CDMA Mobile Communication & IS-95

1
Outline

 Spread Spectrum Basics

 Spreading Codes
 IS-95 Features- Transmitter/Receiver
 Power Control
 Diversity Techniques
 RAKE Receiver
 Soft Handoff

2
Spread Spectrum

 A technique in which the transmission

bandwidth W and message bandwidth R are
related as

W >> R

 Counter intuitive

 Achieves several desirable objectives for e.g.

enhanced capacity
3
Application of Spread Spectrum Systems

 Antijamming
 Multiple access
 Low detectability
 Message Privacy
 Selective calling
 Identification
 Navigation
 Multipath protection
 Low radiated flux density

4
Types of Spread Spectrum Systems

 Frequency Hopping

 Direct Sequence

 Frequency Hopping
– Slow Frequency Hopping - multiple symbols per hop
– Fast Frequency Hopping - multiple hops per symbol

 Care is taken to avoid or minimize collisions of

hops from different users
5
Frequency Hopping

Typical frequency-hopping waveform pattern

6
Direct Sequence

Transmitter side of system

7
Direct Sequence (contd...)

Receiver side of system

8
Code Division Multiple Access - CDMA

 Multiple users occupying the same band by having

different codes is known as a CDMA - Code Division
Multiple Access system
Let
W - spread bandwidth in Hz
R = 1/Tb = Date Rate (data signal bandwidth in Hz)
S - received power of the desired signal in W
J - received power for undesired signals like multiple
access users, multipath, jammers etc in W
Eb - received energy per bit for the desired signal in W
N0 - equivalent noise spectral density in W/Hz
9
 CDMA (contd…)

J N 0W WT b W R
  
S E b Tb E b N 0 E b N 0

J  W R
  
 S max E b N 0 min

What is the tolerable interference over desired signal power?

J  W E b 
   Jamming margin (db)  (db )    (db )
 S max R  N 0 min

10
CDMA (contd…)

 In conventional systems W/R  1 which means, for

satisfactory operation J/S < 1

 Example Let R = 9600; W = 1.2288 MHz

(Eb/N0)min = 6 dB (values taken from IS-95)
Jamming margin (JM) = 10log10(1.2288*106/9.6*103) - 6
= 15.1 dB  32

 This antijam margin or JM arises from Processing Gain

(PG) = W/R = 128

 If (Eb/N0)min is further decreased or PG is increased, JM

can be further increased
11
CDMA (contd…)

 JM is a necessary but not a sufficient condition

for a spread spectrum system. For eg. FM is
not a spread spectrum system
 JM can be used to accommodate multiple users
in the same band
 If (Eb/N0)min and PG is fixed, number of users
is maximized if perfect power control is
employed.
 Capacity of a CDMA system is proportional to
PG.
12
Universal Frequency Reuse

 Objective of a Wireless Communication System

– Deliver desired signal to a designated receiver
– Minimize the interference that it receives

 One way is to use disjoint slots in frequency or time in

the same cell as well as adjacent cells - Limited
frequency reuse

 In spread spectrum, universal frequency reuse applies

not only to users in the same cell but also in all other
cells

 No frequency plan revision as more cells are added

13
Universal Frequency Reuse (contd...)

 As traffic grows and cells sizes decrease,

transmitted power levels in both directions can
be reduced significantly

 Resource allocation of each user’s channel is

energy (instead of time and frequency)

 Hence interference control and channel

allocations merge into a single approach
14
Spreading Codes

 It is desired that each user’s transmitted signal

appears noise like and random. Strictly speaking,
the signals should appear as Gaussian noise
 Such signals must be constructed from a finite
number of randomly preselected stored
parameters; to be realizable
 The same signal must be generated at the
receiver in perfect synchronization
 We limit complexity by specifying only one bit per
sample i.e. a binary sequence
15
Desirable Randomness Properties

 Relative frequencies of “0” and “1” should be ½

(Balance property)

 Run lengths of zeros and ones should be (Run

property):
– Half of all run lengths should be unity
– One - quarter should be of length two
– One - eighth should be of length three
– A fraction 1/2n of all run lengths should be of length n
for all finite n
16
Desirable Randomness Properties
(contd…)

