample volume.

5.2.1 Minimum Signal Level

Radar signal processing normally removes the radar system noise effects from the reflectivity
and spectrum width products so that the system can provide these products when the signal level

68
is below the receiver noise level. The signal to noise ratio (SNR) threshold, i.e., the lowest level
for which the return signal is processed, is selectable by the radar operator between the limits of -
12 dB SNR and +6 dB SNR. With the present version of signal processing, the lower values are
generally not used due to limitations with noise removal but the system does provide useful
products down to -3 dB SNR. The interference level that compromises the system is related to
the minimum signal level of -3 dB SNR and the product characteristics themselves, as described
below. Excessive interference will adversely impact data quality, degrade the meteorological
products, and compromise the system’s ability to accomplish its mission of providing data
necessary for public weather forecasting, severe weather warning, and rainfall measurement for
flash flood prediction and water management.

5.2.2 Reflectivity

Reflectivity is used in multiple applications; the most important of these is rainfall rate
estimation. Reflectivity is calculated from a linear average of return power and is subject to
contamination by interference as an unknown increase in the measured reflectivity. Reflectivity
is seriously contaminated if the bias exceeds 1 dB. A 1-dB bias is twice the radar calibration
accuracy and equal to the standard deviation of the reflectivity estimate specified in the radar
technical requirements. Bias in terms of interference to signal ratio is given by [21]:

⎛S + I⎞
dBbias = 10log⎜ ⎟, (17)
⎝ S ⎠

and a 1-dB bias occurs at:

I/S = 0.26
10 log I/S = -6 dB.

Therefore, reflectivity is biased 1 dB at an interference level 6 dB below the signal. Since the
minimum signal level has an SNR of -3 dB and the maximum I/S level for the reflectivity
product is -6 dB, the maximum I/N is:

(-3 dB) + (-6 dB) = -9 dB I/N.

5.2.3 Mean Radial Velocity

Mean radial velocity is calculated from the argument of the single lag complex covariance. The
complex covariance argument provides an estimate of the Doppler signal vector angular
displacement from pulse to pulse. The displacement divided by the time interval between the
pulses is the Doppler vector angular velocity.

If it has the characteristics of broadband noise, an interference signal vector has uniform
probability over the complex plane and thus does not introduce a systematic rotation nor a bias in

69
the estimate. However, the randomness of the composite signals plus interference vector due to
the interference increases the variance of the Doppler signal estimate.

The Doppler frequency variance, retaining all terms except those inversely proportional to the
number of samples squared can be calculated as [21]:

⎛ N ⎞2 ⎛ N ⎞
⎜ ⎟ + 2⎜ ⎟[1− β (2T )]
2π 2WT
3

ˆ
()
var f = 2
8π Mβ (T )T
2 2
+
⎝S⎠ ⎝S⎠
8π Mβ 2 (T )T 2
2
, (18)

where:

fˆ = frequency estimate, Hz;

W = standard deviation of frequency spectrum, Hz;
W = 4 m/s for NTR benchmark = 80 Hz at fc = 2995 MHz;
T = sampling interval, sec;
T = 10-3 sec for NTR benchmark;
M = number of samples in estimate;
N = noise power;
S = signal power;
β = signal correlation at lag T;
β = exp [-2π 2 W2 T2 ] for the assumed Gaussian spectra.

The first term is the variance contribution due to the signal characteristics and the second term is
the variance contribution due to the noise.

The frequency variances are severely compromised if the interference increases the variance by
more than 50%. The uncertainty in the data degrades all velocity based products and the velocity
shear measurements in particular. (Velocity shear is a velocity difference over some distance.) A
50% increase in variance increases the reliably detected shear value approximately 25% above
the severe weather event formative stage value.

An expression for the interference to noise ratio resulting in a 50% variance increase of the
technical requirements benchmark parameters and SNR = -3 dB is given by [21]:

⎛ I + N ⎞2 ⎛ I + N ⎞ 3
2
3⎛ N ⎞ 3 ⎛ N ⎞
2π WT + ⎜
3
2

⎝ S ⎠
⎟ + 2⎜
⎝ S ⎠ 2
3
(
2
)
⎟[1− β (2T )] = 2π WT + ⎜ ⎟ + (2)⎜ ⎟[1− β (2T )],
2⎝ S ⎠ 2 ⎝ S ⎠
(19)

where:
W = 80 Hz;
T = 10 –3 sec;
2π 3/2WT = 0.89;
[1 - β ( 2T)] = 0.4;
S = 0.5 N.

Substituting and solving for I/N yields the quadratic expression:

70
 (I/N)2 + 2(I/N) - 1.21 = 0
I/N = 0.49 .

Therefore, the interference must not exceed the minimum signal value.

5.2.4 Spectrum Width

The spectrum width is calculated from the single lag correlation assuming a Gaussian spectral
density. The algorithm is expressed as [21]:
1

R2
2
Va
W = ln 2 , (20)
π S

where:

W = Spectrum width (standard deviation);

Va = Nyquist velocity, 25m/s from the radar technical requirements document;
R = single lag covariance power;
S = signal power.

Interference causes both a bias and a variance increase in spectrum width estimation but the bias
is more detrimental. Spectrum width is compromised when the interference-induced bias exceeds
the radar technical requirement width accuracy of 1 m/s. The interference to noise ratio at which
this bias level occurs can be calculated by solving for the covariance at 4 m/s and signal power of
N/2, then solving for the (S+I) level producing a 5 m/s spectrum width. 4 m/s is the base value
given in the radar technical requirements document. To calculate I/N, the equation above is
solved for the 4 m/s and 5 m/s cases.
For W = 4 m/s: For W = 5 m/s:
25 / π ⎢ln (R2/S2)⏐(1/2) = 4 25 / π ⎢ln (R2 / (S + I)2)⏐(1/2) = 5
ln (R2/S2) = -0.25 ln ( R2 / (S + I)2 ) = -0.39
R/S = 0.88 R/(S+I) = 0.82
R = 0.88 S
R = 0.88(N/2)
Substitute: R = 0.88(N/2), S = N/2:
[0.88 (N/2)] / [(N/2) + I] = 0.82
0.82 [(N/2) + I] = 0.88 (N/2)
I/N = 0.0366
10log(I/N) = -14.4 dB.

So in this example, an I/N ratio of -14.4 dB will cause a spectrum width error of 1 m/s.

71
 5.3 System Operation, Output Products, and Interference Sensitivity

5.3.1 System Operational Mode for Testing

The radar that was used for the tests and measurements has multiple modes of operation that
utilize different antenna rotation rates, antenna elevations and prfs. The operational mode
selected for the tests was one that is commonly used, and is optimized for system sensitivity,
leading to high susceptibility of interference. Table 8 provides the characteristics of the radar
mode used in testing.

Table 8. Technical Characteristics of the Meteorological Radar

Parameter Value
Frequency 2995 MHz
Pulse power (peak) 750 kW
Pulse width 4.7 µs
322 Hz (first cut)
Pulse repetition frequency
446 Hz (second cut)
Maximum coverage range 290 mi.
RF bandwidth (at 3-dB points) 13 MHz
IF bandwidth (at 3-dB points) 630 kHz
System noise figure 4.9 dB
System noise floor (in receiver bandwidth of 630 kHz) -111 dBm
Main beam antenna gain [45 dBi]
Antenna pattern type pencil
Antenna scan interval (scan period) 0.84 rpm (71.4 sec)
Antenna height above ground 30 m
Antenna beamwidth (vertical and horizontal) 0.90 deg
Polarization linear horizontal

In the mode used, the antenna rotation starts at an elevation of 0.5 deg., the radar transmits a
4.7-µs pulse every 3.1 ms for the first rotation, and then transmits a pulse every 2.24 ms for the
second rotation. These correspond to prfs of 322 Hz and 446 Hz, respectively. Each revolution
covers 360 deg. in azimuth. In normal operation, the radar also performs antenna rotations at
several higher elevation angles before returning to 0.5 deg. For the purposes of this test, the two
elevation cuts at the single antenna elevation provided sufficient data for analysis and the cuts at
higher elevations were not performed. The first antenna rotation is used to measure reflectivity
and the second rotation is used to measure mean radial velocity and spectrum width (see below).
For each location in the atmosphere, multiple pulses are transmitted and received. Due to the
duration of the transmit pulses compared to the time between pulses, the system is in receive
mode more than 99.5% of the time. The magnitude of the received pulses is approximately

72
200 dB lower than the transmitted pulses because the pulses are scattered by small airborne
objects (on the order of millimeters in diameter or smaller) at distances up to hundreds of
kilometers from the radar. The received signal is downconverted from 2,995 MHz to the IF
frequency of 57 MHz where it is then applied to a synchronous detector. The detected I-Q
baseband signals are digitized to a 16-bit level for use in the processing subsystems.

5.3.2 Output Products

The returned pulses from each location are used by the processing subsystems to derive the three
meteorological base moments of base reflectivity, mean radial velocity, and spectrum width. The
base moments are displayed as products to users and are used to develop other meteorological
products representing rainfall accumulation, tornadoes, wind shear, etc. Reflectivity is derived
from the amplitude (or power) of the received signal. Mean radial velocity is the mean radial
speed and is derived from the differences in the I and Q vectors caused by the Doppler shift.
Spectrum width is the variance between pulses of the velocities received from the same location.

5.3.3 Interference Sensitivity

Base products are affected by interference in two different ways. First, values can be biased
which decreases the accuracy of the system, and second, the variance of the outputs can be
affected. In the presence of interference, reflectivity is sensitive to bias, mean radial velocity is
sensitive to variance errors, and spectrum width is affected by both bias and variance errors. For
spectrum width, the errors due to biasing are more significant than the errors due to variance
because the bias, or offset, represents a velocity measurement error while the variance represents
the uncertainty of the velocities measured. Table 9 shows which interference-induced errors, bias
and variance affect the base products.

Table 9. Sensitivity of Meteorological Products to Interference Induced Error

Interference Induced Errors

Meteorological Product Bias Variance

Reflectivity X
Velocity X
Spectrum Width X X

To calibrate the test set-up for a known interference level at the radar receiver input, the receiver
noise floor was measured, without interference, at the 57.55 MHz IF output of the receiver.
When the noise floor was recorded at the IF output, the interference signal was activated and its
level was increased until the radar IF output noise floor increased by 3 dB. The point at which the
noise increased by 3 dB corresponded to the interference level within the radar passband being
equal to the radar receiver noise within the passband, and an I/N ratio of 0 dB. The signal source
output was recorded for the 0-dB I/N ratio and the actual level being injected into the radar

73
receive path was also measured and recorded. Knowing the signal source setting for the 0 dB
I/N, the signal source could be set for any other desired I/N by adjusting the signal source output
level. Testing was conducted at interference level points corresponding to the following I/N
levels: -15 dB, -12 dB, -10 dB, -6 dB, -3 dB, 0 dB, +3 dB and +6 dB. Figure 12 shows an
example result of the receiver noise level measurement with and without the presence of
interference for the W-CDMA signal testing; Figures B-6 and B-7 show spectra of CDMA
interference signals used for this testing.

The radar was set to scan the atmosphere (hereafter called a “volume scan”) at one antenna
elevation without interference followed by a volume scan with interference. For each volume
scan, with or without interference, the antenna made two complete rotations allowing elevation
cuts at the same elevation using two different prfs. The prf used on the first rotation was a low
value optimized for collecting the base reflectivity product. The prf used on the second rotation
was a high value and was used for collecting the mean radial velocity and spectrum width data.
This alternating pattern of volume scans, with and without interference, was continued for
interference levels ranging from -15 dB to +6 dB. This test approach provided a volume scan
immediately before and after each interference volume scan that could be used as baseline
references for determining the statistical effects of the interference. During the entire test, base
product radar data was recorded for analysis.

Figure 33 shows the test setup, which consisted of a signal generator feeding an RF coupler
where the interference signal is combined with the received radar return signal at input to the
receiver. The receiver amplified and downconverted the signal to IF where it was monitored on a
spectrum analyzer. The I and Q outputs were digitized and processed to provide the
meteorological base products of base reflectivity, mean velocity and spectrum width. The base
products were recorded for statistical analysis. Testing with each of the CDMA signal types, at
all data rates and modulation schemes, was not feasible due to the large number of available
permutations. Representative modulations (CDMA and TDMA, for example) and a
representative range of data rates were used instead.

74
Figure 33. Meteorological radar test set-up block diagram.

5.4 Data Analysis Methodology and Results

Radar tests described in other sections of this report concern systems that locate and track
discrete, point targets. Meteorological radars, in contrast, collect a completely different type of
data, namely on extended, diffuse phenomena. Meteorological radars such those addressed in this
section perform volume scans of the atmosphere and present data on the atmosphere for a full
360 degrees in azimuth and through elevations as high as about 60 degrees.

For discrete-target radars, analysis of the effects of interference on the probability of detection
are usually sufficient, and interference effects in those types of systems will tend to mask desired
targets and/or created false targets.

75
In meteorological radars, where data are collected and analyzed for a volume of the atmosphere,
the system performance is not characterized with the use of probability of detection. Although
visual inspection of the radar display can show some effects due to interference, such inspection
does not provide a scientific analysis of the results on the meteorological products generated such
as rain fall estimates, wind speed measurement or shear detection. The data analysis for
meteorological radar outputs must take a much different approach in order to provide meaningful
results, and it has been determined that in fact an extensive statistical analysis must be performed
on the low level meteorological data for each range gate response that is received.

5.4.1 Assumptions

• As stated above, the test procedure used to inject interference signals into the radar
receiver called for injecting a known interference level at the radar receiver’s input.
• The minimum usable signal, with current technology, is 3 dB below the noise floor.
• The required maximum I/N ratio is equal to the interference level below the signal that
resulted in a 1-dB bias plus the minimum signal level that needed to be retrieved.

The system used processing to remove the effects of noise, allowing the radar to process signals
below the noise floor. In a system that would contain no residual effects, one would expect the
interference that was injected at the receiver input to linearly track the interference level that was
detected through the data analysis. Figure 34 compares the relative levels of the interference that
was injected at the receiver input to the interference level that was detected through the data
analysis. A divergence can be seen at -6 dB.

Figure 34. Injected versus detected interference level in the meteorological radar receiver.

76
This residual effect impacts the analysis in the following way:

• Reflectivity: No effect as the graphical technique that was used to determine the level at
which a 1-dB bias in the reflectivity occurs is relative and is not impacted by this residual
effect.
• Spectrum width: Residual noise (be it present because it is an artifact of the radar’s
calibration, inherent to the radar’s performance, or as a result of graphical analysis errors)
impacts the analysts’ ability to graphically determine the level at which the spectrum
width difference exceeds 1 m/s.

This residual effect can be characterized, and the data analysis can be compensated accordingly,
using the curve shown in Figure 35. These data were used to compensate for analysis errors that
were introduced by using an absolute graphical technique to determine the level at which the
spectrum width difference is 1 m/s. The data were not required for the reflectivity analysis.
Additional variability that adds to the data analysis errors comes into play as a function of
graphically estimating the mean and associated data points. An analysis follows, based on the
topics of reflectivity and spectrum width.

Figure 35. Detected minus injected interference levels as a function of injected level.

5.4.2 Reflectivity Analysis Methodology

Figure 36 represents the regression of the (interference+signal) level to the signal level for the
TDMA-GMSK interference signal. This regression analysis has been used to graphically derive
an interference level relative to the received signal level that results in a 1-dB bias in the
reflectivity measurement. The interference level that results in a 1-dB bias is set equal to the
difference between SNR (without interference) corresponding to the 3-dB and 1-dB SNR
difference points. Adding this number to the -3-dB SNR level that was derived above results in
the true I/N level. As an example, an average has been drawn on the data in Figure 36. The 1-dB
and 3-dB points are noted on the y-axis and horizontal lines have been drawn to intersect the
average. At the points of intersection, vertical lines have been drawn to intersect the x-axis (SNR
without interference). The difference between these points of intersection represents the
interference level below the signal level that results in a 1-dB bias.

