Act 2 Cdma2011 Beamer

Principles of Code Division Multiple Access
(CDMA)
Professor A. Manikas
Chair of Communications and Array Processing
Imperial College London
EE401: Advanced Communication Theory
Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 1 / 94

Table of Contents
1 Introduction
De…nition of a SSS
Classi…cation of SSS
Modelling of b (t ) in SSS
Applications of Spread Spectrum Techniques
De…nition of a Jammer
De…nition of a MAI
Processing Gain (PG)
Equivalent EUE
2 Principles of PN-sequences
Comments on PN-sequences Main Properties
An Important "Trade-o¤"
3 m-sequences
Shift Register and Primitive Polynomials
Implementation of an m-sequence
Autocorrelation of ‘m-sequences’
4 Gold Sequences
5 Basics of CDMA
Basic Properties of CDMA Systems
DS-CDMA: Synchronization
Mobile Cellular Systems: Conventional & CDMA
Channel Reuse and Reuse Distance
Signal Overlay
6 Analysis of a Direct Sequence CDMA System
Main Assumptions
DS/BPSK CDMA System: Modelling and Analysis
SNIRout as a function of EUE, Nc and K
BER as a function of EUE, Nc and K
BPSK Examples
DS/QPSK CDMA System: Modelling and Analysis
Using one PN-code per QPSK-user
SNIRout and BER as a function of EUE, Nc and K
Symbol Error Probability pe,cs
QPSK Example with One PN-code.
7 Some Important CDMA System Components
Power Control
Voice Activity Factor
Sectorization
8 Appendices
Appendix-1: Proof of DS-BPSK/CDMA SNIRout (Equ. 29)

Introduction
Introduction
General Block Diagram of a Digital Comm. System (DCS)

Introduction
Most of the current cellular systems, such as GSM, use frequency

division multiplex - time division multiplex (FDM-TDM) techniques
to improve the system capacity.
Gaussian MSK
456 bits
M=4
Message 20 msec operating on
signal Uniform i.e. rb=22.8kbits/s 148 bits
quantizer per TDMA frame
bandwidth
Fg=4kHz Q=213
BUE=0.3
148bits
0.577ms plus 8.25 guard bits
Sampling 160 levels TDMA FRAME=4.615ms

Frequency 20 msec 260 bits
Fs=8kHz 20 msec SPECTRUM
= 2080 bits i.e. rb=13kbits/s
20 msec 200kHz
i.e. rb=104kbits/s f
890MHz 915MHz 935MHz 960MHz
0 Downlink 25MHz Uplink=25MHz

Introduction
HSCDS: High Speed Circuit Switched Data

GPRS: General Packet Radio Systems (2+)
EDGE: Enhanced Data Rate GSM Evolution (2+)
UMTS:Universal Mobile Telecommunication Systems (3G)
Introduction

Introduction
Industry Transformation and Convergence [from Ericsson 2006, LZT

123 6208 R5B]
WCDMA (Wideband CDMA) is a 3G mobile comm system. It is a

wireless system where the telecommunications, computing and media
industry converge and is based on a Layered Architecture design.
(Note: CDMA Systems 2 the class of SSS).

Introduction De…nition of a SSS
De…nition of a SSS
When a DCS becomes a Spread Spectrum System (SSS)
LEMMA
8 1: CS , SSS
< Bss message bandwidth (i.e. BUE=large)
i¤ Bss 6= f{message}
:
spread is achieved by means of a code which is 6=f{message}
where Bss =transmitted SS signal bandwidth
our AIM: ways of accomplishing LEMMA-1.
NB:
I PCM, FM, etc spread the signal bandwidth but do not satisfy the
conditions to be called SSS
I Btransmitted-signal Bmessage
)SSS distributes the transmitted energy over a wide bandwidth
) SNIR at the receiver input is LOW.
Nevertheless, the receiver is capable of operating successfully because

the transmitted signal has distinct characteristics relative to the noise

(a) SSS: (b) CDMA (K users):

The PN signal b (t ) is a function of a PN sequence of 1’s fα[n ]g
I The sequences fα[n ]g must agreed upon in advance by Tx and Rx and

they have status of password.
I This implies that :

F knowledge of {α[n ]})demodulation=possible
F without knowledge of fα[n ]g )demod.=very di¢ cult
I If fα[n ]g (i.e. “password”) is purely random, with no mathematical

structure, then
F without knowledge of fα[n ]g )demodulation=impossible
I However all practical random sequences have some periodic structure.
This means:
α[n ] = α[n + Nc ] (1)
where Nc =period of sequence
i.e. pseudo-random sequence (PN-sequence)
Introduction Classi…cation of SSS
Classi…cation of SSS

Introduction Modelling of b (t ) in SSS
Modelling of b(t) in SSS

DS-SSS (Examples: DS-BPSK, DS-QPSK):
b (t ) = ∑ α[n].c (t nTc ) (2)

n
where fα[n]g is a sequence of 1’s;

c (t ) is an energy signal of duration Tc
FH-SSS (Examples: FH-FSK)
b (t ) = ∑ exp fj (2πk[n]F1 t + φ[n])g .c (t nTc ) (3)

n
where fk[n ]g is a sequence of integers such that
fα[n]g 7! fk[n]g (4)
and fα[n]g is a sequence of 1’s;

c (t ) is an energy signal of duration Tc
Introduction Applications of Spread Spectrum Techniques
Applications of Spread Spectrum Techniques

