Lecture

 3
Cellular  Systems

I-Hsiang Wang
[email protected]

3/13, 2014
Cellular  Systems:  Additional  Challenges
• So far: focus on point-to-point communication
• In a cellular system (network), additional issues arise:
Multiple access

Inter-cell interference
management

2
 Issues  Less  Emphaized  in  the  Lecture
• Handoff (focus of the network layer)

• Duplexing between uplink and downlink:

- Frequency Division Duplex (FDD)
- Time Division Duplex (TDD)

• Sectorization

• Focus mainly on licensed cellular systems

- WiFi, various wireless personal communication systems, are not
discussed here

3
 Some  History
• Cellular concept (Bell Labs, early 70’s)

• AMPS (analog, early 80’s)

• GSM (digital, narrowband, late 80’s)

• IS-95 (digital, wideband, early 90’s)

• 3G/4G systems

4
 Plot
• Three cellular system designs as case studies to
illustrate approaches to multiple access and (inter-cell)
interference management

• Both uplink and downlink will be mentioned

Downlink Uplink

5
 Outline
• Narrowband (GSM)

• Wideband system: CDMA (IS-95, CDMA 2000, WCDMA)

• Wideband system: OFDMA (Flash OFDM, LTE)

6
Narrowband  Systems
 Basic  Ideas
• Total bandwidth divided into narrowband sub-channels
- GSM: 25 MHz → 200 kHz × 125 sub-channels
- Uplink (890 – 915 MHz) and Downlink (935 – 960 MHz): the same

• Time Division Multiple Access (TDMA)

- Users share time slots in a sub-channel; each user per time slot
- Multiple access is orthogonal: intra-cell users never interfere with
each other

• Partial Frequency Reuse

- Neighboring cells uses disjoint sets of sub-channels
- Careful frequency planning → essential no inter-cell interference

8
 Time  Division  Multiple  Access
Point-to-point
GSM: 8 users communication
share a 200 kHz sub-channel, time slot: 577 μs

125 sub-channels

25 MHz
200 kHz

577 μs

TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7

8 users per sub-channel

Figure 3.9 The 25-MHz band of a GSM system is divided into 200-kHz sub-channels, which are
further divided into time slots for eight different users.

Since one time slot occurs every 4.615 ms for each user, this translates
into a delay of roughly 40 ms, a delay9 judged tolerable for voice. The eight
 Partial  Frequency  Reuse
124 Cellular systems

• Neighboring cells uses

Figure 4.2 A hexagonal
disjoint sets ofarrangements
sub-channelsof cells and a 7
possible reuse pattern of
• Each cell gets only 1/7 of the
channels 1 through 7 with the
condition that a channel
1
3
6
1

total bandwidth cannot be used in one 6 4 5

concentric ring of cells around 5 2 7
the cell using it. The frequency 2 7 3 1
• Frequency reuse factor = 1/7
reuse factor is 1/7.
3 1 6
1 6 4 5
4 5
• High SINR, but price to pay: 5 2 7
2
3
- Reducing the available 7 3 1
degrees of freedom 1 6 4

- Higher complexity in
6 4 5

network planning in real

world cells, a channel is allocated to a cell only i
centric rings of neighboring cells. Assumin
10 arrangement, Figure 4.2 depicts cells that can
Time-­‐Frequency  Resource  Allocation
Frequency

user  index   1 2 3 4 5 6 7 8 cell 4

within  a  cell
cell 3
9 10 11 12 13 14 15 16

cell 2

cell 1

Time

11
 Time  and  Frequency  Diversity
• Time diversity: Coding + Interleaving

• Frequency diversity
- Within a narrowband sub-channel: flat fading ⟹ no diversity
- Obtained via frequency hopping
Frequency

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

9 10 11 12 13 14 15 16 1 2 3 4 5 6 7 8

Time

12
Why  Full  Frequency  Reuse  won’t  Work
2
|h| P
• Signal-to-Interference-plus-Noise Ratio SINR =
N0 + I
• Limiting factor: interference power I
- I is due to the single interferer from the neighbor cell
- I is random since the location of the single interferer is uncertain
- Variance of I is quite large and I can be comparable with |h|2P
- Like deep fade, but can’t be handled by current diversity schemes
• Interference averaging is desired:
- If interference come from multiple interferers with smaller power,
then a similar effect in diversity schemes will emerge due to LLN!
XN N
X
becomes
I ! Ik , E [I] = E [Ik ]
k=1 k=1