77
Figure 36. Reflectivity regression for interference in the meteorological radar.

For this example analysis, the I/N calculation yields the following results:

I/N = (The interference level below the signal that results in a 1 dB bias) + (The minimum
signal level that needs to be retrieved).

Table 10 illustrates the results of this analysis for one test case. The results presented in this table
are consistent with the expected values as derived in Section 5.3.2.

5.4.3 Spectrum Width Analysis Methodology

A similar analysis approach was taken for determining the level at which a 1m/s bias occurred in
the spectrum width product. In this case the regression that was used for graphically determining
this value is shown in Figure 37.

78
Table 10. Reflectivity Results for Example Analysis
Interference Level Relative to Signal I/N Level at which 1-dB Bias Occurs (dB)
(dB)
+6 -9.5
+3 -9.5
0 -9.5
-3 -10
-6 -7.5
-10 -9
-12 -7.5
-15 -8

Mean -8.9
Standard Deviation 0.93

5.4.4 Spectrum Width Regression

The process for analysis of spectrum width regression is very similar to the method used to
determine the level at which a 1-dB bias occurred in the reflectivity data. An absolute level is
derived from the reflectivity plot at a SNR difference level of 3 dB. A visual estimate of the
mean is drawn onto the spectrum width plot and a spectrum width difference of 1 m/s is
identified on the y-axis. A horizontal line is drawn to intersect the mean, and the SNR without
interference level at which the spectrum width difference equals 1 m/s second is identified on the
x-axis. To compensate for the residual effect that was described earlier, the values from Table 11
need to be algebraically added to this number. The results of using this technique at I/N level of -
3 dB are shown in Table 12. The data set supports the theoretical analysis results in Section 5.2.4.

Figure 37. Spectrum width regression for the meteorological radar.

79
Table 11. Error Reduction Values
Interference Level Relative SNR w/o SNR – IL (dB)
to Signal (dB) Interference (dB)
6 5.5 -.5
3 2.5 -.5
0 -.5 -.5
-3 -3.5 -.5
-6 -3.75 2.25
-10 -5.25 4.75
-12 -7 5
-15 -7.6 7.4

Table 12. Spectrum Width Results for Example Analysis

Interference Level Relative I/N Level at which a 1 m/s difference occurs
to Signal (dB) (dB)
6 14
3 11.5
0 7
-3 5.5

5.5 Summary of Measurement Results

The interference test results are summarized in Table 13.

Table 13. Measured I/N Thresholds Necessary for Protection of the Meteorological Radar
Reflectivity Spectrum
Interference Signal
Bias Width Bias
CW -7.5 dB -11.5 dB
W-CDMA 4.096 MS/sec -9.5 dB -10.75 dB
CDMA-2000-3X (fwd link) 3.686 MS/sec -7.0 dB -13.5 dB
CDMA-2000-3X (rev link) 3.686 MS/sec -9.5 dB -13.75 dB
TDMA-GMSK 384 kS/sec -8.75 dB -14.0 dB
TDMA–8PSK 384 kS/sec -8.75 dB -11.75 dB

80
These results support the calculated value required for protection of the reflectivity
measurements. As noted above, the measurements and analysis of the mean velocity were
difficult to perform because the mean radial velocity is the least sensitive to interference, and
those results do not necessarily indicate the overall interference levels that the radar can tolerate.
As discussed above, current limitations in the radar calibration and noise removal process
performed by the low level data processor limit the measurement of the necessary protection
criteria for the spectrum width measurements. However, correction of the data for the limitations
of the radar processing do result in values that support the calculated protection values presented
in Section 5.2.4. Improvements to the radar receiver and processor are currently underway which
will allow the system to approach or exceed its originally intended design criteria. This lower
value will become meaningful when those improvements are made. Section 5.6 addresses those
improvements in more detail.

5.6 Meteorological Radar Improvements

The weather radar system that was used for testing is one that has been operating in the United
States for approximately eleven years. Upgrades to these systems that incorporate advances in
signal-processing systems are currently underway. These upgrades will enable signal detection
10 dB below the current level.

The need for these improvements is driven by several requirements: 1) improved measurement
performance above the planetary boundary layer; 2) detection of small water drops and fine mist
precipitation that can result in aircraft icing; and 3) with the event of dual polarization
measurements, improved monitoring of meteorological growth processes. All of these
requirements call for a detection performance that is about 10 dB better than what is achievable
today with current weather radar systems. To meet these requirements the radar’s performance
can be improved by increasing the transmitter power, reducing the receiver’s noise floor, or
increasing the radar’s computational power for better signal processing capabilities.

Among these possibilities, increasing the transmitter power is not cost effective. The alternative
of noise floor reduction could be accomplished by extending the pulse width, because extension
of the pulse width reduces the required bandwidth of the matched filter and thereby reduces the
noise power within the receiver IF. Increasing the pulse width by a factor of 2 increases the
sample volume by a factor of 3 dB. Matching the receiver bandwidth results in a reduction of the
receiver noise by 3 dB. This would lead to an overall detection improvement of 6 dB.
Unfortunately, design limitations on the transmitter duty cycle will not allow extension of the
pulse width for the system used in these tests. The receiver noise temperature could also be
reduced but a reduction of 1 to 2 dB in noise is all that can be achieved with technology that will
be available in the foreseeable future.

Ultimately, the most cost-effective way to achieve these improvements is through enhanced
signal processing. Increasing the radar processing power with upgraded hardware will enable
implementation of data processing algorithms that were not previously available. This additional
processing will utilize coherent integration and frequency domain detection. The radar currently
collects all the parameters necessary for performing these functions, but limited processing
power has prevented implementation. The planned improvements that are currently under way
will eliminate the processing power limitation.

81
Coherent integration, as implemented on this radar, has demonstrated a 10 dB improvement in
detection. Frequency domain detection, the upgrade that is presently underway, will provide
about 10 dB of improved performance. With frequency domain detection, the spectrum is broken
into discrete coefficients, where the actual number of coefficients is determined by antenna
rotation rate and operating mode. In the present storm modes the number of samples ranges from
41 to 111. Processing in the frequency (spectral density calculation) domain results in the desired
signal being confined to a few spectral coefficients while the noise is spread over all the
coefficients at a much lower level. Although about 20 dB of improvement is anticipated through
the use of these two technologies, 10 dB of degradation is expected due to other engineering
changes, including a new feature in which transmitter power will be divided between two
polarizations. The net change, however, is [(20 dB) - (10 dB)]=10 dB.

The improvements to the radar performance enabled by greater processing power do not reduce
the actual noise floor of the receiver, but the effect is a reduction in the effective noise floor by
providing the ability to recover signals of interest at much lower signal levels. The difference
between the actual noise floor and the effective noise floor is the processing detection
improvement. Ultimately, improvements to the radar performance will also cause the radar to be
adversely affected by interference at even lower levels.

5.7 Summary of Interference Effects on a Weather Radar

Weather radars, designed to track particles in the atmosphere and hydrometeors of submillimeter
size, utilize extensive processing to extract signals from received noise. Tests conducted on one
radar type used worldwide have characterized this processing gain as on the order of 6 to 9 dB. In
addition, meteorological radars detect more than just the presence of return pulse energy; the
processing derives data on return pulse characteristics to determine factors such as wind velocity,
wind shear, turbulence, and precipitation type. This processing makes them very vulnerable to
interference.

The test results demonstrate that depending on the radar base product considered, the necessary
protection criteria vary. Since all three products are necessary for proper operation of the radar,
the most stringent values are applicable to ensuring the radar does not experience harmful
interference. Though testing was conducted on a single type of meteorological radar, most
modern meteorological radar systems employ equally complex processing systems that are as
susceptible to interference.

The test results suggest a requirement for a protection value of -9 dB I/N for the base reflectivity
data. The results also show that while the theory predicts a necessary protection value of -14 dB
I/N for spectrum width, the radar is presently sensitive to interference down to an I/N of
approximately -6 dB for that parameter. Introduction of the radar improvements discussed in
Section 5.6 will provide a net improvement of about 10 dB through improved processing power,
resulting in a reduction in the effective noise floor by a value equivalent to the detection
improvement. The end result is that the point where the radar’s actual sensitivity to interference
deviates from the theoretical curve in Figure 38 will decrease by the amount equal to the
processing power. Figure 38 shows the divergence between test results and theory at I/N= -6 dB.
Near-term improvements will provide 10 dB of processing power.

82
Figure 38. Impact of near-term processing improvements on the weather radar interference
threshold, which is expected to ultimately be about I/N=-14.4 dB.

83
 6 INTERFERENCE MEASUREMENTS ON MARITIME
RADIONAVIGATION RADARS

6.1 Introduction

Maritime radionavigation radars may fail to meet their performance requirements if undesired
signals inflict excessive amounts of various types of interference degradation. Dependent upon
the specific interacting systems and the operational scenarios, those types of degradation may
include:
• diffuse effects, e.g. desensitization or reduction of detection range, target drop-outs and
reduction of update rate;
• discrete effects, e.g. detected interference, increase of false alarm rate.

Associated with these varieties of degradation, interference protection criteria may consist of
threshold values of parameters, e.g., for a collision avoidance system:
• tolerable reduction of detection range and associated desensitization;
• tolerable missed-scan rate;
• tolerable maximum false-alarm rate;
• tolerable loss of real targets.

These protection criteria and the thresholds used to derive them for shipborne radionavigation
systems need to be developed further. The operational requirement for maritime radars is a
function of the operational scenario. This is related to the distance from shore and sea obstacles.
In simplistic terms this can be described as oceanic, coastal or harbor/port scenarios.

There is as yet no international agreement on the protection criteria required for radars installed
on ships for the scenarios identified above. The International Maritime Organization (IMO) is
developing a revision to the operational performance standards for maritime radar that takes
account of ITU-R requirements for limits on unwanted emissions from radio services. The IMO
revision, for the first time, gives recognition to the possibility of interference from other radio
services. Most importantly, the IMO has stated without reservation in its recent update of the
IMO Safety of Life at Sea (SOLAS) Convention that radar remains a primary sensor for the
avoidance of collisions. This statement should be viewed in the context of the mandatory fitting
of Automatic Identification Systems (AIS) to some classes of ships.

AIS technology relies upon external references (e.g., GPS) for the verification of relative position
indication in terms of collision avoidance scenarios. However, AIS cannot take account of many
maritime objects (e.g., icebergs, floating debris, small vessels, and wrecks) that are not fitted with
AIS. These objects may potentially cause collisions at sea, and must be detectable by maritime
radars. Radar will therefore remain the primary system for oceanic collision avoidance for the
foreseeable future.

84
Intensive discussion with maritime authorities, including users, has not yet created any
quantitative consensus regarding the minimum required Pd level for any scenarios. Rather, the
instinctive reaction by such authorities is that during any maritime voyage no interference that
can be controlled by regulation is acceptable. In the absence of additional guidance, a Pd value
of 0.9 for maritime targets (within a single scan) has been used as a nominal minimum in
electromagnetic compatibility studies submitted to the ITU-R.

Tests were performed to assess the effects of pulsed emissions and emissions from digital-signal
communication systems on three maritime radionavigation radars that operate with a primary
allocation in the 2900-3100 MHz band and three radars that operate with primary allocation in
the 9200-9500 MHz band. The systems that were tested are IMO category maritime
radionavigation radars that employ scan rates, pulse widths, prfs, IF bandwidths, noise figure,
and antenna beamwidths typical of those identified in a pertinent ITU-R Recommendation [22].
These radars are representative of the types being used by the United States Coast Guard (USCG)
for shipboard navigation, by the commercial shipping industry, and recreational boaters as well.
The radars operating in the 2900-3100 MHz band are identified as maritime radionavigation
Radars A, B, and C and the 9200-9500 MHz radars are identified as maritime radionavigation
Radars D, E, and F. Tests on radar F concentrated on the effects of pulsed interference.

Radars such as those identified in [22] and used in the NTIA tests typically employ interference
mitigation techniques and processing methods identified in another ITU-R Recommendation [18]
to allow them to operate in the presence of other radionavigation and radiolocation radars.
Techniques of that kind are effective in reducing or eliminating low duty-cycle asynchronous
pulsed interference between radars. All of the radars that were tested have some type of
interference rejection circuitry/processing, which by default was enabled during testing.

These tests investigated the effectiveness of each of the radar’s interference suppression circuitry
and software to reduce or eliminate interference due to the emissions from a communication
systems employing digital modulation schemes. Additional tests were also performed using low
duty-cycle pulsed emissions as an interference source. The tests were performed with the
assistance of the radar manufacturers and experienced mariners. Their guidance was used to
properly set up and to operate the radars. This section describes the results of those studies to
date.

6.1.1 Objectives

The objectives of the maritime radar testing were to:

• quantify the capability of each of the six maritime radionavigation radar’s
interference-rejection processing to mitigate unwanted emissions from digital
communication systems as a function of their power level;
• develop I/N protection criteria that would mitigate interference from digital
communication systems emissions in maritime radionavigation radars;
• observe and quantify the effectiveness of each of the maritime radionavigation
radar’s IR techniques to reduce the number of false targets, radial streaks (strobes),
and background noise or “speckle.”

85
 6.2 Description of Maritime Radars Tested

6.2.1 Maritime Radionavigation Radar A

Maritime radionavigation Radar A, which was introduced circa 2000 and is still being refined, is
designed for commercial applications and is an IMO category radar that operates in the
2900-3100 MHz band. Nominal values for the principal parameters of this radar were obtained
from regulatory type-approval documents, sales brochures and technical manuals. These are
presented in Table 14.

Table 14. Technical Characteristics of Maritime Radionavigation Radar A.

Parameter Value
Frequency (MHz) 3050 (+/-30) MHz
Pulse power (peak) into antenna 30 kW
Range 0.375-1.5 3-6 nmi 12 nmi 24-96
nmi nmi
Pulse width 0.08 µs 0.30 µs 0.60 µs 1.2 µs
Pulse repetition frequency 2.2 kHz 1.028 600 Hz
kHz
IF bandwidth 28 MHz 3 MHz 3 MHz 3 MHz
Spurious response rejection 60 dB
System noise figure 4 dB
Receiver noise in given -96 dBm -105 -105 -105 dBm
bandwidth dBm dBm
Antenna mainbeam gain [xx]
RF bandwidth Unknown
Antenna scan rate (scan period) 26 rpm (2.3 sec/rev)
Antenna horizontal beamwidth 1.9 degrees
Antenna vertical beamwidth 22 degrees
Polarization Linear horizontal

The radar uses a multistage logarithmic IF amplifier/detector. This type of receiver design is very
common in marine radionavigation radars since they have to detect targets that have very small
and large returns. A logarithmic amplifier increases the range of target returns that can be
handled by the radar receiver without it becoming saturated.

86
The noise figure of the radar was measured and was found to be 5.3 dB, which was consistent
with the nominal value of 4 dB. The 3-dB IF bandwidth is about 3 MHz for the range scale used
for the tests. Using those parameters the noise power of the radar receiver is calculated to be
about -104 dBm.

Radar A has extensive signal processing and target tracking capabilities, including an adaptive
local CFAR feature and a scan-to-scan correlation feature. The local CFAR (acting within a
small fraction of one range sweep) is known as an ordered-statistic CFAR, which is a type that
permits the desensitizing effect of interfering pulses to be lessened or avoided. This is done by
discarding a selectable number of background signal samples that would otherwise be used in
establishing the detection threshold. The process discards the samples having the greatest
amplitude. As more samples are discarded which contain the higher amplitude interfering pulses,
the less influence they are likely to have on the sensitivity of valid target detection.

Radar A can also perform a scan-to-scan correlation process that provides an additional means
for discriminating between signals that are present consistently, such as a valid target, and signals
that appear at random times, such as asynchronous pulsed interference.