1 Interference Rejection: to achieve interference rejection due to:
I Jamming (hostile interference). N.B.: protection against cochannel
interference is usually called anti-jamming (AJ)
I Other users (Multiple Access Interefence - MAI): Spectrum shared by
“coordinated “ users.
I Multipath: Self-Jamming by delayed signal
2 Energy Density Reduction (or Low Probability of Intercept LPI). LPI’
main objectives:
I to meet international allocations regulations
I to reduce (minimize) the detectability of a transmitted signal by
someone who uses spectral analysis
I privacy in the presence of other listeners
3 Range or Time Delay Estimation
NB: interference rejection = most important application

Introduction De…nition of a Jammer
Jamming source, or, simply Jammer is de…ned as follows:
Jammer , intentional (hostile) interference
F the jammer has full knowledge of SSS design except the jammer does
not have the key to the PN-sequence generator,
F i.e. the jammer may have full knowledge of the SSSystem but it does
know the PN sequence used.
Introduction De…nition of a MAI
Multiple Access Interference (MAI) is de…ned as follows:
MAI , uniintentional interference

Introduction Processing Gain (PG)
PG: is a measure of the interference rejection capabilities

de…nition:
Bss 1/Tc Tcs
PG , = = (5)
B 1/Tcs Tc
where B=bandwidth of the conventional system
PG is also known as "spreading factor" (SF)
PG = very important in DS-SSS
PG 6= very important in FH-SSS

Introduction Equivalent EUE
Remember:
F Jamming source, or, simply Jammer = intentional interference
F Interfering source = unintentional interference
F With area-B = area-A we can …nd Nj

if
BJ = qBss ; 0 < q 1 (6)
then
Eb Ps .BJ Ps .q.Bss
EUEJ = = = = PG SJRin q (7)
NJ PJ .rb PJ .B
Eb
EUEequ = (8)
N0 + NJ
1
N0
= PG SJRin q +1 (9)
Nj
where
Ps
SJRin , (10)
PJ

SS Transmission in the presence of a Jammer (or MAI)

SS Reception in the presence of a Jammer (or MAI)

Principles of PN-sequences
PN-codes (or PN-sequences, or spreading codes) are sequences of

+1s and -1s (or 1s and 0s) having special correlation properties which
are used to distinguish a number of signals occupying the same
bandwidth.
Five Properties of Good PN-sequences:
Property-1 easy to generate

Property-2 randomness
Property-3 long periods
Property-4 impulse-like auto-correlation functions
Property-5 low cross-correlation

Principles of PN-sequences Comments on PN-sequences Main Properties
Comments on PN-sequences Main Properties
Comments on Properties 1, 2 & 3

I Property-1 is easily achieved with the generation of PN sequences by
means of shift registers,
while
I Property-2 & Property-3 are achieved by appropriately selecting the
feedback connections of the shift registers.

Comments on Property-4
I to combat multipath, consecutive bits of the code sequences should be
uncorrelated.
i.e. code sequences should have impulse-like autocorrelation functions.
Therefore it is desired that the auto-correlation of a PN-sequence is
made as small as possible.
I The success of any spread spectrum system relies on certain
requirements for PN-codes. Two of these requirements are:
1 the autocorrelation peak must be sharp and large (maximal) upon
synchronisation (i.e. for time shift equal to zero)
2 the autocorrelation must be minimal (very close to zero) for any time
shift di¤erent than zero.
I A code that meets the requirements (1) and (2) above is the
m-sequence which is ideal for handling multipath channels.

I The …gure below shows a shift register of 5 stages together with a

modulo-2 adder. By connecting the stages according to the coe¢ cients
of the polynomial D 5 + D 2 + 1 an m-sequence of length 31 is
generated (output from Q5).
The autocorrelation function of this m-sequence signal is shown in the
previous page
Modulo-2 adder Modulo-2 adder
+ +
o/p
Q1 Q2 Q3 Q4 Q5 Q3 Q5
i/p
1 2 3 4 5 i/p
1 2 3 4 5
clock clock
Shift register 1/Tc Shift register
(a) (b)

Comments on Property-5
I If there are a number of PN-sequences
fα1 [k ]g, fα2 [k ]g, ...., fαK [k ]g (11)
then if these code sequences are not totally uncorrelated, there is

always an interference component at the output of the receiver which is
proportional to the cross-correlation between di¤erent code sequences.
I Therefore it is desired that this cross-correlation is made as small as

possible.

Principles of PN-sequences An Important "Trade-o¤"
An Important "Trade-o¤"
There is a trade-o¤ between Properties-4 and 5.

In a CDMA communication environment there are a number of
PN-sequences
fα1 [k ]g, fα2 [k ]g, ...., fαK [k ]g
of period Nc which are used to distinguish a number of signals
occupying the same bandwidth.
Therefore, based on these sequences, we should be able to
F combat multipath
(which implies that the auto-correlation of a PN-sequence
fαi [k ]g should be made as small as possible)
F remove interference from other users/signals,
(which implies that the cross-correlation should be made as small as
possible).