13
 Summary
• Orthogonal narrowband channels are assigned to users
within a cell

• Users in adjacent cells can’t be assigned the same

channel due to lack of interference averaging across
users ⟹ reduces the frequency reuse factor and leads
to inefficient use of the total bandwidth

• The network is decomposed into a set of high SINR

point-to-point links, simplifying the physical-layer design

• Frequency planning is complex, particularly when new

cells have to be added
14
Wideband  System:  CDMA
 Features  of  CDMA
• Universal frequency reuse:
- All users in all cells share the same bandwidth
• Main advantages:
- Maximizes the degrees of freedom usage
- Allows interference averaging across many users
- Soft capacity limit (i.e., no hard limit on the # of users supported)
- Allows soft handoff
- Simplify frequency planning

• Challenges
- Very tight power control to solve the near-far problem
- More sophisticated coding/signal processing to extract the
information of each user in a very low SINR environment

16
 Design  Goals
• Make the interference look as much like a white
Gaussian noise as possible:
- Spread each user’s signal using a pseudonoise sequence
- Tight power control for managing interference within the cell
- Averaging interference from outside the cell as well as fluctuating
voice activities of users

• Apply point-to-point design for each link

- Extract all possible diversity in the channel

17
 Point-­‐to-­‐Point  Link  Design
• Extracting maximal diversity is the name of the game
- Because each user has an equivalent point-to-point link!

• Time diversity is obtained by interleaving across different

coherence time periods and (convolutional/turbo) coding

• Frequency diversity is obtained by the Rake receiver –

combining of the multipaths

• Transmit diversity is supported in 3G CDMA systems

18
 In a CDMA system, users interact through the interference they cause eac
rest are outside the cell. The data of the kth user are encoded into two BPSK
other. We discuss ways to manage that interference and analyze its effect o
CDMA  Uplink
4 I Q
sequences !ak "m#$ and !ak "m#$, which we assume to have equal amplitude
performance. For concreteness, we first focus on the uplink and then mov
for all m. Each sequence is modulated by a pseudonoise sequence, so that the
on to the downlink. Even though there are many similarities in their design
transmitted complex
there aresequence is
several differences worth pointing out.

Cellular systems xk "m# = aIk "m#skI "m# + jaQk "m#skQ "m#% m = 1% 2% & & & % (4.1)
4.3.1 CDMA uplink
user 1 Tx where !skI "m#$ and !skQ "m#$ are pseudonoise sequences taking values ±1.
The general schematic of the uplink of a CDMA system with K users in th
4.3 Schematic of the I Recall that m is called a chip time. Typically, the chip rate is much larger than
{a1[m]} system is shown in Figure 4.3. A fraction of the K users are in the cell and th
uplink. × 5
rest are outside the
user
the data rate. Consequently, information
cell.
1 data
The
Ch.
bits are heavily coded and the coded
of the kth user are encoded into two BPSK
I I Q
{s1[m]} sequences !asequences
k "m#$+ and4 !a "M#$and
!aIkk"m#$ have (1)Qa"m#$,
lot ofwhich
redundancy.
we assumeTheto transmitted
have equal amplitud
h!a k
Q sequence of user k goes
for all through
m. Each a discrete-time
sequence is modulated baseband equivalentsequence,
by{w[m]}
a pseudonoise multipathso that th
{a1[m]}
channel h×'k( and is superimposed
transmitted at the receiver:
complex sequence is
Q
{s1[m]} " #
! K !aI "m#s Q Q
x "m# = 'k( I "m# + jak "m#sk "m#% Σ m= BS1% 2%Rx
&(4.2)
&&% (4.1
I y"m# = k hℓ "m#xk "m − ℓ# + w"m#*
k k
{aK[m]} k=1 ℓ
×
where and !skQ "m#$ are pseudonoise sequences taking values ±1
!skI "m#$
I
{sK[m]} Recall+ that
'k( is called a chip
to beTypically, the chip rate is muchinlarger tha
The fading channels !hm $ are assumed h(K ) time. independent across users,
5
Q
{aK[m]} addition to the
the data rate.
assumption Consequently,
of independence information
across taps bitsmade
are heavily coded
in Section and the code
3.4.3.
× Q
Q
{sK[m]}
sequences !a I
k "m#$ and
user !a K Ch.
k "M#$ have a lot of redundancy. The transmitte
sequence of user k goes through a discrete-time baseband equivalent multipat
4
Since CDMA channel h'k( and
systems operate is superimposed
at very at the ofreceiver:
low SINR per degree freedom, a binary modulation
user K Tx5 Inalphabet is always used. " #
The receiver forIS-95,
user the
k chip rate is 1.2288 MHz and the data!rate'k(is 9.6ofkbits/s
the oroutput
less.
K
multiplies the I and Q ! components
y"m# = I hℓ "m#x Q k "m − ℓ# + w"m#* (4.2
sequence !y"m#$ by the pseudonoise sequences !sk=1 k "m#$ ℓ and !sk "m#$ respec-
tively to extract the coded streams of user k, which are then fed into a
The fading channels
demodulator to recover the information bits. Note $that !h 'k(
areinassumed
practice,tothe be users’
independent across users, i
addition
signals arrive asynchronously to the
at the assumption
transmitter butofweindependence
are makingacross taps made in Section 3.4.3
the ide-
alistic assumption that users are chip-synchronous,
19 so that the discrete-time
 Statistics  of  Interference  (1/2)
• Pseudorandom sequence properties:
- Different users use different random shift of a sequence
generated by maximum length shift register (MLSR):
⇥ ⇤T
s[0] s[1] ··· s[G 1]