6.2.2 Radars B and D

Radars B and D are maritime radionavigation IMO category type of radars produced by the same
manufacturer and are designed for commercial applications. Radar B operates in the
2900-3100 MHz band while Radar D operates in the 9200-9500 MHz band. Radars B and D
locate their transmitter/receiver below deck and use waveguides to send/receive signals from the
antenna. They use different antennas and receiver front-ends, but have a common display along
with common receiver elements, including the interference rejection processing and IF circuitry.
The radars use a multistage logarithmic IF amplifier and a separate video detector. Radars B and
D also use pulse jitter. The transmitted pulse prf can be jittered to prevent second time around
echoes and also to reduce the interference from other transmitters in the vicinity. This function is
automatically set in the transceiver and provides up to ±25µs jitter about the nominal value.
Nominal values for the principal parameters of these radars were obtained from regulatory
type-approval documents, sales brochures and technical manuals. They are presented in Table 15.

87
Table 15. Technical Characteristics of Maritime Radionavigation Radars B and D

Parameter Value
Frequency 3050 (± 10) 9410 (± 30) MHz
MHz
Pulse power (peak) into antenna 30 kW
Range 0.125-1.5 nmi 3-24 nmi 48 nmi 96 nmi
Pulse width 0.070 µs 0.175 µs 0.85 µs 1.0 µs
Pulse repetition frequency 3.1 kHz 1.55 kHz 775 Hz 390 Hz
IF bandwidth 22 MHz 22 MHz 6 MHz 6 MHz
Spurious response rejection Unknown
System noise figure 5.5 dB
Receiver noise in given -95 dBm -95 dBm -101 -101
bandwidth dBm dBm
RF bandwidth Unknown
Antenna mainbeam gain [xx]
Antenna scan rate (scan period) 24/48 rpm (2.5/1.25 sec/rev)
Antenna horizontal beamwidth 2.8 degrees 1.2 degrees
Antenna vertical beamwidth 28 degrees 25 degrees
Polarization Linear horizontal

The values of pulse width and prf in Table 15 are the default settings for that particular range.
The operator can, for some ranges, select pulse widths and prfs that differ from the default
values.

Pulse-to-pulse and scan-to-scan correlators are used by Radars B and D to mitigate interference
from other radars. For pulse-to-pulse correlation, returns from successive pulses are compared to
reduce interference; a target is displayed only if it is present for consecutive pulses. This IR
function is most effective if the transceiver has been set to provide prf jitter. Scan-to-scan
correlation will only display targets if they are present in consecutive scans. These radars do not
have CFAR processing. More discussion of these radar interference mitigation techniques can be
found in Section 1 and Appendix A of this report, as well as in an ITU-R Recommendation [18].

6.2.3 Maritime Radionavigation Radars C and E

Radars C and E are maritime radionavigation IMO category type of radars produced by the same
manufacturer and were designed for commercial applications. Radar C operates in the
2900-3100 MHz band while Radar E operates in the 9200-9500 MHz band. Radars C and E are a
topmast design. The receiver/transmitter (R/T) is encapsulated in a metal housing located directly
below the rotating antenna. The video from the R/T unit is sent to the PPI located below deck via
cables. They use different antennas and receiver front-ends, but have a common display along

88
with common receiver elements including the interference rejection processing and IF circuitry.
Both of the radars use an eight-stage successive approximation logarithmic IF amplifier/detector.

Nominal values for the principal parameters of these radars were obtained from regulatory
type-approval documents, sales documentation, and technical manuals. They are presented in
Table 16. The values of pulse width and prf in the table are the default settings for that particular
range. The operator can, for some ranges, select pulse widths and prfs that differ from the default
values.

Radars C and E use pulse-to-pulse and scan-to-scan correlators to mitigate interference from
other radars, as described above. These radars do not have CFAR processing.

Table 16. Technical Characteristics of Maritime Radionavigation Radars C and E

Parameter Value
Frequency 3050 (+/-10) MHz 9410 (+/-30) MHz
Pulse power into antenna 30 kW
Range 0.125-3 nmi 6-24 nmi 48-96 nmi
Pulse width 0.050 µs 0.25 µs 0.80 µs
PRF (Hz) 1.8 kHz 785 Hz
IF bandwidth 20 MHz 20 MHz 3 MHz
Spurious response rejection Unknown
System noise figure 4 dB
Receiver noise in given bandwidth -97 dBm -97 dBm -105 dBm
RF bandwidth Unknown
Antenna mainbeam gain 30 dBi
Antenna scan rate (scan period) 25/48 rpm (2.4/1.25 sec/rev)
Antenna horizontal beamwidth 2.0 degrees 1 degree
Antenna vertical beamwidth 30 degrees 15 degrees
Polarization Linear horizontal

6.2.4 Maritime Radionavigation Radar F

Maritime radionavigation Radar F is nearly identical to Radar A, except that its RF front end
operates in the 9200-9500 MHz band. Nominal technical characteristics of this radar are
presented in Table 17. The radar uses a summing multistage logarithmic amplifier with the IF
bandwidths given in Table 17 for each pulse width and associated range. A test point was
provided that is located at the output of the third IF amplifier. A CW signal was swept in
frequency to determine the response of the receiver and measure the IF bandwidth. The result is
shown in Figure C-5. The 3 dB IF bandwidth of the radar when set to short pulse mode 1, which

89
uses a pulse width of 200 ns for a maximum range of 3 nautical miles, was measured to be about
6.5 MHz. This mode was used for all of the tests.

Table 17. Technical Characteristics of Maritime Radionavigation Radar F

Parameter Value
Frequency 9410 (+/-30) MHz
Pulse mode Short Short Medium Medium Long
pulse 1 pulse 2 pulse 1 pulse 2 pulse
Pulse width 80 ns 200 ns 400 ns 700 ns 1.2 µs
Range mode 0.125-1.5 nm 0.5-3 nm 1.5- 6 nm 3-24 nm 6-72 nm
Pulse repetition rate 2.2 kHz 1 kHz 600 Hz
IF bandwidth 27 MHz 4.5 MHz 3 MHz
Spurious response 60 dB
rejection
System noise figure 4 dB
Receiver noise in -96 dBm
given bandwidth
Antenna mainbeam 31 dBi
gain
RF bandwidth Unknown
Antenna scan rate 24 rpm (2.5 sec/rev)
(scan period)
Antenna horizontal 1.5 degrees
beamwidth
Antenna vertical 22 degrees
beamwidth
Polarization Linear horizontal

6.2.5 Radar Video Displays

Radars A and F use the same video display unit. Their enhanced signal processing capabilities
can display various types of targets in different combinations. The radars could display
amorphous, raw-video “blips” (known as the image display), synthetic targets that appeared as
“o” symbols, and/or tracked targets that appeared as “x” symbols. The brightness of the video
image targets roughly corresponded to the power of the target echoes. Figure 39 shows an
example of synthetic and raw video targets on the same PPI display.

Synthetic targets require about 2-3 dB of additional desired power compared to raw-video targets
to obtain the same Pd when operating at minimum detectable signal (MDS) level but do not

90
change their brightness in correspondence with echo strength. That is, if the target power for the
0.9 Pd for blip display was -90 dBm, then the power level to achieve the 0.9 Pd for the synthetic
targets would be about -88 dBm. Adding signal power did not change the intensity of the display
of the synthetic targets. For synthetic targets the radar has a built-in target counter that shows the
number of targets per scan and displays that value on the PPI display.

Radars B and D (from the same manufacturer) use a color CRT to display targets and radar
information to the user, including prf, pulse width, range rings, and other parameters. These
radars do not show synthetic targets and only display raw-video blips. Likewise, Radars C and E
(from another manufacturer) only display raw-video blips. However, the displays used with
radars C and E are monochromatic raster scan types. Besides targets, these displays also indicate
various radar parameters. Like Radar A, for these radars the raw-video blips are brighter for
targets that have stronger return echoes.

Figure 39. Example of synthetic and raw video targets on a PPI display.

6.3 Interference Signal Characteristics

6.3.1 Interference Generation for Radars A Through E

Radar A was tested with a 2 MBit/s QPSK waveform as an interference source. Radars B and C
were tested with 64 quadrature amplitude modulation (QAM), 16 QAM, CDMA-2000, and W-
CDMA signals as interference sources. Radars D and E were only tested with the CDMA-2000
and W-CDMA signals. All interfering signals were on-tune (co-channel) with the radars. The
QPSK signal injected into Radar A was continuous, occurring for a full 360 degrees.

The unwanted CDMA signals that were injected into Radars B, C, D and E were gated to occur at
the same time of the target generation within the same azimuth. The gate time was equal to the
length of time that a stationary interference source would be within the radar’s antenna 3-dB
horizontal beamwidth as it rotates. The QAM signals were not gated. A measured emission
spectrum of the continuous QPSK signal is shown in Figure B-5.

91
Communication test sets were used to generate the DVB-T 16 QAM, DVB-T 64 QAM,
CDMA 2000 and WCDMA signals. Spectrum shots of each of the unwanted signals are shown in
Figures B-8 and B-9. The CDMA-2000 signal was for the reverse link (mobile-to-base) standard
according to interim standard 95 (IS-95) format for cellular mobile telephones. The W-CDMA
signal was for the uplink standard according to the 3GPP 3.5 format. The 16 and 64 DVB-T
QAM signals in Figure B-8 represent the type of modulation scheme that is television camera
crews for electronic news gathering outdoor broadcast (ENG-OB) purposes.

6.3.2 Interference Generation for Radar F

Tests on Radar F concentrated on the effects of pulsed interference from radiolocation radars. In
a departure from most of the work described in this report, and for reasons that will become
clear, effects of pulsed interference were explored up to +40 dB I/N levels. In fact, at one point
during the testing, an attenuator setting was inadvertently set to a lower value than was intended.
As a result, I/N levels as high as +60 dB were injected. At these levels, the LNA at the radar RF
front end was gain-compressed and radar performance was seriously compromised as a result.
The tests were performed with radiolocation waveforms that are representative of the radar
systems that operate in the 9000-9200 MHz and 9300-9500 MHz bands.

Three types of radiolocation waveforms were used for the tests. They are chirped, phase coded,
and unmodulated (simple pulsed) waveforms. The waveforms were gated on for the duration of
the mainbeam dwell time for the radionavigation receiver as if it were scanning past a stationary
object. They were also on-tuned (co-channel) with the radar.

6.3.2.1 Linear Chirped-Pulse Interference Waveform

Table 18 shows the parameters of the chirped interference. They were developed based on the
characteristics of existing 9-GHz radars. The number of pulses per beam dwell is dependent on
the type of radiolocation system being simulated.

Table 18. Chirped-Pulse Interference Waveform Characteristics

Waveform Pulse Prf Pri Duty Chirp Chirp rate
width cycle
Chirp 1 10 µs 750 Hz 1.3 ms 0.8% 10 MHz 1 µs/MHz
Chirp 2 10 µs 750 Hz 1.3 ms 0.8% 50 MHz 5 µs/MHz
Chirp 3 13.6/1.65 5 kHz 0.20 ms 0.8% 660/80 MHz 48.5
µs µs/MHz
Chirp 4 10 µs 2 kHz 0.5 ms 0.4% 400/80 MHz 40 MHz/µs
Chirp 5 80 µs 4.5 kHz 0.22 ms 7.2% 400/80 MHz 5 MHz/µs
Chirp 6 10 µs 515 Hz 1.94 ms 0.91% 45/80 MHz 4.5 MHz/µs
Chirp 7 10 µs 5.15 kHz 1.94 ms 0.88% 460/80 MHz 46 MHz/µs

92
Radar F was tested with linear chirped pulsed interference waveforms. These waveforms
primarily use a chirped modulation scheme as shown in Table 18. The duty cycles are calculated
from the scaled pulse widths.

In some cases, the frequency sweep range of the chirp-pulse generation system used in these tests
was limited by hardware to less than the full chirp range of the corresponding radar emission
being modeled. In such cases, the tests were still performed to fully and accurately replicate the
response of radar receivers to the specified chirp parameters. To accomplish this goal, the chirped
pulses used in the tests were swept across at least twice the -20-dB frequency response range of
the Radar F receiver, at the same rate as the sometimes wider-bandwidth chirp pulses from
potentially interfering sources.

For example in Table 18, the 660-MHz chirp in a 13.6 µs pulse (Rc = (660 MHz/13.6 µs) = 48.5
MHz/us) was not possible to generate with available test equipment. But an equivalent
interference effect was generated with an 80-MHz chirp pulse in an interval of 1.65 µs (Rc = (80
MHz/1.65 µs) = 48.5 MHz/µs), provided that the -20-dB radar IF pass-band of the Radar F
victim receiver is less than 50 MHz wide.

In these tests, the value of Rc was always preserved and the Radar F receiver always saw the
chirped interference across its full receiver IF passband in exactly the same way as it would have
if the chirped interference had been generated across wider bandwidths. That was the key
element in accessing the effects of the interference.

For Radar F, the interference signals were injected into the radar at the same azimuth as the
targets for a duration time equal to the antenna beam sweeping across a stationary object. The
receiver’s noise power measured at the IF test point using a spectrum analyzer in zero span mode
without any targets, radiolocation, or linear chirped pulsed waveforms present in the bandwidth
was about -57 dBm. This indicated a nominal gain of about 40-42 dB in the receiver. The gain
compression point23 using an on-tune CW signal was found to be -25 dBm on the generator panel
display, or -43 dBm at the LNA input. The radar receiver interference suppression circuitry and
software cannot mitigate the effects of receiver saturation.

The Radar F receiver IF output response (amplitude and pulse width) to interference from
chirped pulses is a function of the rate at which the chirped frequency sweeps through the victim
radar receiver pass-band. This rate, called chirp rate (Rc), is given by: Rc = (Bc/τ), where Rc is the
sweep rate in megahertz per microsecond, Bc is the chirp frequency range in megahertz and τ is
the pulse duration in microseconds. Victim radar receivers should not respond to interference on
frequencies outside the -20-dB pass-band of their IF circuitry, assuming that the interference is
below the RF front-end overload threshold.

6.3.2.2 Phase Coded Pulsed Interference Waveform

Table 19 shows the parameters of the phase coded pulsed interference waveform, based on a 13-
bit Barker code sequence.
23
The gain compression point is the value where the LNA is saturated by the input signal and will no longer give a
linear relationship between the input and output signals.

93
Table 19. Phase Coded Pulsed Interference Waveform Characteristics
Waveform Pulse width Prf Sub-pulse width
Phase 1 0.64 µs 1.6 kHz 0.049 µs
Phase 2 20 µs 1.6 kHz 1.54 µs

6.3.2.3 Unmodulated Pulsed Interference Waveform

Table 20 shows the parameters of the unmodulated pulsed interference waveforms. They are
based on the characteristics of existing radars along with higher duty cycles that may be used in
future systems. The starred (*) waveforms did not replicate particular radar signals; they were
used determine the effectiveness of radar signal processing at high duty cycles.

Table 20. Unmodulated Pulsed Interference Waveform Characteristics

Waveform Pulse Prf Pri Duty
width Cycle
Unmod 1 1 µs 8 kHz 125 µs 0.8%
Unmod 2 1 µs 19 kHz 52.63 µs 1.9%
Unmod 3 1 µs 35 kHz 28.57 µs 3.5%
Unmod 4* 1 µs 50 kHz 20.0 µs 5.0%
Unmod 5* 1 µs 75 kHz 13.3 µs 7.5%
Unmod 6* 1 µs 100 kHz 10 µs 10%
Unmod 7* 1 µs 200 kHz 5 µs 20%

6.3.2.4 OFDM/BPSK Interference Waveforms

In addition to the radiolocation waveforms, Radar F was subjected to an on-tune orthogonal

frequency division multiplexing (OFDM/BPSK) waveform. It was used to demonstrate the radar
response to interference of a continuous nature (non-pulsed). These signals were gated for the
duration of the mainbeam dwell time. The OFDM/BPSK waveforms, although not specific to any
type of communication system in the 9-GHz band, are a typical modulation of those types of
waveforms used by communications systems in other bands for broadband, high data rate
applications. It was used to show the contrast in interference suppression capabilities between the
effects of pulsed interference in the radar receiver and a digital modulation that appears noise-
like.

94
 6.4 Target Generation

6.4.1 Target Generation for Maritime Radars A Through E

Ten simulated, equally-spaced, equi-amplitude targets were generated along a radial using RF
signal generators, AWGs, and other miscellaneous equipment (combiners, attenuators, etc.) for
each of the radars out to a 3-nmi range as shown in Figure 40. The target generation system
provided groups of RF pulses with the width and timing to appear as ten individual targets on the
radar’s PPI display. The targets had identical signal power in the radar receiver, simulating RCS
that increased with distance. The number of pulses needed to generate each target depended on
the radar’s characteristics.