Principles of PN-sequences An Important "Trade-o¤"
However
2
Rauto 2
+ Rcross > a constant which is a function of period Nc (12)
i.e. there is a trade-o¤ between the peak autocorrelation and

cross-correlation parameters.
Thus, the autocorrelation and cross-correlation functions cannot be
both made small simultaneously.
The design of the code sequences should be therefore very careful.
N.B.:
I A code with excellent autocorrelation is the m-sequence.
I A code that provides a trade-o¤ between auto and cross correlation is
the gold-sequence.

m-sequences
m-sequences - de…nition
m-seq.: widely used in SSS because of their very good autocorrelation

properties.
PN code generator: is periodic
i.e. the sequence that is produced repeats itself after some period of
time
De…nition : A sequence generated by a linear m-stages Feedback
shift register is called a maximal length, a maximal sequence, or
simply m-sequence, if its period is
Nc = 2m 1 (13)
(which is the maximum period for the above shift register generator)
The initial contents of the shift register are called initial conditions.

m-sequences Shift Register and Primitive Polynomials
Shift Register and Primitive Polynomials
The period Nc depends on the feedback connections (i.e. coe¢ cients

ci ) and Nc = max, i.e. Nc = 2m 1, when the characteristic
polynomial
c ( D ) = cm D m + cm 1D
m 1
+ .... + c1 D + c0 with c0 = 1 (14)
is a primitive polynomial of degree m.
0 =) no connection
rule: if ci = (15)
1 =) there is connection
De…nition of PRIMITIVE polynomial = very important

(see SSS topic Comm Systems E3)

m-sequences Shift Register and Primitive Polynomials
Some Examples of Primitive Polynomials
degree-m polynomial
3 D3 + D + 1
4 D4 + D + 1
5 D5 + D2 + 1
6 D6 + D + 1
7 D7 + D + 1
Please see Comm Systems LNs (Spread Spectrum Topic) for some
tables of irreducible & primitive polynomial over GF(2).

m-sequences Implementation of an m-sequence
Implementation of an m-sequence
use a maximal length shift register
i.e. in order to construct a shift register generator for sequences of any
permissible length, it is only necessary to know the coe¢ cients of the
primitive polynomial for the corresponding value of m
1
fc = = chip-rate = clock-rate (16)
Tc
c ( D ) = cm D m + cm 1D
m 1
+ .... + c1 D + c0 (17)
with c0 = 1 (18)

m-sequences Implementation of an m-sequence
Example: c (D )= D 3 + D + 1 = primitive=)power= m = 3
I coe¢ cients=(1, 0, 1, 1) ) Nc = 7 = 2m 1 i.e.period= 7Tc
o/p
1st 2nd 3rd
initial condition 1 1 1
clock pulse No.1 0 1 1
Note that the sequence of 0’s and 1’s is transformed to a sequence of
1s by using the following function
o/p = 1 2 i/p (19)
m-sequences Autocorrelation of ‘m-sequences’
Autocorrelation of ‘m-sequences’
An m-sequ. fα[n ]g has a two valued auto-correlation function:
(
Nc
Nc k = 0 mod Nc
Rαα [k ] = ∑ α[n ]α[n + k ] = (20)
n =1 1 k 6= 0 mod Nc
This implies that Rbb (τ ) is also a "two-valued"

Rbb (τ ):
Remember that a sequence fα[n ]g of period Nc = 2m 1, generated

by a linear FB shift register, is called a maximal length sequence.
Gold Sequences
Gold Sequences
Although m-sequences possess excellent randomness (and especially

autocorrelation) properties, they are not generally used for CDMA
purposes as it is di¢ cult to …nd a set of m-sequences with low
cross-correlation for all possible pairs of sequences within the set.
However, by slightly relaxing the conditions on the autocorrelation

function, we can obtain a family of code sequences with lower
cross-correlation.
Such an encoding family can be achieved by Gold sequences or Gold

codes which are generated by the modulo-2 sum of two m-sequences
of equal period.

Gold Sequences
The Gold sequence is actually obtained by the modulo-2 sum of two

m-sequences with di¤erent phase shifts for the …rst m-sequence
relative to the second. Since there are Nc = 2m 1 di¤erent relative
phase shifts, and since we can also have the two m-sequences alone,
the actual number of di¤erent Gold-sequences that can be generated
by this procedure is 2m + 1.
These sequences, however, are not maximal length sequences.

Therefore, their auto-correlation function is not the two valued one
given by Equ. (20). The auto-correlation still has the periodic peaks,
but between the peaks the auto-correlation is no longer ‡at.