- I and Q channels of the same user can use the same sequence
(
G
X1
- Near-orthogonal property:
m=0
s[m]s[m + l] =
G,
1,
l=0
l 6= 0
XX
• Effective interference for user 1: I[m] :=
k>1 l
(k)
hl xk [m l]

- Circular symmetric because each hl(k) is

• Second-order statistics: approximately white

( P c ⇥ ⇤X h i
⇤ = E
k>1 k , l=0 Ekc := E |xk [m]| 2 (k)
E |hl [m]|2
E [I[m]I[m + 1] ]
⇡ 0, l 6= 0 l

20
 Statistics  of  Interference  (2/2)
• Due to central limit theorem (CLT), further approximate
the interference as a Gaussian random process

• Hence, the effective noise + interference for each user

can be viewed as an additive white Gaussian noise!

• Remark: the assumption that each interferer contributes

a roughly equal small fraction to the total interference is
valid due to tight power control in CDMA

21
 Processing  Gain
• Received energy per chip:
⇥ ⇤ X h i
(k)
Ekc := E |xk [m]| 2
E |hl [m]|2
l

• SINR per chip: small

E1c
SINR1,c := P c+ 2
E
k6=1 k

• SINR per bit: E1b

||u||2 E1c GE1c
SINR1,b := P c+ 2
=P c+ 2
E
k6=1 k E
k6=1 k
⇥ ⇤T
u= sI1 [0] sI1 [1] ··· sI1 [G 1]

• G: Processing Gain

22
 IS-­‐95  Uplink  Architecture
136 Cellular systems

Processing gain PN Code

= 1238.8/9.6 = 128 Generator
for I channel

1.2288 Mchips/s
Baseband
Forward Link Shaping
Data Filter
9.6 kbps –90˚
4.8 kbps Rate = 1/3, K = 9 64-ary Output
Block Repetition 1.2288 Mchips/s Carrier
Convolutional Orthogonal CDMA
2.4 kbps Interleaver ×4 Generator
Encoder Modulator Signal
1.2 kbps
28.8
ksym / s Baseband
Shaping
Filter
1.2288 Mchips/s

PN Code
Generator
for Q channel

Figure 4.4 The IS-95 uplink. processing gain G increase proportionally as well. This means that CDMA is
an inherently scalable multiple access scheme.9

IS-95 link design

The above scheme is based
23
on repetition coding. By using more sophisti-
 Power  Control
• Maintain equal received power for all users in the cell
• Tough problem since the dynamic range is very wide.
Users’ attenuation can differ by many 10’s of dB

• Consists of both open-loop and closed loop

- Open loop sets a reference point
- Closed loop is needed since IS-95 is FDD
• Consists of 1-bit up-down feedback at 800 Hz

• Consumes about 10% of capacity in IS-95

• Latency in access due to slow powering up of mobiles

24
 Power  Control  Architecture
139 4.3 Wideband systems: CDMA

Initial downlink
power Transmitted Received
measurement Estimate power signal
uplink power Channel
required Measured
SINR

Outer loop
Inner loop
Open loop

±1dB Measured
Measured Update error probability Frame
SINR < or > β β > or < target decoder
rate