The train of transmitter trigger pulses (A) was used to trigger the simulated-target generator. A
free-running pulse generator was used to produce gate pulses (B) representing the amplitude
modulating effect on target return due to the antenna beam. Those pulses gated the train of
transmitter triggers in an AND gate circuit, producing bursts (C) of trigger pulses containing
from 6 to 23 pulses each. Each trigger pulse was applied to an arbitrary waveform generator,
which delayed the trigger appropriately and generated a burst of ten pulses (D), each having the
width of one of the radar’s short pulses. All ten of these occurred within one “sweep” of the
radar, i.e., within the displayed fraction of one pulse repetition interval or PRI. Each of those
pulses in turn modulated an RF signal generator set to a frequency near 3050 or 9410 MHz to
produce a simulated-target-return pulse train. The specific RF signal generator frequency was
adjusted to maximize the radar’s response.

Figure 40. Target generator instrumentation for maritime radionavigation radar tests.

6.4.2 Non-Fluctuating Target Generation for Radar F

The targets for these tests were generated using the same basic hardware as for the other
maritime radars, with ten equally spaced targets generated on a radial and the farthest target

95
being located at the maximum range of 3-nmi. Each target was comprised of 18-19 pulses with
the characteristics of the short pulse 2 mode setting in Table 17. Each target on the radial had the
same power level in the radar receiver.

The ten target pulses triggered by each radar trigger all occur within the return time of one of the
radar’s short range scales, i.e., one “sweep”. Consequently, the pulses simulate ten targets along a
radial, i.e., a single bearing. For adjustment of the display settings, the RF power of the target
generator was set to a level so that all ten targets were visible along the radial on the PPI display
with the radar’s video controls set to positions representative of normal operation. The pulse
repetition rate of the target generator (waveform B) was adjusted so the targets would appear at
the same azimuth on consecutive scans of the PPI. The target generator timing diagram is shown
in Figure 41.

A number of trials were performed to determine the target signal power that would result in a Pd
of 90 percent without the interference radiolocation or linear chirped pulse waveforms being
present. This value was found to be about -70 dBm at the panel display of the target generator.
With RF losses, the target power supplied to the LNA input of the radar receiver was -88 dBm.
The noise figure was measured to be about 9 dB. This results in a calculated noise power of
about -97 dBm in the (approximately) 6-MHz IF bandwidth of the radar receiver. Therefore, the
signal-to-noise value to achieve the Pd of 90 percent was about 8-10 dB. Note that the accuracy
of this measurement is probably within ±2 dB.

While the Radar F antenna was not used for the tests, appropriate signals from the antenna
positioner circuitry were supplied to the radar receiver to mimic the antenna’s normal rotation.

Figure 41. Target generator timing diagram for maritime radionavigation radar tests.

96
6.4.3 Fluctuating Target Generation for Radar F

For maritime Radar F, tests were also performed with fluctuating-power targets that changed
their power levels on a scan-by-scan basis according to Swerling case 1 statistics24 (see Section
2.3 of this report). All ten targets on the PPI radial were varied by the same amount; each
Swerling level was held constant for all ten targets for a total of twenty scans, then the next
Swerling level was used for all ten targets for the next twenty scans, and so forth. Thus each data
point represented the counting of (10 targets per scan x 20 scans per level)=200 injected targets at
each Swerling level. A total of twenty Swerling levels were used, and were programmed to run in
a random order from one level to the next. The Swerling power levels were applied relative to the
nominal value of RF power that gave Pd=0.9 without interference; their order and values are
shown in Table 21. For example, the Pd=0.9 nominal target power value was -88 dBm and on the
first scan the power was adjusted by +2.1 dB to be -85.9 dBm, on the second scan the power was
-86.5 dBm, and so on.

For the tests, the signal levels of all targets were adjusted to produce stationary target detections
consistent with a fixed Pd of about 0.9. This Pd value was chosen to reflect the case that the Pd
can never be 100 percent due to propagation effects, interference and other factors. As of the time
that this report was written, the IMO has not specified a minimum Pd for marine radionavigation
radars. The IMO performance standard25 does specify target types, RCS, and the minimum
ranges to detect them. The IMO is developing a minimum Pd for these types of radars and should
publish this value in the near future.

The target power level output was controlled by a computer and sequenced through the twenty
Swerling values of Table 21 without interruption to accurately reproduce the true effect of a
scan-to-scan scenario. Note that in other interference tests the radar was allowed about five scans
to ‘settle’ between each data point. A number of baseline runs through the sequence were
performed without interference, versus with the OFDM signal being injected. Effects of
fluctuating levels on target Pd without any interference are noted in Table 21.

24
Swerling case 1 target levels were generated by summing the squares of pairs of Gaussian-distributed real
numbers.
25
Extracts from IMO Resolutions A222(VII), A278(VIII), A477(XII) for radar equipment required by Regulation
12, Chapter 5 of the IMO-SOLAS Convention.

97
Table 21. Swerling Case 1 Target Power Levels (Relative to Nominal Value of -88 dBm)
Signal levels in dB Relative to Median Power Level and Effect on Target Pd
Without Interference
Scan number dB Effect Scan number dB Effect
Scan 1 +2.1 Pd 100% Scan 11 +2.7 Pd 100%
Scan 2 +1.5 Near 100% Scan 12 +0.90 Pd higher
Scan 3 +3.3 Pd 100% Scan 13 -5.6 No targets
Scan 4 +0.3 Baseline Scan 14 +7.3 Pd 100%
Scan 5 +4.5 Pd 100% Scan 15 -0.3 Reduced Pd
Scan 6 -7.2 No targets Scan 16 -2.5 No targets
Scan 7 -1.7 Few targets Scan 17 -3.3 No targets
Scan 8 -14.4 No targets Scan 18 -9.5 No targets
Scan 9 +4.0 Pd 100% Scan 19 -4.3 No targets
Scan 10 +5.7 Pd 100% Scan 20 -1.0 Reduced Pd

6.5 Maritime Radar Test Conditions

The tests were performed with the following parameters set on the maritime radionavigation
radars as shown in Table 22.

For all of the radars, the STC and FTC features could be activated at the operator’s discretion. As
noted in Sections 1.10.1 and 1.10.2, STC suppresses sea-clutter returns by attenuating strong
echoes at short ranges, while FTC suppresses rain clutter by differentiating echo returns after
envelope detection.

For each of the radars that were tested, baseline values for the software functions that controlled
the target and background brilliance, hue, and contrast settings were found through
experimentation by test personnel and with the assistance of the manufacturers and with
professional mariners that were experienced with operating these types of radars on ships of
various sizes. Once these values were determined, they were used throughout the test program for
that radar.

Table 22. Maritime Radar Test Control Settings

Parameter Value
Sensitivity time control (STC) Disabled
Fast time constant (FTC) Disabled (default)
Interference rejection (IR) On (default)
Automatic gain control On (default)
Radars A and F image selected Raw video (“image”) and/or
synthetic targets
Radars B, C, D, E Raw video
Range scale 3 nmi

98
6.5.1 Interference Suppression Features in Radars A and F

In addition to conventional STC and FTC features for clutter reduction, Radars A and F deserve
special mention of the following capabilities to minimize clutter and RF interference: ordered
statistic CFAR, spike suppression, clutter mapping, and scan-to-scan correlation. As shown in
Table 22, STC and FTC were disabled for the tests, since they are used to discriminate sea clutter
returns from target returns and to offset the effects of rain, respectively. Obviously, sea clutter
and rain did not affect these measurements as they were performed on a test bench. A brief
description of the other interference mitigation features follows.

The spike suppression circuitry provides a means of instantaneously filtering interference from
other transmitters by detecting and eliminating spikes that occur at a given range over three
adjacent sweeps based on a maximum rise and fall criteria. The circuit substitutes the spike value
with the average of the amplitudes on the previous and subsequent sweeps.

The CFAR uses an ordered statistic (OS) technique. This adaptive technique minimizes clutter
breakthrough in large homogeneous clutter areas. It permits target returns near or above the peak
noise plus clutter level to be detected while eliminating the bulk of the noise and clutter. The OS
CFAR operates by automatically adjusting the detection threshold based on an instantaneously
derived estimate of the predominant noise plus clutter level in the vicinity of the test cell. A
programmable guard cell region allows the OS CFAR to accommodate targets of extended range
run length, such as supertankers, without loss of sensitivity while still discriminating against
clutter.

The clutter mapping compensates for unique spatially distributed clutter situations, such as ones
that might be caused by multi-path, grazing angle versus sea state, or a combination of such
conditions. The radar operator is provided a means of biasing the OS CFAR derived threshold on
an area-by-area basis through the use of a threshold bias map (TBM). The threshold offset stored
in the TBM for a particular area is added to the OS CFAR computed threshold for all detections
occurring in that area. There are sixteen TBM areas centered around the radar’s position which
permit the specification of an annulus clutter filter area, and which may be used to counter effects
that are range dependent but bearing independent. The TBM area may be defined to negatively
bias the OS CFAR generated threshold to increase sensitivity or completely mask returns such as
those from land mass. The TBM is organized as a table of 1024 range cells by 1024 azimuth
cells.

The scan-to-scan correlator takes advantage of the fact that sea clutter is correlated on a pulse-by-
pulse basis, but de-correlated on a scan-by-scan basis. Clutter that is permitted to pass through all
previous clutter suppression stages is processed through a temporal-spatial de-correlation filter,
the retrospective processor. This processor performs scan-to-scan correlation, maintaining as
many as nine scans of data.

99
 6.6 Maritime Radar Test Procedures (Non-Fluctuating Targets)

For each radar that was tested, the RF power output of the target generator system was adjusted
so that the target Pd was about 90 per cent without unwanted signals being present, with the
baseline PPI target and background display settings. Table 22 lists the target power at each
radar’s RF input that was required to obtain a Pd of 0.90. Once these values were determined,
they were used throughout the tests.

Table 23. Target Power Levels (Non-Fluctuating) Required to Achieve a Pd of 0.90

Radar Target power at RF input for a Pd of 0.90

A -90 dBm
B -89 dBm
C -77 dBm
D -89 dBm
E -86 dBm
F -88 dBm

For Radars A, C and E, the appropriate levels of unwanted signal powers that were required to
produce the I/N levels within the radar receivers was determined using the calculated receiver
noise power calibrated to the receiver’s waveguide input. The receiver noise power was
calculated using the IF bandwidth and noise figure. Any differences in bandwidths between the
radar receiver and the test signals were accounted for in setting the I/N levels.

The appropriate levels of unwanted signal powers that were required to produce the I/N levels
within radar receivers B and D were determined by monitoring the output of the IF circuitry at a
test point located at the detector input with the spectrum analyzer. The spectrum analyzer was set
to zero-span mode and the value of the radar receiver noise power at the IF test point, without
any unwanted signal being present, was measured and recorded. The unwanted signal was then
injected into the radar RF front-end and the noise power at the IF test point was monitored for a
3-dB increase as the power level of the unwanted signal was also increased. A 3-dB increase in
the receiver noise power is equal to an I/N of 0 dB. Once the value of the unwanted signal that
generated the I/N of 0 dB was found, the unwanted signal power levels that generated the other
I/N values were easily determined. The power levels of the unwanted signals were controlled
using step attenuators or the test set panel display.

For Radars B and D, the number of targets on each radial was counted for 50 simulated rotations
of the antenna for each I/N level for each type of unwanted signal. The Pd was calculated by
dividing the number of counted targets by the total number of targets that were generated.

For Radars A, C and E, observations of the relative strength or brightness of the targets displayed
on the PPI were performed and documented at the various I/N levels. The nature of the effect of
the interference on Radars A, C, and E target displays prevented performing an actual “count” of
the targets because all of the targets tended to “fade” at the same rate. These effects included a

100
“dimming” of the targets, an increase in the number of false targets, radial streaks (“strobes”),
and an increase in background “speckle” or noise.

6.7 Test Results for Maritime Radionavigation Radars

6.7.1 Radar A (3 GHz)

Figure 42 shows a digital photograph of Radar A’s PPI baseline operating state (no interference
injected). Note that the raw-video targets appear along a radial at about 320 degrees. Local clutter
returns from buildings and slight speckling are also visible on the radar display.

Observations of video image targets on the radar’s PPI display were made with emissions from
the QPSK generator applied to its receiver. The power level of the QPSK emission was adjusted
until the appearance of the radar’s PPI was in a baseline condition. The power level of the QPSK
waveform was adjusted within a range of values to find the level where the QPSK emissions did
not adversely affect the performance of the radar in displaying video targets. Figures 43 and 44
are photographs of the radar’s PPI that show the effects of the QPSK waveform at power levels
of -112 and -102 dBm (measured within a 3-MHz bandwidth), respectively. The radar’s receiver
noise power was approximately -104 dBm. The resulting I/N ratios were -8 dB and +2 dB.

101
Figure 42. Radar A baseline state (no interference) with video targets.

Figure 43. Radar A with QPSK interference at I/N = -8 dB. Targets have gradually precessed
on the display since the time that the picture of Figure 42 was taken.

102
Figure 44. Radar A with QPSK interference at I/N = +2 dB. Targets are on same radial as in
Figure 43. Interference was turned off during the indicated segment of the scan at upper left. The
black lines (Moiré pattern) in that sector were not present on the radar screen; they are an artifact
of the reproduction of this image.

Table 24 summarizes the results for Radar A. The images show that the QPSK emissions caused
an increase in the background noise or speckle. In comparing Figure 41, which is the radar
baseline state without interference, to Figure 42 (which has an I/N of -8 dB) the background
speckle has increased but the targets are still detected and displayed. In Figure 44 the I/N is +2
dB and the QPSK emissions have increased the background noise to the extent that the targets are
becoming indistinguishable from the speckle.

Table 24. Radar A Responses to QPSK Interference

I/N Ratio of Interference Effect

-8 dB Slight increase in background speckle; targets distinguishable
-7 dB More increase in background speckle; targets distinguishable
-6 dB Target visibility degraded due to increasing speckle effect
+2 dB Targets becoming indistinguishable from background speckle

103
The power level of the QPSK emissions was adjusted to find the point where the video targets
were still clearly visible and the background speckle was similar to the baseline level. That power
level was found to be about -111 dBm at the receiver input, for an I/N ratio of about -7 dB.

It is important to note that the test targets on the radial are more visible than “real world” targets
that would be distributed anywhere on the radar’s PPI. Therefore, care needs to be taken in
interpreting radar presentations in the presence of noise. The I/N values were not based on one
specific photograph per se. The photographs in this report are representative of the interference
condition. Some of the radar’s scans might show a worse state (denser speckle/false targets)
while others might show a better state (clearer PPI) at the same I/N level. Approximately 20
scans were observed at each I/N level in choosing the I/N values represented in Figures 42
through 44.

6.7.2 Radar B (3 GHz)

For Radar B it was possible to observe the effect that the unwanted signals had on individual
targets. For each unwanted signal, it was possible to count the decrease in the number of targets
that were visible on the PPI as the I/N level was increased. Target counts were made at each I/N
level for each type of interference. A baseline target Pd count was performed before the
beginning of each test. The results of the tests on Radar B are shown in Figure 45, which shows
the target Pd versus the I/N level for each type of interference. The baseline Pd in Figure 45 is
0.93 with the 1-sigma error bars 0.016 above and below that value. Note that each point in Figure
45 represents a total of 500 desired targets.

Figure 45. Radar B Pd curves.

104
Figure 45 shows that, except for the case of the pulsed interference, the target Pd was reduced
below the baseline Pd used in these tests minus the standard deviation for I/N values above
-12 dB for all of the unwanted signals that used a digital modulation. The QAM interference
caused the quickest drop in the Pd as the I/N was increased. Data was not taken for higher I/N
values above of -3 dB for QAM because all of the targets were gone on the PPI above that level.
The CDMA 2000 had the least effect on the target Pd, but it was still causing a drop in the target
Pd at I/N values above -12 dB.