Gold Sequences
Example of a Gold Sequence of Nc = 127 = 27 1

Rbb (τ ):
Example of an m-sequence of Nc = 127 = 27 1

Rbb (τ ):

Basics of CDMA
Basics of CDMA
BLOCK DIAGRAM

Basics of CDMA
Example: DS-BPSK CDMA System
SISO = Scalar-Input Scalar-Output Channel

Basics of CDMA
SISO Multipath Channel
SISO Multipath channel of the i-th user
In the absence of multipaths the above diagram has only τ i 1 and βi 1

terms.
For the simplicity we will drop the second subscript and we will use τ i
and βi ,and thus the BPSK/DS-CDMA in the absence of multipaths
may be represented as follows:&

Basics of CDMA

Basics of CDMA Basic Properties of CDMA Systems
Basic Properties of CDMA Systems
CDMA is one of the applications of spread spectrum communications

which is used in civilian, commercial and military communication.
Two systems: DS-CDMA (i.e averaging system) and FH-CDMA (i.e.
avoidance system).
In this course only DS-CDMA will be considered.
Assign a speci…c PN-code to each user
PN-code (having the status of ‘password’) acts like a ‘channel’
DS-CDMA: two main cases
I PN-signal period = Nc Tc = Tcs (known as ‘short codes’CDMA)
I PN-signal period = Nc Tc Tcs (known as ‘long codes’CDMA)

Basics of CDMA Basic Properties of CDMA Systems
DS-CDMA: two main types

1 synchronous
8 DS-CDMA
< i-th user
>
>
: j-th user
2 non-synchronous
8 DS-CDMA
> i-th user
<
>
:
j-th user

Basics of CDMA DS-CDMA: Synchronization
DS-CDMA: Synchronization
The Rx requires a replica of the PN code, with the correct clock
phase, in order to despread the signal.
Therefore, Rx =“synchronization circuits” + “demod. circuits”
The process of synchronizing the receiver to the transmitter’s PN
code consists of two stages:
I Acquisition (coarse synchronization).
I Tracking (…ne synchronization).
Operation: acquisition; tracking + demodulation; loose tracking;

acquisition; tracking+demodulation; ......etc..........
Basics of CDMA Mobile Cellular Systems: Conventional & CDMA
Mobile Cellular Systems: Conventional & CDMA
A mobile cellular system consists of base stations, cells (a cell is the

area serviced by a base station) and mobiles (subscribers). When a
call originates, the base station negotiates with the mobile on various
aspects (such as the channel used etc.), before establishing
communications. After this, as the mobile moves from cell-to-cell, the
service is handed (hand-o¤ or handover) from one base station to
another.
Only one base station will service a mobile at any one time.
Note:
I base station to mobile is known as FORWARD LINK
I mobile to base station is known as REVERSE LINK

Type of channels:
UPLINK DOWNLINK
Tra¢ c Channel Tra¢ c Channel
Access Channel Pilot Channel
Synchron.Channel
Paging Channel
Frequency Division Duplex (FDD) and Time Division Duplex (TDD)

Channel Reuse and Reuse Distance

There is interference from other cells sharing the same channels. The
reuse distance D, in these systems, is determined by the worst case
interference situation.
Current cellular systems = FDMA/TDMA
Most of the current cellular systems, such as GSM, use frequency
division multiplex - time division multiplex (FDM-TDM) technique to
improve the system capacity. In these systems, each user is assigned
one time-frequency slot.
I When the system gets larger,
slots 6= unique for each and every user
as this will limit the system capacity. Therefore these slots

(time/frequency) have to be reused
(reused in cells separated by D (cells), which is the reuse distance of
the system).

The system capacity could be increased by increasing the number of

channels available in a single cell, i.e. reducing the reuse distance D.
But this reduction is limited by the co-channel interference, (i.e. the
interference from other cells sharing the same channels).
The reuse distance D, in these systems, is determined by the worst
case interference situation.

In a CDMA system, the available spectrum and time are not split into
distinct slots. Instead the whole (available) spectrum is used by each
user.
Since the same frequency channel could be used by all the
users/subscribers, the reuse distance D could be reduced to 1, i.e.
if CDMA then D = 1 (21)

Signal Overlay
The spread spectrum signal, from a CDMA system, has a very low power
spectral density and, therefore, a CDMA system can overlay on top of
existing narrow-band mobile cellular systems (of the same frequency
band).
This is because the interference (due to CDMA signals), added to a
narrow-band mobile system channel, is very low and, therefore,
the presence of CDMA signal will hardly a¤ect the performance of the
narrow-band mobile system.
The CDMA system, however, needs to perform some extra processing to
reject the narrowband interference due to the presence of the narrow-band
signals.
Comment:
The capacity and performance of a mobile cellular system could be
signi…cantly improved by using CDMA techniques. In the paper “On the
Capacity of a Cellular CDMA”, IEEE Transactions on Vehicular
Technology, Vol.40, 1991 (by Gilhousen et al) the improvement in the
capacity is discussed and it is stated that “no other proposed scheme
appears to even approach this (CDMA) performance”.
Analysis of a Direct Sequence CDMA System
Analysis of a Direct Sequence CDMA System
Two systems will be analysed

F DS/BPSK
F DS/QPSK
objective:
to relate the BER pe with the total number of users K as well as with
the EUEequ at the receiver.
i.e.
pe = f{EUE equ
,K } (22)

Analysis of a Direct Sequence CDMA System Main Assumptions
Main Assumptions
single cell system of K users,

@ multipaths
PN code period = Nc = PG
System=perfectly power-controlled
(all SS signals arrive at the receiver with the same power)
NB: power control can often be implemented in practice with great
accuracy.
System = totally asynchronous (there is no common timing reference
for the transmitters/users)
F NB: This is actually an advantage of CDMA over other multiple access
techniques, because all users can transmit independently and no
signalling information is required.