Closed loop

Figure 4.5 Inner and outer adjusts the SINR threshold as a function of frame error rates (Figure 4.5).
loops of power control. An important point, however, is that even though feedback occurs at a high
rate (800 Hz), because of the limited resolution of 1 bit per feedback, power
control does not track the fast multipath fading of the users when they are at
vehicular speeds. It only tracks the slower shadow fading and varying path
loss. The multipath fading is dealt with primarily by the diversity techniques
25
 Interferene  Averaging
• The received SINR for a user:
P
SINR = P
N0 + (K 1)P + i2cell
/ Ii

• In a large system, each interferer contributes a small

fraction of the total out-of-cell interference
- Made possible due to power control

• This can be viewed as providing interference diversity

• Same interference-averaging principle applies to voice

bursty activity and imperfect power control

26
Soft handoff
Soft  Handoff
Handoff from one cell to the other is an important mechanism in cellular
systems. Traditionally, handoffs are hard: users are either assigned to one

• cell or the other but not both. In CDMA systems, since all the cells share
Provides another form of diversity: macrodiversity
the same spectrum, soft handoffs are possible: multiple base-stations can
- Two basesimultaneously
stations can decodesimultaneously
the mobile’s data, with the switching center
decode choosing
the data

Switching
center

± 1 dB Power control bits ± 1 dB

Base-station 1 Base-station 2

Figure 4.6 Soft handoff. Mobile

27
 Uplink  vs.  Downlink
• Near-far problem does not exist in DL ⟹ power control
is less crucial
• Tx can make DL signals for different users orthogonal
- Still, due to multipaths, not completely orthogonal at the receiver
• Rake is highly sub-optimal in the downlink
- Equalization is beneficial as all users’ data go through the same
channel and the aggregate rate is high
• Less interference averaging in the downlink
- Interference comes from a few high-power base stations as
opposed to many low-power mobiles

28
 Issues  with  CDMA
• In-cell interference reduces capacity

• Power control is expensive, particularly for data

applications where users have low duty cycle but require
quick access to resource

• In-cell interference is not an inherent property of systems

with universal frequency reuse
⟹ We can keep users in the cell orthogonal, and still
have universal frequency reuse

29
Wideband  System:  OFDMA
 Basic  Ideas
• Lecture 2: OFDM as a point-to-point modulation scheme,
converting an ISI channel into parallel channels

• It can also be used as a multiple access technique!

- By assigning different time/frequency slots to users, they can be
kept orthogonal within a cell
- Equalization is no longer needed

• How to deal with inter-cell interference?

• ⟹ Interference averaging
• Achieved by careful design of hopping matrices (a way of
subcarrier allocation)

31
Hopping  Sequences  as  Virtual  Channels
• Basic unit of resource: a virtual channel
• – Hopping sequence over time-frequency plane

• Coding across the symbols in a hopping sequence

- If there were no coding and coding across subcarriers, the OFDM
system would behave like narrowband systems due to lack of
interference averaging!

• Hopping sequences are orthogonal within a cell

• Each user is assigned a number of virtual channels
depending on their data rate requirement

32
 Design  Principles
• Spread out the subcarriers for one user to gain
frequency diversity
• Hop the subcarrier allocation every OFDM block
Nc = 5, and 5 users Hopping Matrix
Frequency (Latin square)
2 3
0 1 2 3 4 0 1 2 3 4
2 3 4 0 1 62 3 4 0 17
6 7
4 0 1 2 3 !6
64 0 1 2 377
41 2 3 4 05
1 2 3 4 0
3 4 0 1 2
3 4 0 1 2
Each row/column is a
Time
permutation of [0:Nc–1]

33
 Hopping   S
4.4 Wideband systems: OFDM
equences
4.9 Virtual channel Virtual Channel 0 Virtual Channel 1 Virtual Channel 2
g patterns for Nc = 5.

Virtual Channel 3 Virtual Channel 4

For example, we see that the virtual channel 0 is assigned the OFDM symbo
time and sub-carrier pairs (0, 0), (1, 2), (2, 4), (3, 1), (4, 3). Now users could
34
 Hopping  Matrix  Design
• Each base station has its own hopping matrix
• Design rule: maximize the number of interferers that one
user encountered ⟹ min. overlap of hopping matrices
- Latin squares with this property are called orthogonal
Bad Choice Good Choice
Cell A Cell B Cell A Cell B
2 3 2 3 2 3 2 3
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
62 3 4 0 17 62 3 4 0 17 62 3 4 0 17 61 2 3 4 07
6 7 6 7 6 7 6 7
64 0 1 2 37 64 0 1 2 37 64 0 1 2 37 62 3 4 0 17
6 7 6 7 6 7 6 7
41 2 3 4 05 41 2 3 4 05 41 2 3 4 05 43 4 0 1 25
3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 4 0 1 2 3