The performance of Radar B was highly robust against in the presence of pulsed interference up
to I/N levels of +39 dB. The low duty cycle of the pulsed interference allowed the radar’s
interference rejection mechanism (which relies on non-coherence between the desired and the
undesired pulse repetition sequences) to work effectively.

6.7.3 Radar C (3 GHz)

For Radar C it was difficult to count the decrease in target Pd as the interference was injected into
the radar’s receiver. The interference caused all of the targets to fade at the same rate no matter
where they were located in the string of targets. It was not possible to make individual targets
“disappear” as the interference power was increased, and count the number of lost targets in
order to calculate the Pd. Therefore, the data taken for Radar E reflects whether or not the
appearance of all the targets was affected at each I/N level for each type of interference. The data
for Radar C are summarized in Tables 25 and 26.

Table 25. Radar C Responses to Continuous QAM Interference

I/N Ratio 64 QAM 16 QAM
-12 dB No effect No effect
-10 dB No effect No effect
-9 dB Targets slightly dimmed Targets slightly dimmed
-6 dB Targets dimmed Targets dimmed
-3 dB Targets not visible Targets not visible
0 dB Targets not visible Targets not visible
+3 dB Targets not visible Targets not visible
+6 dB Targets not visible Targets not visible

The data in Table 25 show that the unwanted QAM signals affected the visibility of the targets
for Radar C on its PPI at an I/N level of -9 dB. At that level the brightness of the targets on the
PPI was slightly dimmed from their baseline state. At I/N levels of -6 dB they were dimmed more
and for I/N levels above -3 dB the targets had dimmed so much that they were no longer visible
on the PPI display.

The data in Table 26 show that the unwanted CDMA signals affected the visibility of the targets
for Radar C on its PPI at an I/N level of -6 dB. At that level the brightness of the targets on the

105
PPI was noticeably dimmed from their baseline state. At I/N levels of -3 dB and above, the
targets had dimmed so much that they were no longer visible on the PPI.

For Radar C, the gated 2.0 and 1.0 µs pulsed interference with duty cycles of 0.1 and 1.0 percent
did not affect the visibility of the targets on the PPI at the highest I/N level, which was 40 dB.

Table 26. Radar C Responses to Gated CDMA Interference

I/N Ratio W-CDMA CDMA-2000

-12 dB No effect No effect
-10 dB No effect No effect
-9 dB No effect No effect
-6 dB Targets dimmed Targets dimmed
-3 dB Targets not visible Targets not visible
0 dB Targets not visible Targets not visible
+3 dB Targets not visible Targets not visible
+6 dB Targets not visible Targets not visible

6.7.4 Radar D (9 GHz)

For Radar D it was possible to observe the effect that the unwanted signals had on individual
targets. For each unwanted signal, it was possible to count the decrease in the number of targets
as the I/N level was increased. Target counts were made at each I/N level for each type of
interference. A baseline target Pd count was performed before the beginning of each test. The
results of the tests on Radar D are shown below in Figure 46 with the target Pd versus the I/N
level for each type of interference. The baseline is shown at a Pd of 0.92 with the 1-sigma error
bars 0.016 above and below. Note that each point in Figure 46 represents a total of 500 desired
targets.

The performance of Radar D was highly robust against in the presence of pulsed interference up
to I/N levels of +39 dB. The low duty cycle of the pulsed interference allowed the radar’s
interference rejection mechanism (which relies on non-coherence between the desired and the
undesired pulse repetition sequences) to work effectively.

The data in Figure 46 show that, except for the case of the pulsed interference, the target Pd was
reduced below the baseline Pd used in these tests minus the standard deviation for I/N values
above -12 dB for the unwanted CDMA signal.

106
Figure 46. Radar D Pd curves.

6.7.5 Radar E (9 GHz)

As in the case of Radar C, for Radar E it was difficult to count the decrease in target Pd as the
interference was injected into the radar’s receiver. The interference caused all of the targets to
fade at the same rate no matter where they were in the string of targets. It was not possible to
make individual targets “disappear” as the interference power was increased. Therefore, the data
taken for radar E reflects whether or not the appearance of all the targets was affected or not at
each I/N level. The data for Radar E are summarized in Table 27.

Table 27. Radar E Responses to Gated CDMA Interference

I/N Ratio WB CDMA CDMA-2000

-12 dB No effect No effect
-10 dB No effect No effect
-9 dB No effect No effect
-6 dB Targets dimmed Targets dimmed
-3 dB Targets dimmed Targets dimmed
0 dB Targets not visible Targets not visible
3 dB Targets not visible Targets not visible
6 dB Targets not visible Targets not visible

107
The data in Table 27 show that the unwanted CDMA signals affected the visibility of the targets
for Radar E on its PPI at an I/N level of -6 dB. At that level the brightness of the targets on the
PPI was noticeably dimmed from their baseline state. At I/N levels of 0 dB and above, the targets
had dimmed so much that they were no longer visible on the PPI.

For Radar E, the gated 2.0 and 1.0 µs pulsed interference with duty cycles of 0.1 and 1.0 percent
did not affect the visibility of the targets on the PPI at the highest I/N level, which was 40 dB.

6.7.6 Radar F (9 GHz)

The test results for the maritime radionavigation Radar F are contained in Table 28. The table
shows that the radar did not suffer any degradation to its performance with any of the chirped or
phase-coded waveforms up to an I/N of +40 dB. For the unmodulated pulses the radar did not
suffer any degradation at an interference duty cycle of 5 percent and an I/N of +40 dB. At the
higher duty cycles for the unmodulated radiolocation waveform, the radar produced a strobe on
the PPI display at the target azimuth when the I/N exceeded +20 dB. At lower I/N ratios the
strobe on the PPI was transformed into an increase in the amount of false targets or speckle.
These effects are shown in photographs of the radar’s PPI in Figures 47 and 48. These
photographs are representative examples of the PPI condition; there was variation from scan-to-
scan in the number of false targets.

Figure 47. 1-µs pulsed interference at 7.5% duty cycle and I/N= +40 dB.

108
Figure 48. 1-µs pulsed interference at 7.5% duty cycle and I/N= +12 dB.

Regarding Figures 47-48, the observers knew in advance where the targets would be displayed
and that knowledge made it easier for them to distinguish the targets from interference speckling.
Radar operators viewing ordinary targets in an operational environment would not have that
condition available to them, making the targets harder to distinguish from background speckle (as
seen off the target radial) in these in real-world scenarios. The off-radial speckling is due to
random radar receiver noise.

For the OFDM interference (Figures 49-50), the effects on the PPI were noticeable at I/N = +3
dB as false targets, and strobes were produced above that I/N value. At I/N values of 0 dB and -3
dB, the radar’s CFAR circuitry performed exactly as it was supposed to, reducing the Pd as the
overall noise level increased. (The OFDM signal would appear to the radar as broadband noise.)
At I/N = 0 dB the Pd was 0.83 and at I/N = -3 dB the Pd was 0.85. At I/N = -6 dB the radar
appeared to have recovered to its baseline state. This response demonstrates that the radar was
able to reject most of the pulsed interference, but it did not effectively mitigate the OFDM
interference effects.

109
Figure 49. OFDM interference effect at I/N= +3 dB.

Figure 50. OFDM interference effect at I/N= +6 dB.

110
6.7.6.1 Radar F with Fluctuating Targets

Interference tests were performed with the targets fluctuating in power as described in 2.4.1 of
this report using an OFDM interference source. The results showed that there was a small gain in
the target Pd at low I/N levels when the target power fluctuated “upwards” (scans 1-5, 9-12, and
14 in Table 21). That is, the target Pd increased because the target power was above the level that
produced the Pd of 0.90 percent. However, this effect has limits. When the I/N value was around
0 to 3 dB, an increase in target power did not further enhance the target Pd even when the target
power increased to the higher values in Table 3. This was the effect even when the target power
was increased to its maximum Swerling value of 7.3 dB (scan 14 in Table 21) above the baseline
power level.

Conversely, when the target power fluctuated “down” from the baseline value that produced the
Pd of 0.90 percent, the targets were already weak and the interference made them disappear. In
fact, when the target power was only 1.7 dB (scan 7 in Table 21) below the baseline state, the
targets were severely degraded without interference. When the interference was injected at low
I/N levels, the targets disappeared.

Table 28. Radar F Responses to Interference*

Interference Effect at Given I/N Ratio (dB)
Waveform -9 -6 -3 0 +3 +6 +9 +12 +20 +40
Chirp 1 None none none none none none none none none none
Chirp 2 None None None None None None None None None none
Chirp 3 None None None None None None None None None None
Chirp 4 None None None None None None None None None None
Chirp 5 None None None None None None None None None None
Chirp 6 None None None None None None None None None None
Chirp 7 None None None None None None None None None None
Phase1 None None None None None None None None None None
Phase2 None None None None None None None None None None
Unmod None None None None None None None None None None
pulsed 1
Unmod None None None None None None None None None None
pulsed 2
Unmod None None None None None None None None None None
pulsed 3
Unmod None None None None None None None None None None
pulsed 4
Unmod None None None None None None None False strobe strobe
pulsed 5 targets
Unmod None None None None None None None False strobe strobe
pulsed 6 targets
Unmod None None None None None None False False strobe strobe
pulsed 7 targets targets
OFDM Pd= Pd= Pd= Pd= False strobe strobe strobe strobe strobe
0.93 0.90 0.85 0.83 targets
* When no effect is indicated, the interference Pd was within ±3 percent of the baseline Pd, and the number
of false targets or “speckle” was at the baseline state as well.

111
The results of these tests with fluctuating targets have demonstrated that:

• Trying to establish interference criteria using fluctuating targets is pointless because

changing the I/N value while changing the target power only leads to an uncontrolled
experiment without a baseline measure or state;

• Adding target power to the radar can only help at low values of I/N levels even when the
target power is increased to 7.3 dB above the baseline state. Once the interference causes
a high increase of false targets and strobing, the radar cannot recover the targets even
when the target power increases to its maximum Swerling Case 1 value.

Both of these points are very important and the second one is a key element for radar analyses.
There is currently a misconception that by merely increasing the target power, a radar can
overcome any level of interference. These test results have disproved that assumption. It would
be true if target power were adjusted along with detection levels, but that would tantamount to
building a new radar with a more powerful transmitter or a bigger antenna.

6.8 UWB Interference Tests on a Maritime Radar

This section describes the results of UWB interference susceptibility tests that were performed on
maritime radionavigation Radar A. The purpose of the tests was to inject impulse UWB
interference into the radar’s receiver and observe and document the degradation of the radar’s
performance as the power level of the UWB signal in the radar receiver was varied. The tests
were performed closed-loop, as described above.

6.8.1 UWB Interference Test System

The test system and target generator were the same as described for Radar A in Section 6. These
tests used UWB signals with prfs of 100 kHz, 1 MHz, and 10 MHz. The UWB transmitter unit
that was used for these tests was supplied by the Time Domain Corporation. This UWB device
had been used in previous UWB/GPS testing by NTIA. The UWB pulser was connected to an
AWG that provided triggering for dithered and non-dithered UWB waveforms. The RF outputs
of the target generator and the UWB pulser were combined and connected to a cable-to-
waveguide adapter that was then connected to the mixer input of the radar receiver. In effect, this
removed the radar antenna from the system but still allowed for normal operation of the radar.
The IF output of the radar was monitored with a spectrum analyzer by connecting to a test point
on the receiver circuit board.

112
6.8.2 UWB Interference Test Parameters

The RF power output of the target generator system was adjusted so that the target Pd was as
close as possible to 0.90 without UWB interference being present. This value was approximately
-90 dBm at the waveguide input of the receiver. Figure 42 shows a photograph of the baseline
operating state of the radar. The targets appear along a radial, and local clutter returns from
buildings are visible on the radar display.

The IR feature of the radar was activated for these tests. The STC and FTC functions were turned
off and the auto gain function was selected. This configuration is a normal operating mode for
the radar, as confirmed by USCG personnel who are certified to operate the radar on Coast Guard
vessels.

6.8.3 UWB Test Method

The power of the target generator was held at a constant -90 dBm at the waveguide input. The
UWB signal power was varied with an adjustable attenuator and was gated to coincide with the
target radial. This simulated the radar’s mainbeam sweeping across a UWB device. The radar
display was then observed for interference effects. UWB interference effects were documented
with digital photographs of the ppi. Between tests, the interference power was reduced until the
radar returned to baseline operation.

6.8.4 UWB Test Results

The test results for each UWB prf are discussed below. Overall, the results showed that for
moderate UWB power levels, the UWB interference caused strobes along the radial of the
targets. The targets were impossible to distinguish within the strobes when the UWB interference
occurred. As the UWB power was decreased, the strobes become less intense and at some point
transformed into numerous false targets. So the effect of the UWB interference at low levels was
an increase in the false targets, while at higher levels the desired targets were masked altogether.
Figures 51-65 show the PPI display under various interference circumstances. The results
showed that for dithered UWB waveforms, independent of the UWB prf, the maximum
allowable UWB power in the radar receiver which would let the radar operate at the baseline
performance level was about -115 dBm/MHz (root mean square (RMS) average power detected).

6.8.5 Effect of UWB Interference on Synthetic Targets

Figures 66-68 show radar PPI displays with synthetic targets in the presence of UWB
interference. As shown in those figures, UWB interference was observed to suppress synthetic
targets on desired targets while generating false synthetic targets at other locations on the display.
Synthetic target generation did not perform well in the presence of UWB interference.

113
6.8.6 Comparison to I/N Values

The noise power of the radar receiver for an IF bandwidth of 3 MHz and noise figure of 5.3 dB is
-104 dBm. In a 1-MHz bandwidth this is about -109 dBm/MHz. The maximum allowable UWB
power was about -115 dBm/MHz, equating to an I/N of -6 dB to -7 dB. This value matches the
interference threshold used in the UWB analyses in [23].

6.8.7 Additional Observations on UWB Interference Effects

Adding more power to the desired targets, i.e., making the targets “stronger,” did not help the
radar overcome the effects of UWB interference. When the target power was increased by 3 dB
the UWB interference still caused false targets to appear. This effect is shown in Figures 66-67.
The UWB interference did not resemble clutter returns.

6.9 Summary of Interference Effects on Maritime Radars

The results of interference tests on maritime radionavigation radars show that when interference
with the characteristics of digital communication signal modulations is coupled into these radars
at I/N levels of -6 dB, some of the radars begin to show degradation effects such as dimmed
targets, lost targets, or false targets. For maritime radars with logarithmic IF amplifier/detectors
(radars A, C, E, and F), these effects began to be manifested at slightly higher I/N levels; the
targets were either not visible or dimmed at the I/N levels of -3 dB and -6 dB, as indicated in
Tables 24-27 (and Table 28 for OFDM interference into Radar F). The effects of interference on
Radars A, C, and E were maximized (i.e., the targets had disappeared from the PPI and no other
effects were visible) at I/N levels between 0 dB and -10 dB. For Radars B and D (which use a
logarithmic amplifier and separate video detector), at the I/N level of -6 dB, the target Pd values
dropped below the baseline 1-sigma error. These test results show that at an I/N of -10 dB, for
Radars A, C, and E, the targets were no longer dimmed and for radars B and D, the target Pd
values were slightly below the baseline 1-sigma error. For Radar A the synthetic targets required
about 2 dB to 3 dB of additional desired signal power compared to the raw-video targets to
obtain the same Pd level when operating at a minimum detectable signal level, but the appearance
of the targets was not brighter on the PPI display, as shown in Figure 66.
The test results for Radars B, D, and F show that the maritime radars can withstand low-duty
cycle, asynchronous pulsed interference at extremely high (+30 dB to +40 dB) I/N levels due to
the inclusion of radar-to-radar interference mitigating circuitry and/or signal processing. The
radar-to-radar interference mitigation techniques of scan-to-scan and pulse-to-pulse correlators
and CFAR processing [18, 19] have been shown to work well. But the test results show that the
same techniques do not mitigate continuous emissions that appear noise-like or CW-like within
the radar receiver.