Analysis of a Direct Sequence CDMA System DS/BPSK CDMA System: Modelling and Analysis
DS/BPSK CDMA System: Modelling and Analysis
p
Note that the carrier of i th transmitter is Pi . exp (j (2πFc t + ψi ))

i-th user’s data signal mi (t ) and PN-signal bi (t ):

8
>
> mi (t ) ∑ ai [n ]. c1 (t n.Tcs ) ; nTcs < t (n + 1)Tcs
>
< n
bi (t ) = ∑ αi [k ]. c2 (t kTc ) ; kTc < t (k + 1)Tc (23)
>
> k j k
>
: with Tcs = Nc Tc ; Nkc = n;
The period of each user’s PN-sequence is selected as Nc = TTcsc , and

therefore there is one code period per data bit (or Nc chips per bit).
Thus, for the BPSK case, the processing gain PG is:
Tcs
PG = Nc = (24)
Tc
The transmitted signal si (t ) of the i-th user is therefore
p
si (t ) = Pi .mi (t ).bi (t ). exp (j (2πFc t + ψi )) (25)
where Fc is assumed common for all carriers.

Since the transmitters are not time-synchronous, there is a di¤erent

time delay τ i for each signal si (t) before it reaches the receiver, with
0 τ i < Tcs for i = 1, 2, 3, ..., K . The carrier phases ψi are also
assumed di¤erent so that 0 ψi < 2π for i = 1, 2, 3, ...K . Thus,
ignoring the band-pass …lters at the transmitters and the receiver, the
received signal r (t ) can be described as follows:
K p
r (t ) = ∑ |βi {zP}i .mi (t τ i ).bi (t τ i ). exp(j (2πFc t + φi )) + n(t )
i =1 p
, P
(26)
where
φi = ψi 2πFc τ i (27)
with n(t ) denoting the additive white Gaussian channel noise of
double sided power spectral density N0 /2

If the transmitted signals are all synchronized, the delays τ i , 8i,are

neglected. Synchronizing all transmitted signals requires a common
timing reference and compensation for transmission delays in various
transmission paths. This complicates the system requirements and
has no clear advantage.
The receiver has a local PN-signal generator as well as a carrier
generator that generate exact replicas of those used in the transmitter.
The receiver is in perfect synchronism with the transmitter by using
acquisition and tracking techniques (i.e. τ 1 , φ1 are known).
Thus, without loss of generality, let us assume that τ 1 = 0, φ1 = 0.
The output of the 1-st correlator (desired) at t = Tcs is thus
Z nT cs
1
G1 = r (t ).b1 (t ). exp( j (2πFc t )dt (28)
Tcs (n-1 )T cs

The SNIR at the output of the desired receiver can be found to be

(for proof see Appendix-1, page 85)
P
SNIRout = K Z T cs
(29)
N0 P
2Tcs
+ 2
Tcs ∑ 0
Rb2i b1 (τ i ) + R̃b2i b1 (τ i ) d τi
| {z } | i =2
{z }
,P n,out ,P MAI
( R τi
Rbi b1 (τ i ) =
b (t τ i ).b1 (t )dt even cross-corr.
where R0T cs i (30)
R̃bi b1 (τ i ) = τi bi (t τ i ).b1 (t )dt odd cross-corr.
It is obvious from Equ 29 that the contribution of the i -th interfering

signal on to SNIRout , is inversely proportional to the (even/odd)
periodic cross-correlation functions between the i -th and desired-user’s
code waveforms. Therefore, designing the code waveforms so that
these cross-correlation functions are as small as possible is essential for
reducing the total other-user interference (known as multiple access
interference) and thus enhancing the system performance.
1-st user (desired)

a1 [ n 1 ] a1 [ n ] a1 [ n + 1 ]
i-th user (MAI)

ai [ n 1 ] ai [ n ] ai [ n + 1 ]

SNIR-out as a function of EUE, Nc and K

Equ 29 may be rewritten as:
Z T cs
! 1
K
N0 1
SNIRout =
2PTcs
+ 2
Tcs ∑ 0
i =2
= for you ...
1
K 1 1
+
3Nc 2 EUE

BER as a function of EUE, Nc and K
We have seen that a good approximation to Equation 29 is:

1
K 1 1
SNIRout + (31)
3Nc 2 EUE
However,
pe = T npSNIR out
o
(32)
therefore, 8 9
>
> >
>
< =
pe = T >> r K 1
1
1 > >
(33)
: + ;
3Nc 2 EUE

Alternatively by using the concept of EUEequ we have

=E b
z }| {
P (K 1).PTcs (K 1).Eb
Nj = ( K 1 ) . = (K 1).P.Tc = =
Bss Nc Nc
(34)
which gives us the following SNIRout (not necessarily equal to
Equation-29),
E Eb Eb 1
SNIRout = 2 =2 = 2 (K 1 ).E b
= (K 1 )
Nequ NJ + N0 + N0 + 1
| {z } Nc 2N c 2 EUE
,N equ
(35)
which implies that
8 9
np o < 1 =
pe = T 2EUEequ =T q (36)
: K 1
+ 1 ;
2N c 2 EUE

BPSK Examples



Analysis of a Direct Sequence CDMA System DS/QPSK CDMA System: Modelling and Analysis
DS/QPSK CDMA System: Modelling and Analysis

Using one PN-code per QPSK-user
The data bits of each user are now grouped in symbols of 2-bits. For
instance,
input output QPSK phase equivalent ‘symbol’

