2016 5th International Conference on Computer Science and Network Technology (ICCSNT)

Comparative Study of Spectrum Sensing Techniques

Base on Techniques Non-cooperative in Cognitive
Radio Networks

Prince SEMBA YAWADA1 and An Jian Wei2

University of Science and Technology Beijing
Department of Communication Engineering
School of Computer and Communication Engineering
Beijing, China
[email protected], [email protected]

Abstract²The growing demand of the wireless Cognitive radio by definition is therefore a system with
communications introduced the challenge of the efficient use of cognitive abilities that allow a device to acquire knowledge
the spectrum. Cognitive radio is an emerging technology in the on its environment, to have a vision of the radio available
field of wireless access, in order to considerably improve the use resources and finally to give him the opportunity to exploit
of the radio spectrum by allowing access of opportunistic these unused bands to achieve a more effective management
manner. The detection of the spectrum is the basic functionality of the radio spectrum.
in a cognitive radio that is why in this paper a particular
emphasis is placed on the comparison of the different methods The detection of free resources is one of the very crucial
of detection of the spectrum in the case of detection non- steps in cognitive radio. Several methods allow a secondary
cooperative. We note that the different techniques do not user to detect spectrum holes. The methods of the most
present the same performance. In this paper the CFD method known and used detection in the framework of the cognitive
presents a good performance by report to the MFD and ED. radio non-cooperative are: the Detection of Energy, the
The result of the simulation was obtained through the Software Cyclostationary Detection and the Matched Filter Detection
Matlab. [7]. It is important to note that these methods have their own
advantages and disadvantages. In this paper, we propose to
Keywords: Performance, Comparison, Cognitive Radio, make a performance analysis (comparison) of these different
Spectrum sensing techniques. techniques for the detection of spectrum in the framework of
detection non-cooperative.
I. INTRODUCTION
To address the problems mentioned above, we adopt the
Current wireless networks are based on policies for the following plan: in section II, we describe the different
management of static spectrum. However, the need for detection techniques of the hole of spectrum. In section III,
dynamic access solutions to spectrum (DSA) [1-3], in order we present the problem of detection. Section IV presents the
to exploit the spectrum resources available better, has clearly comparison and analysis of performance and the conclusion
been imposed as a result of the increase in bandwidth. follows in section V.
Cognitive radio is a promising technology in terms of
dynamic and opportunistic access to spectrum whose II. TECHNIQUES OF DETECTION
essential features for the proper functioning are detection,
access and sharing of the spectrum and the spectral mobility The detection of the spectrum is the basic functionality in
[4]. a cognitive radio. It is the detection of presence of primary
users in a spectrum under license; it is to detect the bands
The concept of cognitive radio was introduced by Joseph that are not used in the spectrum and to share them without
Mitola III [5], during a seminar at the Royal Institute of interference with other users. This is a fundamental problem
Technology in Sweden (KTH) in 1998. Its research had in cognitive radio. Several detection techniques have been
focused on the language of representation of the radio proposed to find unused portions of the spectrum namely: the
(RKRL) and on the way to increase flexibility at the level of Cyclostationary Detection, the Matched filter Detection and
wireless services. Cognitive radio represents a new approach the Detection of Energy. Each of these methods presents
in the field of wireless communications that Mitola described strong points and weak points.
as a radio which employs a model of reasoning to reach a
level of competence which specified in the areas related to
the radio [6].

978-1-5090-2129-1/16/$31.00 ©2016 IEEE 517 Changchun, China

A. The Matched Filter Detection (MFD) receivers can perform the timing, the estimate of channel,
In the majority of the wireless system cases, the pilot bits etc. In addition, this periodicity may be used for the detection
are transmitted on a periodic basis in order to allow the of a primary user. The basic approach is based on the density
estimation of the channels and synchronization. The bits of the function of auto-correlation and power spectrum. The
drivers are the information that can be used to detect the function of spectral correlation presents a cyclic spectrum
presence of the primary user. When knowledge of the signal with a frequency of cycle. An advantage is that the
emitted is available in the first place, the matched filter interference and the stationary noise does not present a
detection will be the technique of optimal detection, because spectral correlation. Therefore, the detector Cyclostationary
it can establish a correlation between the received signals can operate in a region of low SNR. The implementation of
with the primary signal known for the detection. The this method is very complex and it requires a priori
matched filter has the advantage of short detection time and information about the issued signal.
it works well in the regime of low SNR. But it requires
perfect knowledge on the characteristics of the signals of the
primary user, for example, the type of used modulation, the
central frequency and bandwidth etc. Any bad information
on the signal of the able will lead to a strong degradation of
detection performance. Fig.3. The synoptic diagram of Cyclostationary Detector