As most maritime radionavigation radars in the 2900-3100 and 9200-9500 MHz bands are very
similar in design and operation, it is expected that these test results should apply equally well to
other models of marine radars in these bands.

114
Determining an exact boundary for acceptable levels of interference effects for these types of
radars can be somewhat subjective due to such variables as the eyesight and experience of radar
operators, possible fatigue due to staring at PPI displays for long periods, and grading of the
brightness and distinguishability of the targets relative to the background noise on the PPI
displays. But the design of these radars permits no alternative test techniques, and the tests as run
do replicate actual operational conditions.

Experienced, trained radar operators may be better able to discern real targets from false targets,
interference and/or clutter than inexperienced ones. But for these tests, the radar manufacturers
provided radar design engineers and the UK Maritime and Coast Guard Agency (MCA) provided
experienced radar operators and instructors. Those personnel all concurred with the results and
conclusions of these tests.

The radar could only marginally process out UWB interference and then only at low power levels
such as -115 dBm/MHz for a 1-MHz prf signal. This was probably due to the high prf of the
UWB signals.26 UWB interference did not resemble clutter returns.

Figure 51. Radar A PPI display with 100-kHz prf gated UWB signal at -85 dBm/MHz (RMS
detection).

26
At high prf values, there is a high probability that a UWB pulse will occur at the beginning of the IR
gate and that another UWB pulse will occur at the end of the IR gate.

115
Figure 52. Radar A PPI display with 100-kHz prf gated UWB signal at -95 dBm/MHz (RMS
detection).

Figure 53. Radar A PPI display with 100-kHz prf UWB signal at -105 dBm/MHz (RMS
detection).

116
Figure 54. Radar A PPI display with 100-kHz prf UWB signal at -110 dBm/MHz (RMS
detection).

Figure 55. Radar A PPI display with 1-MHz prf UWB signal at -115 dBm/MHz (RMS
detection).

117
Figure 56. Radar A PPI display with 1-MHz prf UWB signal at -110 dBm/MHz (RMS
detection).

Figure 57. Radar A PPI display with 1-MHz prf UWB signal at -105 dBm/MHz (RMS
detection).

118
Figure 58. Radar A PPI display with 1-MHz prf UWB signal at -95 dBm/MHz (RMS
detection).

Figure 59. Radar A PPI display with 1-MHz prf UWB signal at -85 dBm/MHz (RMS
detection).

119
Figure 60. Radar A PPI display with 1-MHz prf UWB signal at -75 dBm/MHz (RMS
detection).

Figure 61. Radar A PPI display with 10-MHz prf UWB signal at -116 dBm/MHz (RMS
detection).

120
Figure 62. Radar A PPI display with 10-MHz prf UWB signal at -111 dBm/MHz (RMS
detection).

Figure 63. Radar A PPI display with 10-MHz prf UWB signal at -106 dBm/MHz (RMS
detection).

121
Figure 64. Radar A PPI display with 10-MHz prf UWB signal at -86 dBm/MHz (RMS
detection).

Figure 65. Radar A PPI display with 10-MHz prf UWB signal at -66 dBm/MHz (RMS
detection).

122
Figure 66. Radar A PPI display with high power video targets and synthetic targets, baseline
state (no interference). Compare to Figure 67, which shows more false targets when UWB
interference was injected.

Figure 67. Radar A PPI display with 1 MHz UWB interference suppressing desired raw video
targets and synthetic targets while generating false synthetic targets. (Compare to the baseline
state of Figure 66.)

123
Figure 68. Radar A PPI display showing a loss of video targets and generation of false
synthetic targets due to 1-MHz UWB interference.

124
 7 INTERFERENCE MEASUREMENTS ON AN AIRBORNE METEOROLOGICAL
RADAR

7.1 Introduction

This report describes the methodology and results of interference tests on a 9-GHz airborne
weather radar. Interference waveforms were injected into the radar at its RF frequency and the
effects of the interference were recorded as a function of I/N ratio in the radar receiver IF stage.
Interference waveforms that were tested included: OFDM, unmodulated pulsed, chirped pulsed,
phase-coded pulsed, and BPSK.

7.2 Characteristics of the Airborne Weather Radar

The pertinent technical characteristics of the airborne radar are shown in Table 29. The radar is
widely utilized on large-capacity, multi-engine turboprop transport aircraft, and it may operate on
large jet aircraft in the future. It has functions for identification of active precipitation and
identification of wind shear. It has operational modes for short-range, medium-range, and long-
range weather surveillance.

Table 29. Characteristics of the Airborne Weather Radar

Radar Characteristic Value
Radar Modes Weather surveillance and wind shear alert
Frequency Agility Range 9310-9410 MHz, spaced every 5 MHz
Frequency Used for Testing 9335 MHz
Transmitter Type Solid state
Transmitter Peak Power 10 kW (?)
Pulse Modulations Unmodulated pulses (short range)
5-Chip Barker Coded (medium range)
13-Chip Barker Coded (maximum range)
Minimum range Maximum range
Pulse Repetition Rates 2000 pps 230 pps
Pulse Width Range 0.19 µs 234 µs
IF Bandwidth Range 5 MHz 52 kHz
Antenna Type Gimbaled, flat-plate slotted array
Antenna Gain +32 dBi
Antenna Sector Scan Angular Range ±15° (minimum), ±135° (maximum)
Antenna Sector Scan Rate 60°/sec
Antenna Beamwidth 2.7° horizontal, 4° vertical
Radar Receiver Noise Figure 5 dB
Radar Output Color-coded display, with colors based on
the strength of echo returns

125
The radar normally operates with automatic frequency agility, hopping among twenty
frequencies across 100 MHz of spectrum between 9310-9410 MHz. Available frequencies are
spaced 5 MHz apart. During frequency-agility operations, the radar changes frequency from
pulse to pulse. Frequencies are selected pseudo-randomly, with a uniform distribution. The
frequency-hopping design is intended to minimize frequency-dependent effects in weather echo
returns, as well as to enhance operational effectiveness in multi-radar environments. For the tests,
the radar was operated in a single-frequency mode.

The radar transmitter generates RF pulses from a set of solid-state modules. The pulses are
unmodulated in short-range operations, 5-chip Barker coded when the radar is operating at
medium range, and 13-chip Barker coded when the radar is operating at long range.

The radar antenna is a flat-plate, slotted array that sector-scans on a mechanically gimbaled
mount inside a radome on the nose of the aircraft. The elevation angle of the scanning (relative to
horizontal) is manually controlled from the cockpit during flight.

The radar receiver is a conventional triple-stage heterodyne design in which the RF front-end
LNA noise figure determines the noise figure of the entire receiver. The overall receiver noise
figure is 5 dB. RF energy from the LNA is downconverted to the input frequency of the first IF
stage, at 870 MHz. That signal is downconverted to the frequency of the second IF stage, at 90
MHz. The 90-MHz signals are processed through a quadrature modulator stage that produces I-Q
outputs. Final downconversion brings the echo signals to baseband video between 0-50 MHz,
where they undergo narrowband filtering to match operational pulse widths. Baseband video is
detected and sampled with either an 8- or 11-bit A-D converter, depending upon the selected
radar range scale. This radar utilizes a CFAR processor to reduce the effects of noise and clutter.
Final IF bandwidths are adjusted automatically to match the radar pulse width (or sub-pulse
width, in Barker-coded pulse modes). The portion of the radar IF that is significant for this report
is shown schematically in Figure 69.

Figure 69. Block diagram of part of the airborne weather radar IF stage.

The radar video output is digitally processed into a color-coded, high resolution cockpit display.
Colors correspond to the intensity of precipitation and wind shear ahead of the aircraft. An
example of a weather surveillance display when the radar was operated in a receive-only mode,
in which it was processing against only its internal noise, is shown in Figure 70. In this figure, as
in the interference tests, the radar was operating in its 40 nm maximum range weather
surveillance mode; the range rings are 10 nm apart.

126
The weather radar used in this study represents the current state-of-the art for this technology. Its
hardware is highly integrated and modular, and its signal-processing software for echo returns is
highly sophisticated. Its capabilities are believed to represent the best hardware and signal
processing that are currently available in airborne platforms performing weather surveillance and
wind shear detection missions.

Figure 70. Example of the airborne weather radar display in the weather surveillance mode.
There is no interference and the radar is running as receive-only. Green and yellow colors
indicate that the receiver is processing only its internal noise.

7.3 Interference Measurement Protocol for the Airborne Weather Radar

The goal of the interference tests was to observe the effects of interference on a radar operational
frequency as a function of both the type of interference and the interference-to-noise (I/N) ratio
of the interference in the radar IF stage. Characteristics of the interference waveforms are listed
in Table 30.

127
Table 30. Characteristics of Interference Waveforms for Airborne Weather Radar Tests
Sub-pulse Pulse Bandwidth
Interference Pulse or scaled repetition or chirp
waveform width width rate width Duty cycle
(µs) (µs) (pps) (MHz) (%)
Chirped 1 10 n/a 750 10 0.8
Chirped 2 10 n/a 750 50 0.8
Chirped 3 13.6/1.65 n/a 750 660/80 0.8
Phase coded 1 0.64 0.049 1600 1.56 0.1
Phase coded 2 20 1.54 1600 0.05 3.2
Unmod pulse 1 1 n/a 8000 1 0.8
Unmod pulse 2 1 n/a 19000 1 1.9
Unmod pulse 3 1 n/a 35000 1 3.5
BPSK 1 10 2 2000 400/80 0.4
BPSK 2 80 16 4500 400/80 7.2
BPSK 3 10 17.7 515 45/80 0.91
BPSK 4 10 1.7 5150 460/80 0.88
*
OFDM Continuous n/a n/a 20 100*
*
The OFDM interference was continuous within the 45-ms beam-scanning interval of the radar antenna.

Since most airborne weather radars operate on single frequencies, the primary goal of the tests on
this radar was to observe the effects of interference when it was likewise operated on a single
frequency. Some additional interference tests were also performed when the radar was operating
in its wideband, frequency-agility mode. For both of the radar modes (single-frequency and
frequency agility), the protocol for injecting interference and observing its effects was the same.
The interference injection arrangement is shown as a block diagram in Figure 71.

The entire radar transmitter/receiver system (including the antenna and all of the radar hardware
except for the cockpit monitor display) was mounted at the top of a 20-meter high tower, the
tower itself being located on the summit of the highest hill in the area. The radar transmitter was
turned off, so that the radar operated in a receive-only mode. The maximum range was set to 40
nmi to eliminate the effect of STC. Processed data were displayed on a cockpit monitor. When no
interference was injected, the radar showed only the effects of its internal noise on the cockpit
display, as shown in Figure 70.

Interference waveforms were generated with a vector signal generator (VSG)27 that had been
programmed for the waveform characteristics of Table 30. The RF output of the VSG was routed
via a hardline from the radar control room to a directional coupler at the radar antenna. Thus the
interference energy was injected into the radar ahead of the first RF front-end LNA. The
interference energy passed through the radar receiver RF front end, was mixed down and passed
through the multi-stage IF (as described above), was quadrature-demodulated to generate I-Q
outputs, and then passed through a digital signal processor. The processor output was routed to
the cockpit display in the radar control room.

27
A programmable Agilent VSG, model 8267C.

128
Figure 71. Block diagram of the airborne weather radar interference test setup.

Simultaneously with the observation of interference effects on the cockpit display, the (I+N)/N
level was monitored on a spectrum analyzer connected to the first stage of the IF (tapped
internally as depicted in Figure 1). Thus the effect of interference could be correlated directly
with its measured (I+N)/N level.28 An example of the spectrum analyzer data derived from the
tap point in the radar IF stage is shown in Figure 72. The interference was gated on for 45-ms
intervals, equal to the time required for the radar antenna beam to scan a single azimuth. This
made the interference interval equal to the interval it would occupy if it were emitted by a point
source in the antenna’s far field.

During normal operations, the radar receiver gain automatically decreased from one complete
sector scan (that is, a left-right plus a right-left antenna scan) to the next. Each change in receiver
gain caused the I/N level to change by an equal number of decibels. The decrease from scan to
scan was from one to two decibels. Thus, if the initial I/N level for an interference waveform
was, e.g., +20 dB, that level would normally decrease from scan to scan to levels that would
successively look like, e.g., +19 dB, +18 dB, +16 dB, +15 dB, +13 dB, etc. After a succession of
about 12 of these decreases, (i.e., after about 12 complete sector scans), the radar automatically
performed a complete recalibration of the receiver, the gain level reverted to its original, highest
value, and the entire gain sequence was repeated.

28
Subsequent to the tests, the measured (I+N)/N values were converted to the I/N values that are
reported here.

129
Figure 72. Interference burst as observed in the airborne radar IF stage. The trace is peak-
detected, so the RMS level of the radar noise is 10 dB lower (relative to the amplitude of the
interference burst) than shown here.

The test personnel used this built-in, gain-varying behavior of the radar to their advantage by
generating a high I/N level from the VSG at the beginning of each radar gain sequence and then
watching the decreasing effect of the interference on the cockpit display as the internal radar gain
was automatically decreased from scan to scan. When no interference effects were observed on
that display, the corresponding I/N level was noted. The VSG power output level was kept
constant through each of the gain-variation sequences.

Because this type of radar does not process discrete targets, the criterion of probability of
detection of targets is not applicable. Instead, its display was assessed qualitatively as a function
of interference. The qualitative levels that were used were: strongly visible; marginally visible;
barely visible; and no visible effect. Figures 73-75 show examples of the radar display when an
interference effect was strong, marginal, and barely visible, respectively.

130
Figure 73. Example of a strongly visible interference strobe (caused by phase-coded waveform
2 at I/N = +60 dB).

Figure 74. Example of a marginal interference strobe (caused by unmodulated pulse

waveform 1 at I/N = +24 dB).

131
Figure 75. Example of barely visible interference (caused by unmodulated pulse waveform 1 at
I/N = +20 dB.

7.4 Results of Interference Measurements on the Airborne Weather Radar

Results for the airborne weather radar in the single-frequency mode are summarized in Table 31.
The results show that the chirp-pulsed waveforms, phase coded waveform 1, BPSK 1, BPSK 3,
and BPSK 4 did not cause strobes on the radar’s display at I/N ratios ranging from +43 to +63
dB. (Those power levels were the maximum that the test equipment could generate, the available
maximum depending upon the modulation of each interference waveform.) BPSK 2 caused an
effect at an I/N of +30 dB. The unmodulated pulsed waveforms 1 and 2 caused effects at I/N
levels of +18-19 dB, and the unmodulated pulsed waveform 3 caused effects at an I/N of +25 dB.

In contrast, the airborne weather radar was affected at relatively low I/N levels by the OFDM
signal. This result was not surprising due to that signal’s high duty cycle. The OFDM signal was
20 MHz wide and continuous throughout a 45-ms dwell time. Therefore, it filled the entire IF
filter with energy the entire time the waveform was gated on. The I/N had to be below -2 dB
before the OFDM signal no longer affected the display.

132
Table 31. Results of Interference Tests on the Airborne Weather Radar*
Interference Signal Peak-to-RMS I/N29 Interference Effect On Radar
Modulation (dB)
Chirped 1 +63 No visible effect on display
Chirped 2 +52 No visible effect on display
Chirped 3 +43 No visible effect on display
Phase Coded 1 +47 No visible effect on display
Phase Coded 2 +60 Strongly visible strobes
+36 Strongly visible strobes
+30 Marginally visible strobes
+28 Barely visible strobes
+25 No visible effect on display
Unmodulated 1 +57 Visible strobes
+22 Marginally visible strobes
+19 No visible effect on display
Unmodulated 2 +57 Visible strobes
+21 Marginally visible strobes
+18 No visible effect on display
Unmodulated 3 +32 Visible strobes
+29 Marginally visible strobes
+25 No visible effect on display
BPSK 1 +44 No visible effect on display
BPSK 2 +52 Strongly visible strobes
+32 Marginally visible strobes
+30 No visible effect on display
BPSK 3 +54 No visible effect on display
BPSK 4 +43 No visible effect on display
OFDM +33 Strongly visible strobes
-2 Marginally visible strobes
-6 No visible effect on display
*
Where “no visible effect on display” is indicated, the corresponding I/N level indicated is the maximum
power level that could be achieved with the test equipment.