user 0 in cell A always user 0 in cell A interferes with user

interferes with user 0 in cell B! 0, 3, 1, 4, 2 in cell B respectively

35
 Mutually  Orthogonal  Latin  Squares
• For a prime Nc, a simple construction of a family of Nc–1
mutually orthogonal Latin squares are as follows:
For a 2 {1, 2, . . . , Nc 1}, define an Nc ⇥ Nc matrix Ra
a
with (i, j)-th enrty Rij = ai + j mod Nc ,
where i, j 2 {0, 1, . . . Nc 1}

• It can be shown that a≠b ⟹ Ra and Rb are orthogonal

36
 Out-­‐of-­‐Cell  Interference  Averaging
• The hopping patterns of virtual channels in adjacent cells
are designed such that any pair has minimal overlap

• This ensures that a virtual channel sees interference

from many users instead of a single strong user

• This is a form of interference diversity

37
 Example:  Flash  OFDM
• Bandwidth = 1.25 Mz
• # of data sub-carriers = 113
• OFDM symbol = 128 samples = 100 μ s # #
• Cyclic prefix = 16 samples = 11 μ s delay spread

• OFDM symbol time determines accuracy requirement of

user synchronization (not chip time, better than CDMA)

• Ratio of cyclic prefix to OFDM symbol time determines

overhead (fixed, unlike power control in CDMA)

38
 States  of  Users
• Users are divided into 3 states:
- Active: users that are currently assigned virtual channels (<30)
- Hold: users that are not sending data but maintain
synchronization (<130)
- Inactive (<1000)
• Users in hold state can be moved into active state very
quickly

• Because of the orthogonality property, tight power control

is not crucial and this enables quick access for users
- Important for certain applications (requests for http transfers,
acknowledgements, etc.)

39
 Bandwidth (MHz) 1.25 2.5 5.0 10.0
and data. Efficiency and low latency are therefore paramount. As we will see in the following section, OFDMA is

OFDMA  in  LTE

superior to packet-oriented schemes in both of these critical dimensions. Subcarrier bandwidth (kHz) 15

Physical resource block (PRB) 180

2.3.2 OFDMA and the LTE Generic Frame Structure bandwidth (kHz)

OFDMA is an excellent choice of multiplexing scheme for the 3GPP LTE downlink. Although it involves added Number of available PRBs 6 12 25 50
complexity in terms of resource scheduling, it is vastly superior to packet-oriented approaches in terms of efficiency and

•
latency. In OFDMA, users are allocated a specific number of subcarriers for a predetermined amount of time. These are
In LTE, OFDMA is used in downlink The total number of available subcarriers depends on the overall transmission bandwidth of the
referred to as physical resource blocks (PRBs) in the LTE specifications. PRBs thus have both a time and frequency
specifications define parameters for system bandwidths from 1.25 MHz to 20 MHz as shown in
dimension. Allocation of PRBs is handled by a scheduling function at the 3GPP base station (eNodeB).

-
defined as consisting of 12 consecutive subcarriers for one slot (0.5 msec) in duration. A PRB is
resource allocation assigned by the base station scheduler.
Basic unit of resource is a 12 sub-carrier × 7 OFDM symbol time block
Figure 2.3.2-1 LTE Generic Frame Structure
Figure 2.3.2-2 Downlink Resource Grid

Tslot
downlink slot
1 Frame (10 msec)

1 Sub-Frame (1.0 msec) 1 Slot (0.5 msec)

0 1 2 3 10 11 19

0 1 2 3 4 5 6 0 1 2 3 4 5 6

7 OFDM Symbols
(short cyclic prefix)
cyclic prefixes Resource Block:

- Interference averaging is achieved

In order to adequately explain OFDMA within the context of the LTE, we must study the PHY layer generic frame
7 symbols X 12 subcarriers (short CP), or;

NBW subcarriers

12 subcarriers
6 symbols X 12 subcarriers (long CP)
structure. The generic frame structure is used with FDD. Alternative frame structures are defined for use with TDD.
However, TDD is beyond the scope of this paper. Alternative frame structures are therefore not considered.
by hopping over different blocks
As shown in figure 2.3.2-1, LTE frames are 10 msec in duration. They are divided into 10 subframes, each subframe
over
being 1.0 msec long. time
Each subframe is further divided into two slots, each of 0.5 msec duration. Slots consist of either 6
Resource Element

-
or 7 ODFM symbols, depending on whether the normal or extended cyclic prefix is employed.