III. PROBLEM OF DETECTION

The problem of detection of a free band is to know if
there is a signal present or not. It boils down to the following
hypothesis test:
Fig.1. The synoptic diagram of Matched Filter Detector
(1)
B. The Energy Detection (ED)
The detection of the energy is focused on the fact that the Hypothesis 1 (H1): indicates that the signal is present
level of the energy of the signal is generally higher than the Hypothesis 0 (H0): indicates that the signal is absent
noise. To determine the existence of the primary user, the
detector of energy compares its output with a predefined where h is the gain of canal, n(t) is considered as the additive
threshold, which is based on the average of noise. If the white Gaussian noise (AWGN), x(t) is the useful signal.
output is above the threshold, then the value of the energy
takes the decision that the primary user is present; otherwise, IV. COMPARISON AND ANALYSIS OF PERFORMANCES
it takes the decision that the primary user is absent. The
The analysis of the performance of the detection of the
energy detection is the most common type of technology for
spectrum is focused on the two main following parameters,
the detection of spectrum for the following reasons: It is
namely: the probability of detection (PD) and the probability
simple to implement; it does not require a priori information
of false alarm (PFA).
concerning the primary user signal; the detection time is
relatively short [8][9]. Although a detector of energy is very The probability of false alarm is the probability that H1 is
simple to implement, it has several drawbacks; Firstly, the actually true.
spectrum sensing speed is relatively slow. Secondly, the
threshold of detection is very sensitive to the high level of The Probability of Detection is the probability that
noise and interference in the band. It is even worse in the decides that a signal is present while H1 is true.
frequency-selective and channels varying in time. Thirdly, a In order to analyze the performance of different detection
detector of energy cannot differentiate the modulated signals, techniques of the spectrum above, in data conditions, it is
noise and interference. often useful to draw the curve which binds the probability of
detection with the probability of false alarm. This curve is
called ROC, Receiver Operating Characteristic.
Figure 4 and Figure 5 show the comparison between
CFD and MFD for the different values of SNR= -6 dB and
SNR= 2 dB. With these figures, we can observe that the CFD
Fig.2. The synoptic diagram of energy detector presents the best performance that MFD. This method is
reasonable because, in the CFD detection, information on the
signal of the primary user are known a priori, and
C. The Cyclostationary Feature Detection (CFD) additionally CFD detection presents better performance even
The modulated signals are in general Cyclotationary, with low SNR.
given that the frequency integrated often occurs in the
sequence of training, of cyclic prefixes, etc. This periodicity
is introduced in the signal of the main users so that the

518
 ROC for comparison of Cyclostationary Detection and Matched Filter Detection, SNR= -6 dB ROC for comparison of ED, MF Detection and CF Detection, SNR/dB =-2.3
1 1

0.9 0.9

0.8 0.8

0.7 0.7

Probabilty of Detection
Probabilty of Detection

0.6 0.6

0.5 0.5
Matched Filter Detection
Cyclostationary Detection 0.4
0.4
Energy Detection
0.3 Matched filter Detection
0.3
Cyclostationary Detection
0.2
0.2
0.1
0.1
0
0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Probabilty of False Alarm
Probabilty of False Alarm

Fig.6. Comparison of Cyclostationary Detector, Matched Filter Detector

Fig.4. Comparison of Cyclostationary Detector and Matched Filter
and Energy Detector for SNR= -2.3 dB
Detector for SNR= -6 dB
ROC for comparison of ED, MF Detection and CF Detection, SNR =3 dB
ROC for comparison of Cyclostationary Detection, Matched Filter Detection, SNR = 2 dB 1
1
0.9

0.9
0.8

0.8 0.7

Probabilty of Detection
0.7 0.6
Probabilty of Detection

0.5
0.6
0.4
0.5
0.3 Energy Detection
Matched Filter Detection
0.4 Matched Filter Detection 0.2 Cyclostationary detection
Cyclostationary Detection
0.3 0.1

0.2 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Probabilty of False Alarm
0.1

0 Fig.7. Comparison of Cyclostationary Detector, Matched Filter Detector

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Probabilty of False Alarm and Energy Detector for SNR= 3 dB