Other than for OFDM, the airborne weather radar was affected at some level or another by the
following interference waveforms: unmodulated pulsed 1, 2, and 3; BPSK 2; and phase coded 2.
The threshold for these waveforms was from +21 dB to +32 dB. (In contrast, the OFDM
interference threshold was from I/N = -6 dB to -2 dB.) All but one of these waveforms had duty
cycles of 1.9% or greater, the sole exception being a duty cycle of 0.8% for unmodulated pulsed
1. This observed relationship between a duty cycle of about 1% or greater and the eventual
occurrence of visible interference effects was consistent with the results of tests on other types of
radars in this study.

29
The I/N levels in this table have been corrected from measured (I+N)/N levels.

133
Figures 76-77 illustrate the contrast between the relatively robust radar performance at high I/N
levels for signals with duty cycles of a few percent or less versus the relatively low threshold for
the high duty cycle OFDM signal. Figure 76 shows the effect of OFDM interference at I/N levels
of -2 dB and +16 dB, while Figure 77 shows the effect of the Phase 2 waveform at I/N levels of
+28 dB and +45 dB. Comparison of these figures shows that the high duty cycle OFDM
interference waveform caused effects that were quantitatively similar to those of the low duty
cycle (3.2%) phase coded pulsed 2 waveform, but that there is a difference of about 30 dB
between the I/N levels where the onset of interference effects begins for these two waveforms.

Figure 76. Barely visible (left) and strongly visible (right) strobes due to high duty cycle
OFDM interference at I/N levels of -2 dB and +16 dB, respectively.

Figure 77. Barely visible (left) and strongly visible (right) strobes due to low duty cycle (3.2%),
phase-coded pulsed waveform 2 interference at I/N levels of +28 dB and +45 dB, respectively.
The effects are similar to those of Figure 76, but occur at I/N levels that are about 30 dB higher
than the OFDM interference thresholds.

When the radar was operated in its frequency agility mode, all interference was mitigated. This
was due to the radar design, in which the radar ceases operations on a frequency on which
interference is detected; this behavior was observed during the tests. Interference would appear
on the screen as it did for the single frequency case (e.g., the chirp-pulsed waveforms had no

134
effect at even the highest levels, while the OFDM signal had a strong effect at low levels). But as
soon as one full sector-scan was completed, the receiver would stop hopping to the frequency (or
frequencies) of the interference and the display would revert to its baseline appearance until the
frequency of the interference was changed to match the radar’s frequency.

7.5 Summary of Interference Effects for the Airborne Weather Radar

The airborne weather radar showed no effects from interference with the lowest duty cycles, on
the order of 0.1%, at even the highest I/N levels that could be achieved, on the order of +50 dB to
+60 dB. Interference waveforms with higher duty cycles, on the order of 1-3% duty cycles,
caused effects to occur at I/N levels on the order of +21 dB to +32 dB. But for a communication-
type signal (OFDM) with a nearly 100% duty cycle (within the 45 ms mainbeam dwell interval
of the radar antenna scan), the I/N levels where interference effects were manifested for OFDM
signals were substantially lower, between -6 dB to -2 dB.

These results are consistent with the results from other radars in this study. In general, radar
receivers seem to perform robustly in the presence of signals such as those emitted by other
radars (with duty cycles less than 1-3%), but are not robust in the presence of interference with
high duty cycles (near 100%) such as are characteristic of communication signals.

135
 8 SUMMARY OF INTERFERENCE EFFECTS ON RADARS

This report describes the results of wide-ranging interference measurements that have been
performed on many representative radar receivers. The receiver types included long range air
search radionavigation radars, a short range air search radionavigation radar, a fixed ground-
based weather radar, an airborne weather radar, and several maritime radionavigation radars.
Missions performed by these radars include en-route (long range) air traffic control and air
defense; airport traffic control; weather surveillance and warning; and maritime navigation and
surface search. The radars that have been tested operate in several microwave spectrum bands:
1200-1400 MHz; 2700-2900 MHz; 2900-3100 MHz; and 8500-10500 MHz. Five major results
have been observed and documented; they are summarized here.

8.1 Radars are Vulnerable to the Effects of Communication Signal Interference

The results of interference tests with communication signal modulations are summarized in
Table 32. Interference with communication-type modulations caused degradation in radar
performance at I/N levels between -9 dB to -2 dB, well below the noise floor of each radar
receiver. One radar lost targets at an I/N level of -10 dB, and I/N = -6 dB caused most of the
radars to lose targets. Future improvements to meteorological radars are predicted to render them
vulnerable at I/N levels as low as -14 dB.

Table 32. I/N Levels of Communication Signal Modulations at which Performance Decreased
for All Radars Tested
Radar Tested I/N Threshold for
Decreased Pd
Long Range Air Search Radiolocation Radar 1 (installation 1) -9 dB
Long Range Air Search Radiolocation Radar 1 (installation 2) -9 dB
Long Range Air Search Radiolocation Radar 2 -6 dB
Short Range Air Search Radionavigation Radar -9 dB
Fixed Ground-Based Meteorological Radar -9 dB*
Maritime Radionavigation Radar A -7 dB
Maritime Radionavigation Radar B -10 dB
Maritime Radionavigation Radar C -6 dB to -9 dB
Maritime Radionavigation Radar D -9 dB
Maritime Radionavigation Radar E -6 dB
Maritime Radionavigation Radar F -6 dB
Airborne Meteorological Radar Between -6 to -2 dB
*
-14 dB is the predicted threshold for the upgraded version of the fixed meteorological radar.

136
 8.2 Radars Perform Robustly in the Presence of Interference from Other Radars

In contrast to the effects of interference from communication signal modulations, pulsed

interference at low duty cycles (less than about 1-3%) was tolerated at I/N ratios as high as
+30 dB to +63 dB. The radar receivers that were tested performed robustly in the presence of
signals transmitted by other radars.

8.3 Low-Level Interference Effects in Radar Receivers are Insidious

The loss of targets at low interference levels is insidious; there is no overt indication to radar
operators or even to sophisticated radar software that losses are occurring. No dramatic
indications such as flashing strobes on radar PPI screens are observed. This insidiousness can
make low-level interference more dangerous than higher levels that will generate strobes and
other obvious warning indications for operators or processing software.

Even when radars experience serious performance degradation due to low-level interference, it is
very unlikely that such interference will be identifiable as such. It is therefore unlikely that such
interference will ever result in reports to spectrum management authorities even when it causes
loss of desired targets. Since low-level interference is not expected to be identified or to generate
reports when it occurs, lack of such reports cannot be taken to mean that such interference does
not occur.

8.4 Low-Level Interference Can Cause Loss of Radar Targets at Any Range

Interference can (and will) cause loss of targets at any distance from any radar station; loss of
targets due to radio interference is not directly related to distance of targets from radar stations.
When radar performance is reduced by some number of decibels, X, then all targets that were
within X decibels of disappearing from coverage will be lost. Range from the radar is not a factor
in this equation. Interference can cause loss of desired, large cross section targets (such as
commercial airliners, oil tankers, and cargo ships) at long distances, as well as small cross section
targets at close distances. Low-observable targets that could be lost include, for example, light
aircraft; business jets; incoming missiles; missile warheads; floating debris including partially
submerged (and extremely dangerous) shipping containers; life boats; kayaks, canoes, dinghies;
periscopes; and swimmers in life jackets.

Because any radar target can potentially be lost at any distance in the presence of radio
interference, radio interference does not translate directly into an equivalent radar range
reduction. “Range reduction” should therefore not be used as a metric in discussions of adverse
effects of interference on radars, unless such range reduction is qualified with a reference to a
(unrealistic) condition of fixed, constant cross sections for all desired targets.

137
 8.5 Radar Interference Waveforms and Test Reporting Should be Standardized

During radar interference measurements, the interference signals should be standardized to

include: broadband noise; CW; impulse UWB (dithered and non-dithered); direct sequence
UWB; multiband OFDM UWB; CDMA; TDMA; BPSK; QPSK; QAM; GMSK; OFDM; and
pulsed signals (both linear FM-modulated (chirped) and unmodulated).

Plots of Pd versus I/N are a useful and effective method of displaying interference data and
should be adopted as a standard for future radar interference measurements.

For pulsed signals, the parameters of pulse width and prf which comprise the duty cycle should
be varied to determine the point at which interference effects transition from pulse-like to noise-
like or CW-like. For linear-FM pulsed interference signals, the chirp rate and chirp bandwidth
should be varied to observe the onset of the same effects.

138
 9 REFERENCES

[1] F.H. Sanders and V.S. Lawrence, “Broadband spectrum survey at Denver, Colorado,”
NTIA Report 95-321, Sep. 1995.

[2] F.H. Sanders, B.J. Ramsey, and V.S. Lawrence, “Broadband spectrum survey at San
Diego, California,” NTIA Report 97-334, Dec. 1996.

[3] F.H. Sanders, B.J. Ramsey, and V.S. Lawrence, “Broadband spectrum survey at Los
Angeles, California,” NTIA Report 97-336, May 1997.

[4] F.H. Sanders, B.J. Ramsey, and V.S. Lawrence, “Broadband spectrum survey at San
Francisco, California,” NTIA Report 99-367, Jul. 1999.

[5] C.A. Filippi, R.L. Hinkle, K.B. Nebbia, B.J. Ramsey, and F.H. Sanders, “Accommodation
of broadcast satellite (sound) and mobile satellite services in the 2300-2450 MHz band,”
NTIA Technical Memorandum TM-92-154, Jan. 1992.

[6] P.E. Gawthrop, F.H. Sanders, K.B. Nebbia, and J.J. Sell, “Radio spectrum measurements
of individual microwave ovens, volume 1,” NTIA Report 94-303-1, Mar. 1994.

[7] P.E. Gawthrop, F.H. Sanders, K.B. Nebbia, and J.J. Sell, “Radio spectrum measurements
of individual microwave ovens, volume 2,” NTIA Report 94-303-2, Mar. 1994.

[8] M.G. Biggs, F.H. Sanders, and B.J. Ramsey, “Measurement to characterize aggregate
signal emissions in the 2400-2500 MHz frequency range,” NTIA Report 95-323, Aug.
1995.

[9] M.I. Skolnik, Introduction to Radar Systems, 2nd ed., New York, NY: McGraw-Hill,
1980.

[10] M.I. Skolnik, ed., Radar Handbook, 2nd ed., New York, NY: McGraw-Hill, 1990.

[11] D.K. Barton, Modern Radar System Analysis, Norwood, MA: Artech House, 1988.

[12] P.A. Lynn, Radar Systems, New York, NY: Van Nostrand Reinhold, 1987.

[13] L.V. Blake, “Recent advancements in basic radar range calculation technique,” pp. 49-59,
in Radars Vol. 2: The Radar Equation, D.K. Barton, ed., Norwood, MA: Artech House,
1974.

[14] L.V. Blake, Summary of “A guide to basic pulse-radar maximum-range calculation,” pp.
151-154, in Radars Vol. 2: The Radar Equation, D.K. Barton, ed., Norwood, MA: Artech
House, 1974.

[15] D. Meyer and H. Mayer, Radar Target Detection: Handbook of Theory and Practice,
New York: Academic Press, 1973.

139
[16] A. Paul, G. Hurt, T. Sullivan, G. Patrick, R. Sole, L. Brunson, C-W. Wang, B. Joiner, and
E. Drocella, with Contributors S. Williams and G. Saam, “Interference protection criteria,
Phase 1: Compilation from existing sources,” NTIA Report 05-432, Oct. 2005.

[17] M.I. Skolnik, Editor-in-chief, Radar Handbook, 2nd ed., New York: McGraw-Hill, 1990.

[18] ITU-R Recommendation M.1372-1, “Efficient use of the radio spectrum by radar stations
in the radiodetermination service,” International Telecommunication Union,
Radiocommunication Sector, 1998.

[19] S. Kingsley and S. Quegan, Understanding Radar Systems, London: McGraw-Hill, 1992.

[20] ITU-R Recommendation M.1464, “Characteristics of and protection criteria for

radionavigation and meteorological radars operating in the frequency band 2700-
2900 MHz,” International Telecommunication Union, Radiocommunication Sector, 1998.

[21] R. Doviak, D. Zrnic, and D. Surmans, “Doppler Weather Radar,” Proceedings of the
IEEE, vol. 67, no. 11, Nov. 1979.

[22] ITU-R Recommendation M.1313, “Technical characteristics of maritime radionavigation

radars,” International Telecommunication Union, Radiocommunication Sector, 2000.

[23] L.K. Brunson, J.P. Camacho, W.M. Dolan, R.L. Hinkle, G.F. Hurt, M.J. Murray, F.A.
Najmy, P.C. Roosa, Jr., and R.L. Sole, “Assessment of compatibility between
ultrawideband devices and selected federal systems,” NTIA Special Publication SP-01-
43, Jan. 2001.

140
 APPENDIX A: EXAMPLE INTERFERENCE REJECTION (IR) RESPONSES OF A
MARITIME RADIONAVIGATION RADAR

The IR feature of radars is discussed at length in Section 1. IR is fundamentally a correlator

function that operates on either a pulse-to-pulse basis (on either the repetition interval or the
pulse width) or else on a scan-to-scan basis. For pulse correlation techniques, repetition intervals
and pulse widths of return echoes that match the transmitter’s nominal characteristics are allowed
to pass through the correlator and are displayed, whereas other received pulses are rejected. IR
only works against pulsed, asynchronous interference. It is not effective against high duty cycle
interference, such as communication signals.

This appendix shows examples of the interference rejection performance in the presence of
various types of pulsed interference inputs for maritime radionavigation Radar A. The radar
characteristics are described in detail in Section 6.

In these tests, the interference was always pulsed, with a pulse width of 10 µs. But the duty cycle
of the interference varied from one test to the next. The results shown here were obtained with
duty cycles of 0.1%, 1%, and 5%, corresponding to prfs of 100 Hz, 1 kHz, and 5 kHz. For each
duty cycle, the amplitude of the interference was gradually increased and the performance of the
radar receiver was assessed by examining the PPI display with the IR function turned on versus
turned off. “Gated” refers to the case where the interference is created on the azimuth of the
targets, for the duration required for a radar mainbeam to sweep across stationary targets.
“Ungated” means that the interference is generated for the duration of an entire 360-degree
sweep of the radar beam.

IR was found to be an effective method for mitigating the effects of pulsed interference up to a
duty cycle as high as 1-3%. The effectiveness of IR decreased for higher duty cycles. IR was
found to be totally ineffective against high duty cycle interference such as communication
signals.

141
Figure A-1. +50 dB I/N, ungated,10 µs pw, 0.1% duty cycle, IR off.

Figure A-2. +50 dB I/N, ungated, 10 µs pw, 0.1% duty cycle, IR on.

142
Figure A-3. +50 dB I/N, ungated, 10 µs pw, 1% duty cycle, IR off.

Figure A-4. +50 dB I/N, ungated, 10 µs pw, 1% duty cycle, IR on.

143
Figure A-5. +80 dB I/N, ungated, 10 µs pw, 1% duty cycle, IR off.

Figure A-6. +80 dB I/N, ungated, 10 µs pw, 1% duty cycle, IR on.

144
Figure A-7. +10 dB I/N, ungated, 10 µs pw, 5% duty cycle, IR off.

Figure A-8. +10 dB I/N, ungated, 10 µs pw, 5% duty cycle, IR on.

145
Figure A-9. +15 dB I/N, ungated, 10 µs pw, 5% duty cycle, IR off.

Figure A-10. +15 dB I/N, ungated, 10 µs pw, 5% duty cycle, IR on.

146
 APPENDIX B: SELECTED INTERFERENCE EMISSION SPECTRA

Figure B-1. 1.023 MBit/s BPSK interference signal spectrum.

Figure B-2. 10 MBit/s BPSK interference signal spectrum.

147
Figure B-3. 5 MBit/s BPSK interference signal spectrum.

Figure B-4. 0.5 MBit/s BPSK interference signal spectrum.