00 m1 (t ) 0 1 exp (j0 )
01 m2 (t ) 90 j exp (j90 )
11 m3 (t ) 180 1 exp (j180 )
10 m4 (t ) 270 j exp (j270 )
and the i-th user’s binary information may be now represented by two
di¤erent data signals mIi (t ) and mQi (t ) and one PN-signals bi (t ), as
follows:


i-th user’s data signals :

8 9
< inphase: mIi (t ) ,
> ∑ a i [ n ] . c1 ( t n.Tcs ) >
=
n (even )
(37)
:quadrature: mQi (t ) ,
> ∑ a i [ n ] . c1 ( t n.Tcs )>
;
n (odd )
Note that mIi (t ) is formed by the even data bits and mQi (t ) by the
odd ones.

At each transmitter, the QPSK signals/symbols are …rstly multiplied

by the PN-code waveform bi (t ),
bi ( t ) = ∑ αi [k ].c2 (t kTc )
k
and then by the carrier

r
Pi
. exp (j (2πFc t + ψi )) (38)
2
while the spread spectrum signal si (t ) of the i-th user (i.e.
transmitted signal) is:
r
Pi
si ( t ) = .(mIi (t ) + jmQi (t ))bi (t ) exp (j (2πFc t + ψi )) (39)
2
where Fc is assumed common for all carriers.

It is important to pointqout that now the amplitude of the carriers has

p
changed from Pi to P2i so that the power of the transmitted
signal si (t ) still remains equal to Pi . Indeed the power of si (t ) is
Pi = E fsi (t )2 g = E fsi (t ).si (t )g = for .you.....
r !2
Pi
= 2 = Pi
2
Thus the received signal r (t ) can be expressed as
r
K
P
r (t ) = ∑ . (mIi (t τ i ) + jmQi (t τ i )).bi (t τ i ) (40)
i =1 2
exp (j (2πFc t + φi )) + n (t ) (41)
where
r r
P Pi
= βi (42)
2 2
φi = ψi 2πFc τ i (43)
The received signal is applied at the input of a DS/QPSK1 correlation

receiver, matched to the signal s1 (t ).
The outputs of the I and Q correlators (just before the decision
devices) at time t = nTcs are as follows:
Z T cs
1
G1 = r (t ).b1 (t ). exp( j2πFc t )dt (44)
Tcs 0
GI 1 = Re fG1 g (45)
and
GQ 1 = Re fG1 g (46)

Then it can be found (proof for you) that the SNIR at the output of
the inphase branch output of the desired correlator is
1 P
SNIRout = (47)
2 K ZT cs n o
N0 P
2Tcs
+ 2
Tcs ∑ Rb2i b1 (τ i ) + R̃b2i b1 (τ i ) d τ i
i =2 0
By symmetry, the SNIRout of the quadrature branch is also given by

a similar equation.
By comparing Equ 29 with Equ 47, it is clear that the SNIRout for
QPSK system is half that of BPSK system.

A simpli…ed version of the above system is a system which uses the

same PN-signal for both I & Q branches. The block structure of this
system is given below:

SNIR-out and BER as a function of EUE, Nc and K

From Equation 47 it can be seen that the SNIRout of the QPSK
system is half that of the BPSK system. Therefore, an approximation
to Equation 47 can be made as
1
1 K 1 1
SNIRout + (48)
2 3Nc 2 EUE
i.e.
1
2(K 1) 1
SNIRout + (49)
3Nc EUE
The bit error rate of the inphase branch is related to the SNIRout as
np o
pe =
8
SNIRoutT 9
(50)
< =
=) pe = q
: 2K 1 + 1
1
T ;
(51)
3N c EUE

Symbol Error Probability
Since, the bit errors in the two branches are statistically independent
the probability of symbol error is given as follows
pe,cs = 1 (the joint probability of no bit error in both branches)

= 1 ( 1 pe ) 2
2 pe
= (if
Equation 51 is used then ..)
8 9
< =
= 2 qT 1
: 2K 1 + 1 ;
(52)
3N c EUE

QPSK Example with One PN-code
.
The data bits of each user are now grouped in symbols of 2-bits.
For instance, consider a DS-QPSK SSS with φ = 45 (Gray code)
input output QPSK phase equivalent ‘symbol’

p
2
00 m1 (t ) 45 (1+j) exp (j45 )
p2
2
01 m2 (t ) 135 ( 1+j) exp (j135 )
p2
2
11 m3 (t ) 225 ( 1 j) exp (j225 )
p2
2
10 m4 (t ) 315 2 (+1 j) exp (j315 )

At the Transmitter:


Some Important CDMA System Components Power Control
Power Control
In order to achieve the full bene…ts of using CDMA, the transmitted

signal powers Psi should be controlled in such a way that received
signal power, from all the users at a cell, are the same.
This makes power control a key feature of CDMA mobile systems.

Some Important CDMA System Components Voice Activity Factor
Voice Activity Factor

Human speech contains a lot of pauses where there is no data to
transmit.
Thus a speaker is active for about half the time due to listening and
pauses in speech.
The fraction of time that a speaker is active is known as

the voice activity factor a
Extensive studies have shown that 0.35 < a < 0.5.
A popular value used is a = 3/8 = 0.375
The voice activity feature can be taken into account in a
communication system by suppressing the transmission when voice is
absent.
Assuming that we have a scheme where the carrier is turned-o¤
during the speech idle periods then a reduction in interference (by a
factor of the voice activity) can be achieved.