Less averaging than symbol-by-

8 Overview of the 3GPP Long Term Evolution Physical Layer Freescale Semiconductor
symbol hopping but facilitate
channel estimation

40 Freescale Semiconductor, Inc. Overview of the 3GPP Long Term Evolution Physica
 one symbol period and is referred to as a resource element. Note that in MIMO applications, there is a re

Channel  Estimation
each transmitting antenna.

In contrast to packet-oriented networks, LTE does not employ a PHY preamble to facilitate carrier offset
channel estimation, timing synchronization etc. Instead, special reference signals are embedded in the P
in Figure 2.3.2-3. Reference signals are transmitted during the first and fifth OFDM symbols of each slot

• Channel estimation is achieved by interpolating between

CP is used and during the first and fourth OFDM symbols when the long CP is used.
Figure 2.3.2-3 LTE Reference Signals are Interspersed Among Resource Elements
the pilots
Subframe

Slot Slot

R R
12 Subcarriers

R R

R R

R R

Note that reference symbols are transmitted every sixth subcarrier. Further, reference symbols are stagg
time and frequency. The channel response on subcarriers bearing the reference symbols can be comput
Interpolation is used to estimate the channel response on the remaining subcarriers.
41
 Peak-­‐to-­‐Average  Power  Ratio
• OFDM transmitted signal has a high PAPR due to
superposition of many independent sub-carrier symbols

• This leads to significant backoff in the power amplifier

setting and low efficiency
• Particularly significant issue in the uplink

• Several engineering solutions to this problem

• Current version of LTE uplink uses OFDM for multiple
access but single carrier transmission per user.

42
LTE uplink requirements differ from downlink requirements in several ways. Not surprisingly, power consumption is a

LTE  Uplink:  SC-­‐FDMA

key consideration for UE terminals. The high PAPR and related loss of efficiency associated with OFDM signaling are
major concerns. As a result, an alternative to OFDM was sought for use in the LTE uplink.
Single Carrier – Frequency Domain Multiple Access (SC-FDMA) is well suited to the LTE uplink requirements. The
basic transmitter and receiver architecture is very similar (nearly identical) to OFDMA, and it offers the same degree of
multipath protection. Importantly, because the underlying waveform is essentially single-carrier, the PAPR is lower.
Fig. 2.5-1 SC-FDMA and OFDMA Signal Chains Have a High Degree of Functional Commonality

Cyclic
Single
Symbol N-Point Prefix
Bit Carrier S/P M-Point Subcarrier
IDFT & RFE
Stream Constellation Convert DFT Mapping
Block Pulse
Mapping
Shaping
Channel

Freq Cyclic
Bit Const. SC P/S Symbol M-Point N-Point
Detector Convert Domain DFT Prefix RFE
Stream De-map IDFT
Block Equalizer Removal

Functions Common to OFDMA and SC-FDMA

SC-FDMA Only

The block diagram of Figure 2.5-1 shows a basic SC-FDMA transmitter / receiver arrangement. Note that many of the
functional blocks are common to both SC-FDMA and OFDMA, thus there is a significant degree of functional
commonality between the uplink and downlink signal chains. The functional blocks in the transmit chain are:
1. Constellation mapper: Converts incoming bit stream to single carrier symbols (BPSK, QPSK, or 16QAM
depending on channel conditions)
2. Serial/parallel converter: Formats time domain SC symbols into blocks for input to FFT engine

3. M-point DFT: Converts time domain SC symbol block into M discrete tones

4. Subcarrier mapping: Maps DFT output tones to specified subcarriers for transmission. SC-FDMA systems
43 tones (distributed) as shown in Figure 2.5-2. The
either use contiguous tones (localized) or uniformly spaced
 Summary
Narrowband
Wideband CDMA Wideband OFDMA
system

Signal Narrowband Wideband Wideband

Intra-cell bandwidth
Orthogonal Pseudorandom Orthogonal
allocation

Intra-cell interference None Significant None

Inter-cell bandwidth
Partial reuse Universal reuse Universal reuse
allocation
Inter-cell uplink
Bursty Averaged Averaged
interference
Accuracy of power
Low High Low
control

Operating SINR High Low Range: low to high

PAPR of uplink signal Low Medium High

44