148
Figure B-5. 2 MBit/s QPSK interference signal spectrum.

Figure B-6. W-CDMA interference signal spectrum.

149
Figure B-7. CDMA-3X interference signal spectrum.

Figure B-8. QAM interference signal spectra for maritime radar tests.

150
Figure B-9. CDMA interference spectra for maritime radar tests.

Figure B-10. UWB interference spectra as a function of receiver bandwidth.

151
Figure B-11. Chirped-pulse interference spectrum. Chirp width was equal to about 5 MHz.

152
 APPENDIX C: CALIBRATION OF UNDESIRED SIGNALS AND EXAMPLES OF
RADAR IF SELECTIVITY CURVES

CW and interference signal calibrations on radars were performed as follows. An Agilent E-

4440A spectrum analyzer, set in zero span mode and with the RMS detection selected, was
connected to a radar channel IF output. A CW signal tuned to the radar operating frequency was
then injected into the radar RF port and a level was found that produced an [(I+N)/N)] of 3 dB;
equivalent to an I/N ratio of 0 dB. The undesired signal level was recorded, and identified to be
equivalent to an I/N ratio of 0 dB. The process was repeated for channel B (if such a channel was
present), and for each type of interference signal. Using the calibrated I/N ratio of 0 dB, the other
I/N values were determined.

In addition, a CW signal was swept in frequency and the response of the IF circuitry of the radar
receivers was measured for each channel and recorded with a spectrum analyzer. The 3-dB IF
bandwidth of the radar receiver was determined in this way. Selected measurements of radar IF
bandwidths are shown in the figures below, with names of the radars referring to the titles of the
main text sections. The actual frequency of the IF outputs is usually around 32 MHz. However,
in some cases (Figures C-1 and C-2) the frequency axes have been changed to show the
corresponding RF frequencies of the radars.

Figure C-1. Channel A IF response of Long Range L-Band Radar 1, referred back to RF
frequencies. 100 kHz resolution bandwidth, peak detected, maximum-hold.

153
Figure C-2. Channel B IF response of Long Range L-Band Radar 1, referred back to RF
frequencies. 100 kHz resolution bandwidth, peak detected, maximum-hold.

Figure C-3. IF response of one channel of Long Range Radar 2, centered at 36.25 MHz. 300
kHz resolution bandwidth, peak detected, maximum-hold.

154
Figure C-4. IF response curve of the short-range air search radar, centered near 31 MHz. 100
kHz resolution bandwidth, peak detected, maximum-hold.

Figure C-5. IF response curve of a typical maritime radionavigation radar (Radar F). The IF
frequencies are mapped to their RF equivalents. 100 kHz resolution bandwidth, peak detected,
maximum-hold.

155
 APPENDIX D: TEST RESULTS ILLUSTRATING THE EFFECTIVE DUTY CYCLE
OF FM-PULSED WAVEFORMS IN A MARINE RADAR RECEIVER

The measurements in this appendix were performed with marine radionavigation Radar F
described in the main body of the report. The chirped waveforms identified in Tables 18 and 21
were injected into the radar receiver to investigate how the signal’s duty cycle and pulse width
are altered through the receiver, changing from the RF pulse waveform that was originally
transmitted to the form that is manifested in the radar’s detector/processor.

Testing has shown that the effective duty cycle of the interfering waveform is a key element in
the ability of a radar receiver to suppress interference effects from unwanted pulsed signals. The
results of these tests show that the effective duty cycle of chirped pulsed interference is reduced
inside the receiver as compared to the duty cycle that occurs at the radar receiver RF front end.

D.1 Background

The victim receiver’s IF output response (amplitude and pulse width) to interference from
chirped pulses is a function of the rate at which the chirped frequency sweeps through the victim
radar receiver pass-band. This rate, called chirp rate (Rc), is given by: Rc = (Bc/τ), where Rc is the
sweep rate (in megahertz per microsecond), Bc is the chirp frequency range (in megahertz) and τ
is the pulse duration (in microseconds). Victim radar receivers should not respond to interference
on frequencies outside the -20 dB pass-band of their IF circuitry, assuming that the amplitude of
the interference is below the front-end overload threshold of the radar receiver RF front end.

D.2 Measurement Technique

The pulses were injected into the radar receiver at the nominal RF frequency of 9410 MHz at the
LNA input. The receiver was not connected to its antenna, to ensure that no other signals could
couple into the receiver during testing. A measurement point was identified on the IF circuit card,
a spectrum analyzer was set to zero-span mode with a resolution bandwidth commensurate with
the radar receiver, and then the analyzer was connected to the measurement point to monitor the
response of the radar to the injected waveforms30. The radar was placed in stand-by mode so that
the receiver was activated, but its transmitter was not generating pulses.

The radar receiver used a summing multistage logarithmic amplifier (Figure 3 in the main
report). The IF test point that was used is located at the output of the third amplifier. A CW
signal was swept in frequency from 9370 to 9450 MHz to determine the response of the receiver
and to measure the receiver’s IF bandwidth. The result is shown in Figure D-1. The 3-dB IF
bandwidth of the radar was set to short pulse mode 1. This mode, which used a pulse width of
200 ns for a maximum range of 3 nautical miles, was measured at 6 MHz. A spurious receiver
response was noted at 9381 MHz, 20 dB below the peak response at 9410 MHz.

30
The spectrum analyzer was also used to measure and record the time waveform envelopes of
the pulses at the fundamental RF frequency, prior to injection into the radar receiver.

156
 -30

Operational frequency of radar

-40

Power in IF (dBm)

-50

Spurious responce
-60

-70

-80
9 9 9 9 9
9.37x10 9.39x10 9.41x10 9.43x10 9.45x10

Frequency (GHz)

Figure D-1. Frequency response of the IF stage of marine Radar F to a CW input signal when
the radar was in its short-pulse mode, shown as a function of the RF frequency of the CW input.

D.3 Results

The results of the measurements are shown in Figures D-2 through D-9. The figures show time
waveforms of the pulses after they have passed through the receiver’s LNA and IF circuitry.
These are the waveforms that are sent into the receiver’s processor and detector. For these
measurements the -6 dB points are used to specify pulse widths. Figure D-2 shows that
unmodulated 1-µs RF pulses (the baseline signal) were still 1-µs wide in the radar’s IF
bandwidth. However for the chirped pulsed signals, the widths of these pulses in the IF passband
of the receiver were shorter than the ones that were transmitted at the RF level. Since the pulse
repetition interval (pri) did not change, the effective duty cycle was consequently reduced.

Table D-1 summarizes the results of the differences in pulse width between the RF transmitted
pulses and the pulses that are presented to the detector/processor of the radar receiver. The
percent difference was calculated by dividing the received pulse width by the transmitted pulse
width and then multiplying by 100.

D.4 Summary

Due to the chirping characteristics of the waveforms and the response of the radar receiver, the
effective duty cycles and pulse widths of the waveforms at the radar’s detector/processor input
are smaller than those of the transmitted waveforms. The interference susceptibility tests have
shown that a radar’s robustness against the effects of pulsed interference is closely related to the
duty cycle and pulse width of the interfering waveforms. As the effective duty cycles and pulse
widths are lowered, radar receivers are able to better mitigate the effects of pulsed interference,

157
as the CFAR circuitry and other mitigation techniques outlined in ITU-R Recommendation
M.1372 become more effective. Therefore, when assessing the compatibility of radionavigation
radars and chirped pulsed waveforms, tests and measurements along with analyses should take
such effects into account. As the data in this appendix demonstrate, the effective duty cycle and
pulse width in radar receivers should be considered along with the I/N ratio when assessing the
impact of chirped waveforms on radar receivers.

Table D-1. Characteristics of Chirp-Pulse Waveforms Injected into Marine Radar F

Waveform Transmitted Pulse Pulse width (µs) at Detector-processor
width (µs) detector-processor pulse width divided
by full pulse width
Chirp 1 10 5.1 0.51
Chirp 2 10 1.0 0.10
Chirp 3 1.65 0.20 0.12
Chirp 4 2 0.20 0.10
Chirp 5 16 1.0 0.063
Chirp 6 17.7 1.2 0.068
Chirp 7 1.7 0.20 0.118

158
 0

-6 dB Pulsewidth
1.0 µsecond Single Pulse

-25
Receiver Power (dBm)

-50

-75 Nominal Noise Power

-100
0 -6 -6 -6 -6 -6
1x10 2x10 3x10 4x10 5x10

Time (microseconds)

Figure D-2. 1-µs unmodulated pulse in marine Radar F receiver. Because the radar IF
bandwidth is greater than 1 MHz, the pulse appears well-formed in the time domain.

Pulse
Pulse width: 10µs
prf: 750 Hz
-20 Chirp:10 MHz
Chirp rate: 1 MHz/µs
Receiver Power (dBm)

-40 - 6dB Pulse width

5.1 µseconds

-60

Nominal Noise in Receiver

-80

-100
0 -5 -5 -5 -5
0.5x10 1.0x10 1.5x10 2.0x10

Time

Figure D-3. Chirped waveform 1 in marine Radar F receiver. The time waveform is a result of
the Radar IF bandwidth being less than the chirp bandwidth.

159
 Chirp pulse 2 Waveform in radar 5 MHz IF bandwidth

Single Pulse
-20
Pulse width: 10 µs
prf: 750 Hz
Chirp: 50 MHz
Chirp Rate: 5 MHz/µs

-40
-6 dB Pulsewidth
1.0 µsecond

-60

Nominal Noise Power

-80

-100
0 -6 -6 -6 -6 -6
1x10 2x10 3x10 4x10 5x10

Time in microseconds

Figure D-4. Chirped waveform 2 in marine Radar F receiver.

-6 dB pulsewidth
0.20 µsecond
Single Pulse
-20
Pulse width: 1.65 µs
prf: 500 Hz
Chirp: 80 MHz
Chirp Rate: 48.5 MHz/µs
Receiver Power (dBm)

-40

-60

Nominal Receiver Noise

-80

-100
0 -6 -6 -6 -6 -6
1x10 2x10 3x10 4x10 5x10

Time (microseconds)

Figure D-5. Chirped waveform 3 in marine Radar F receiver.

160
 0

-6 dB Pulsewidth
0.20 µsecond
Pulse width: 2 µs
-20 prf: 2000 Hz
Chirp: 80 MHz
Chirp rate: 40 MHz/µs
Receiver Power (dBm)

Pulse
-40

-60

Nominal Noise Power

-80

-100
0 -6 -6 -6 -6 -6
1x10 2x10 3x10 4x10 5x10

Time (microseconds)

Figure D-6. Chirped waveform 4 in marine Radar F receiver.

- 6dB Pulsewidth
1.0 µsecond

-20 Single Pulse

Pulse Width: 16 µs
prf: 4500 Hz
Chirp: 80 MHz
Chirp Rate: 5 MHz/µs
Receiver Power (dBm)

-40

-60

-80 Nominal Noise Power

-100
0 -5 -5 -5
1x10 2x10 3x10

Time

Figure D-7. Chirped waveform 5 in marine Radar F receiver.

161
 0

-6 dB Pulsewidth
1.2 µsecond

-20
Pulse Width: 17.7 µs
pri: 515 Hz
Chirp: 80 MHz
Chirp rate: 4.6 MHz/µs
Receiver Power (dBm)

Single Pulse
-40

-60

Nominal Noise Power

-80

-100
0 -5 -5 -5
1x10 2x10 3x10

Time

Figure D-8. Chirped waveform 6 in marine Radar F receiver.

0

-6 dB Pulsewidth
0.20 µsecond

-20
Pulse Width: 1.7 µs
prf: 515 Hz
Chirp: 80 MHz
Chirp rate: 46 MHz/µs
Receiver Power (dBm)

-40
Single Pulse

-60

Nominal Noise Power

-80

-100
0 -6 -6 -6 -6 -6
1x10 2x10 3x10 4x10 5x10

Time (microseconds)

Figure D-9. Chirped waveform 7 in marine Radar F receiver.

162
FORM NTIA-29 U.S. DEPARTMENT OF COMMERCE
(4-80) NAT’L. TELECOMMUNICATIONS AND INFORMATION ADMINISTRATION

BIBLIOGRAPHIC DATA SHEET

1. PUBLICATION NO. 2. Government Accession No. 3. Recipient’s Accession No.

TR-06-444
4. TITLE AND SUBTITLE 5. Publication Date
September 2006
Effects of RF Interference on Radar Receivers
6. Performing Organization Code
NTIA/ITS
7. AUTHOR(S) 9. Project/Task/Work Unit No.

Frank H. Sanders, Robert L. Sole, Brent L. Bedford, David Franc, and Timothy
Pawlowitz
8. PERFORMING ORGANIZATION NAME AND ADDRESS
10. Contract/Grant No.
NTIA/ITS.T
U.S. Department of Commerce
325 Broadway
Boulder, CO 80305
11. Sponsoring Organization Name and Address 12. Type of Report and Period Covered

14. SUPPLEMENTARY NOTES

15. ABSTRACT (A 200-word or less factual summary of most significant information. If document includes a significant bibliography or
literature survey, mention it here.)

This report describes the results of interference tests and measurements that have been performed on
radar receivers that have various missions in several spectrum bands. Radar target losses have been
measured under controlled conditions in the presence of radio frequency (RF) interference. Radar
types that have been examined include short range and long range air traffic control; weather
surveillance; and maritime navigation and surface search. Radar receivers experience loss of desired
targets when interference from high duty cycle (more than about 1-3%) communication-type signals is
as low as -10 dB to -6 dB relative to radar receiver inherent noise levels. Conversely, radars perform
robustly in the presence of low duty cycle (less than 1-3%) signals such as those emitted by other
radars. Target losses at low levels are insidious because they do not cause overt indications such as
strobes on displays. Therefore operators are usually unaware that they are losing targets due to low-
level interference. Interference can cause the loss of targets at any range. Low interference thresholds
for communication-type signals, insidious behavior of target losses, and potential loss of targets at any
range all combine to make low-level interference to radar receivers a very serious problem.
16. Key Words (Alphabetical order, separated by semicolons)

radar interference; radar interference vulnerability; radar performance degradation; radar target loss; radar target
detection; RF interference; UWB interference effects
17. AVAILABILITY STATEMENT 18. Security Class. (This report) 20. Number of pages
162
UNLIMITED.
19. Security Class. (This page) 21. Price:
 NTIA FORMAL PUBLICATION SERIES

NTIA MONOGRAPH (MG)

A scholarly, professionally oriented publication dealing with state-of-the-art research or
an authoritative treatment of a broad area. Expected to have long-lasting value.

NTIA SPECIAL PUBLICATION (SP)

Conference proceedings, bibliographies, selected speeches, course and instructional
materials, directories, and major studies mandated by Congress.

NTIA REPORT (TR)

Important contributions to existing knowledge of less breadth than a monograph, such as
results of completed projects and major activities. Subsets of this series include:

NTIA RESTRICTED REPORT (RR)

Contributions that are limited in distribution because of national security
classification or Departmental constraints.

NTIA CONTRACTOR REPORT (CR)

Information generated under an NTIA contract or grant, written by the contractor,
and considered an important contribution to existing knowledge.

JOINT NTIA/OTHER-AGENCY REPORT (JR)

This report receives both local NTIA and other agency review. Both agencies=
logos and report series numbering appear on the cover.

NTIA SOFTWARE & DATA PRODUCTS (SD)

Software such as programs, test data, and sound/video files. This series can be used to
transfer technology to U.S. industry.

NTIA HANDBOOK (HB)

Information pertaining to technical procedures, reference and data guides, and formal
user's manuals that are expected to be pertinent for a long time.

NTIA TECHNICAL MEMORANDUM (TM)

Technical information typically of less breadth than an NTIA Report. The series includes
data, preliminary project results, and information for a specific, limited audience.

For information about NTIA publications, contact the NTIA/ITS Technical Publications Office at
325 Broadway, Boulder, CO, 80305 Tel. (303) 497-3572 or e-mail [email protected].

This report is for sale by the National Technical Information Service, 5285 Port Royal Road,
Springfield, VA 22161,Tel. (800) 553-6847.