Fig.5. Comparison of Cyclostationary Detector and Matched Filter V. CONCLUSION

Detector for SNR= 2 dB
This paper has just highlighted the comparison of
Figure 6 and Figure 7 show the comparison of the different techniques for the detection of spectrum in the case
performance of the three methods, namely: CFD, MFD and of technique for detecting non-cooperative. We note that the
ED for the different values of SNR= -2.3 dB and SNR= 3 technique CFD presents the better performance than other
dB. We note that the CFD always shows the best techniques. This is justified by the fact that this method
performance in relation to other methods. It is because this knows a priori information on the signal to transmit, and
method has a perfect knowledge of the signal and the more for a low value of SNR, just like for a high value of
channel. We are also able to see that the MFD has a better SNR the CFD has a good performance. We can also see that
performance than ED. It is because it does not have the MFD presents a better performance than ED. The simple
information of a priori on the signal to transmit, and for a fact is that the ED does not know a priori information on the
low value of SNR the performance is low. signal to transmit and it is necessary that the value of SNR is
significantly higher so that it can achieve a good
performance.
In future work, we will do a deep analysis of the impacts
of the SNR on each of the techniques of detection of the
spectrum.

519
 ACKNOWLEDGMENT
This paper is supported by the Nature Science
Foundation of China No.61572072 and the National High
Technology Program No.2015AA015702.

REFERENCES
[1] E. Trigui0(VVHJKLUHW/0HUJKHP%RXODKLD³$0RELOLW\6FKHPD
IRU&RJQLWLYH5DGLR1HWZRUNV´12th Annual Mediterranean Ad Hoc
Networking Workshop (MEDHOC-1(7¶, Ajaccio, Corsica,
France, 24 -26 juin 2013.
[2] E. Trigui, M. EssegKLU HW / 0HUJKHP %RXODKLD ³0XOWL-Agent
Systems Negotiation Approach for Handoff in Mobile Cognitive
5DGLR 1HWZRUNV´ IEEE International Conference on New
7HFKQRORJLHV0RELOLW\DQG6HFXULW\,(((1706¶Istanbul, 7- 10
Mai 2012, pp 1-5.I.S. Jacobs and &3 %HDQ ³)LQH SDUWLFOHV WKLQ
ILOPV DQG H[FKDQJH DQLVRWURS\´ LQ 0DJQHWLVP YRO ,,, *7 5DGR
and H. Suhl, Eds. New York: Academic, 1963, pp. 271-350.
[3] E. Trigui 0 (VVHJKLU HW / 0HUJKHP %RXODKLD ³6SHFWUXP $FFHVV
GXULQJ&RJQLWLYH5DGLR0RELOHV¶+DQGRII´International Conference
on Wireless and Mobile &RPPXQLFDWLRQV,&:0&¶Luxembourg,
19 - 24 Juin 2011, pp 221-224.
[4] Ziafat S, Ejaz W, Jamal H. Spectrum sensing techniques for cognitive
radio networks: Performance analysis[C]// Intelligent Radio for
Future Personal Terminals (IMWS-IRFPT), 2011 IEEE MTT-S
International Microwave Workshop Series onIEEE, 2011:1-4.
[5] - 0LWROD ,,, ³&RJQLWLYH UDGLR IRU IOH[LEOH PRELOH PXOWLPHGLD
FRPPXQLFDWLRQ´ IEEE International Workshop on Mobile
Multimedia Communications (MoMuC), Novembre 1999, pp. 3±10.
[6] - 0LWROD ³Cognitive radio: An integrated agent architecture for
software defined radio´ 3K' WKHVLV ,QVWLWXW GH 7FKQRORJLH UR\DOH
(KTH), Suède, Mai 2000.
[7] Kalambe S, Lohiya P, Malathi P. Performance evolution of energy
detection spectrum sensing technique used in cognitive radio[C]//
Signal Propagation and Computer Technology (ICSPCT), 2014
International Conference onIEEE, 2014:786-790.
[8] : =KDQJ 5 0DOOLN DQG . /HWDLHI ³2SWLPL]DWLRQ RI FRRSHUDWLYH
spectrum sensing ZLWKHQHUJ\GHWHFWLRQLQFRJQLWLYHUDGLRQHWZRUNV´
IEEE Transactions on Wireless Communications,vol. 8, no. 12, pp.
5761±5766, 2009.
[9] = <H * 0HPLN DQG - *URVVSLHWVFK ³(QHUJ\ GHWHFWLRQ XVLQJ
estimated noise variance for spectrum sensing in cognitive radio
QHWZRUNV´LQProc. of IEEE WCNC. IEEE, 2008, pp. 711±716.

520