(
TDMA/FDMA: very di¢ cult
implementation of voice activity:
CDMA: very easy
For a large number of users the capacity increases by a factor 1/a.
Therefore, using the voice activity monitoring approach the capacity
and the performance of a CDMA system will be improved
(this improvement cannot be obtained in FDMA/TDMA systems)
In particular the power of a user’s signal at a speci…c time instant can
be expressed 1 Puser with probability a and 0 Puser with
probability (1 a).

Therefore we can model the “on-o¤” activity of each user a binomial

distribution, which implies that the probability that k user (out of K )
are active is given as follows:
k k
Pr(k users are active) = a (1 a )K k
(53)
K
where K is the number of users per cell

Note that as K =")spread of distribution=#
Set a threshold Kth , eth such that:
Pr(number of active users > Kth ) < eth (54)

Some Important CDMA System Components Sectorization
Sectorization is achieved by using directional antennas instead of

omnidirectional antennas.
Each cell is divided to three sectors using three directional antennas
each having 120 beamwidth.
Using sectorization the performance can be improved even more.
(The expected value of the total interference is reduced by a factor of
3 wrt single omnidirectional antenna case)
Eb
BPSK : SNIRout = 2.EUEequ = 2 = (for you) = (55)
Nj + N0
In practice: 3 dB<SNIRout < 15dB

It is used in both TDMA/FDMA and CDMA systems

A better approach is to use three linear antenna arrays (smart

antennas)
antenna array switched beam array system adaptive array system

Appendices Appendix-1: Proof of DS-BPSK/CDMA SNIRout (Equ. 29)
Appendices
Appendix-1: Proof of DS-BPSK/CDMA SNIR-out (Equ.29)
The decision variable G1 is given as follows:

=r (t ). exp ( j 2πF c t )
z( }| ){
nT
Z cs
1 K p
G1 =
Tcs ∑ Pmi (t τ i ).bi (t τ i ) exp(j φi ) .b1 (t )dt
i =1
(n-1 )T cs
nT
Z cs
1
+ n (t ).b1 (t ). exp( j2πFc t ) dt (56)
Tcs
(n-1 )T cs
Without loss of generality, we have assumed that the 1-st (desired)

user is fully synchronised (i.e. τ 1 , φ1 are known) and let us assume
τ 1 = 0, φ1 = 0.

desired term
z Z
}| {
1 nT cs p
G1 = P. m1 (t ) .b1 (t ) .b1 (t )dt
Tcs (n 1 )T cs | {z }
= 1 =a [n ]
MAI term
z }| {
K Z nT cs p
1
+
Tcs ∑ P.mi (t τ i ).bi (t τ i ) exp(j φi ).b1 (t )dt
i =2 (n 1 )T cs
noise term
z Z nT cs
}| {
1
+ n (t ).b1 (t ). exp( j2πFc t )dt (57)
Tcs (n 1 )T cs
R nT cs R T cs
To simplify the notation (n 1 )T cs will be replaced by 0
repeated in
every interval for any n.

1-st user (desired)

a1 [ n 1 ] a1 [ n ] a1 [ n + 1 ]
i-th user (MAI)

ai [ n 1 ] ai [ n ] ai [ n + 1 ]

(
ai [n 1] denotes the previous information bit
For the i-th user
ai [n] denotes the current information bit
Note: ai [n 1] and ai [n] are variables of 1’s
desired term:
a 1 [n ]
Z T cs p z }| {
1
desired = P.m1 (t )b12 (t )dt
Tcs 0
=T cs
p zZ }| {
P T cs
= a1 [ n ] b12 (t )dt
T 0
pcs
= P.a1 [n ] (58)

MAI term:
1 =ai [n 1 ]
K Z τi p z }| {
1
MAI =
Tcs ∑f 0
P mi (t τ i ).bi (t τ i ).b1 (t ) exp(j φi )dt
i =2
1 =ai [n ]
Z T cs p z }| {
+ P.mi (t τ i ).bi (t τ i ).b1 (t ) exp(j φi )dt g
τi
p K Z τi
P
=
Tcs ∑ f ai [ n 1]
0
bi ( t τ i ).b1 (t )dt
i =2
Z T cs
+ ai [ n ] bi (t τ i ).b1 (t )dt g exp(j φi )
τi
p K
P
Tcs i∑
= ai [n 1]Rbi b1 (τ i ) + ai [n]R̃bi b1 (τ i ) exp(j φi )(59)
=2

where Rbi bj (τ i ) and R̃bi bj (τ i ) are the partial cross-correlation

functions de…ned by:
Z τi
Rbi bj (τ i ) = bi ( t τ i ).bj (t ) dt (60)
0
Z T cs
R̃bi bj (τ i ) = bi ( t τ i ).bj (t ) dt (61)
τi
noise term
Z T cs
1
noise term = n (t ).b1 (t ). exp( j2πFc t )dt
Tcs 0
and, …nally,
p
p P K
G1 = a 1 [ n ] P +
Tcs ∑ ai [ n 1]Rbi b1 (τ i ) + ai [n ]R̃bi b1 (τ i ) . exp(j φi )
i =2
Z T cs
1
+ n (t ).b1 (t ). exp( j2πFc t ) dt (62)
Tcs 0

From Equ 62, it is evident that the …rst term is the desired signal
term, the second term the code noise produced by the
cross-correlation with the codes of other users, and the last term the
additive white Gaussian noise introduced by the channel.
It can also be seen that G1 is a random variable that is dependent on
the mutually independent random variables φi , τ i , ai [n 1] and ai [n]
for 1 i K , i 6= j. To simplify, it is reasonable to assume that the
random variables φi and τ i are uniformly distributed on the intervals
[0, 2π ] and [0, Tcs ] respectively, and ai [n 1], ai [n] take values 1
with equal probability.

With the above assumptions, the powers of the three terms in 62 can
be found as follows:
I desired power:
p 2
Pdesired = E P.a1 [n ] =P (63)
I MAI power:
8 !2 9
< pP K =
PMAI = E
: Tcs ∑ ai [ n 1]Rbi b1 (τ i ) + ai [n ]R̃bi b1 (τ i )
;
(64)
i =2
P K n n oo
∑
2
= E ai [ n 1]Rbi b1 (τ i ) + ai [n ]R̃bi b1 (τ i )
Tcs2 i =2
(65)

I Since
E fai [n 1]g = E fai [n]g = 0 (66)

n o n o
E [ai [n 1]]2 = E [ai [n]]2 = 1 (67)
n o
E (exp(jφi ))2 = 1 (68)
Equation 65 becomes
K n o
P 2 2
PMAI = 2
Tcs ∑ E ai [ n 1]Rbi b1 (τ i ) + E ai [n]R̃bi b1 (τ i )
i =2
Z
Pc K T cs
= 3 ∑
Tcs i =2 0
Rb2i b1 (τ i ) + R̃b2i b1 (τ i ) d τi (69)
I noise term:
( Z 2
)
1 T cs N0
Pnoise = E n (t ).b1 (t ). exp( j2πFc t ) dt = ... =
Tcs 0 2Tcs
(70)
Therefore, the SNIRout at the output of the correlator is found to be

P
SNIRout = R T cs
N0 P
+ 2 ∑Ki=2 0
2Tcs Tcs
(71)
1
N.B.: Note that Fc T cs . This condition is always satis…ed in a
practical system.

Act 2 Cdma2011 Beamer

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Act 2 Cdma2011 Beamer

Uploaded by

Copyright:

Available Formats

Principles of Code Division Multiple Access

Imperial College London

EE401: Advanced Communication Theory

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 1 / 94

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 2 / 94

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 3 / 94

Most of the current cellular systems, such as GSM, use frequency

Sampling 160 levels TDMA FRAME=4.615ms

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 4 / 94

HSCDS: High Speed Circuit Switched Data

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 6 / 94

Industry Transformation and Convergence [from Ericsson 2006, LZT

WCDMA (Wideband CDMA) is a 3G mobile comm system. It is a

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 7 / 94

)SSS distributes the transmitted energy over a wide bandwidth

) SNIR at the receiver input is LOW.

Nevertheless, the receiver is capable of operating successfully because

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 9 / 94

(a) SSS: (b) CDMA (K users):

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 10 / 94

The PN signal b (t ) is a function of a PN sequence of 1’s fα[n ]g

I The sequences fα[n ]g must agreed upon in advance by Tx and Rx and

I This implies that :

I If fα[n ]g (i.e. “password”) is purely random, with no mathematical

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 12 / 94

Modelling of b(t) in SSS

b (t ) = ∑ α[n].c (t nTc ) (2)

where fα[n]g is a sequence of 1’s;

b (t ) = ∑ exp fj (2πk[n]F1 t + φ[n])g .c (t nTc ) (3)

where fk[n ]g is a sequence of integers such that

fα[n]g 7! fk[n]g (4)

and fα[n]g is a sequence of 1’s;

Applications of Spread Spectrum Techniques

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 14 / 94

Jamming source, or, simply Jammer is de…ned as follows:

Jammer , intentional (hostile) interference

Multiple Access Interference (MAI) is de…ned as follows:

MAI , uniintentional interference

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 16 / 94

PG: is a measure of the interference rejection capabilities

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 17 / 94

F With area-B = area-A we can …nd Nj

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 18 / 94

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 19 / 94

SS Transmission in the presence of a Jammer (or MAI)

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 20 / 94

SS Reception in the presence of a Jammer (or MAI)

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 21 / 94

PN-codes (or PN-sequences, or spreading codes) are sequences of

Property-1 easy to generate

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 22 / 94

Comments on PN-sequences Main Properties

Comments on Properties 1, 2 & 3

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 23 / 94

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 24 / 94

I The …gure below shows a shift register of 5 stages together with a

Modulo-2 adder Modulo-2 adder

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 25 / 94

fα1 [k ]g, fα2 [k ]g, ...., fαK [k ]g (11)

then if these code sequences are not totally uncorrelated, there is

I Therefore it is desired that this cross-correlation is made as small as

Prof. A. Manikas (Imperial College) EE401: Principles of CDMA 28 Oct 2011 26 / 94

There is a trade-o¤ between Properties-4 and 5.