 If the random sequence is shifted by any nonzero

number of elements, the resulting sequence
should have an equal number of agreements and
disagreements with the original sequence
(Autocorrelation property)

17
PN Sequences

 A deterministically generated sequence that nearly

satisfies these properties is referred to as a
Pseudorandom Sequence (PN)

 Periodic binary sequences can be conveniently

generated using linear feedback shift registers
(LFSR)

 If the number of stages in the LFSR is r, P  2r - 1

where P is the period of the sequence
18
PN Sequences (contd…)

 However, if the feedback connections satisfy a

specific property, P = 2r - 1. Then the sequence
is called a Maximal Length Shift Register
(MLSR) or a PN sequence.

 Thus if r=15, P=32767.

19
Randomness Properties of PN Sequences

 Balance property - Of the 2r - 1 terms, 2r-1 are

one and 2r-1–1 are zero. Thus the unbalance is
1/P. For r=50; 1/P10-15
 Run property - Relative frequency of run length
n (zero or ones) is 1/ 2n for n  r-1 and 1/(2r - 1)
for n = r
 One run length each of r-1 zeros and r ones
occurs. There are no run lengths for n > r
 Autocorrelation property - The number of
disagreements exceeds the number of
agreements by unity. Thus again the
discrepancy is 1/p
20
Randomness Properties of PN Sequences
(contd.)

Autocorrelation function

21
Randomness Properties of PN Sequences
(contd…)

Power Spectral Density

22
SR Implementation of PN Sequences

 The feedback connection should correspond to a

primitive polynomial.
 Primitive polynomials of every degree exist. The number
of primitive polynomials of degree r is given by :

2r  1 J Pi  1 J
N  
r i 1 Pi
where 2  1   Pi
r

i 1
ei

 Simple Shift Register Generator (SSRG) - Fibonacci

configuration.
 Modular Shift Register Generator (MSRG) - Galois
configuration.
23
 SR Implementation of PN Sequences

SSRG configuration of f(x)=1 + c1x + c2x2 + ….. + cixi + … + cn-1xn-1 + xn

MSRG configuration of f(x)=1 + c1x + c2x2 + ….. + cixi + … + cn-1xn-1 + xn

24
PN Sequences Specified in IS-95

 A “long” PN sequence (r =42) is used to

scramble the user data with a different code
shift for each user

 The 42-degree characteristic polynomial is

given by:
– x42+x41+x40+x39+x37+x36+x35+x32+x26+x25+x24+x23+x21
+x20+x17+x16+x15+x11+x9+x7+1

 The period of the long code is 242 - 1  4.4*102

chips and lasts over 41 days
25
PN Sequences Specified in IS-95
(contd…)

 Two “short” PN sequences (r=15) are used to

spread the quadrature components of the
forward and reverse link waveforms

 The characteristic polynomials are given by :

– x15+x10+x8+x7+x6+x2+x (I-channel)
– x15+x12+x11+x10+x9+x5+x4+x3+1 (Q-channel)

 The period of the short code is:

215 - 1 = 32767 chips  80/3 ms
26
Orthogonal Spreading Codes – Walsh
Codes

Walsh functions of order N are defined as a set of N time

functions denoted as {Wj(t); t(0,T), j=0,1,…N-1} such that:
 Wj(t) takes on the values {+1, -1} except at the jumps,
where it takes the value zero
 Wj(t) = 1 for all j
 Wj(t) has precisely j sign changes in the interval (0,T)

0 if j  k
0 W j t Wk t dt  T if j  k
T

 Each Wj(t) is either even or odd with respect to T/2 i.e.

the mid point
27
Walsh Functions

The Walsh Functions of order 8

28
Walsh Functions (contd.)

+1  “0”
-1  “1”

The Walsh Sequence of Order 8

29
Walsh Functions (contd.)

 A set of Walsh functions of order N = 2K possess

symmetry properties (even or odd) about K axes
at T/2, T/22, …., T/2K

 Consider the 13th Walsh function of order N = 24

= 16
W13 = 0101101010100101
– The sequence has odd symmetry about T/24 = T/16
– The sequence has odd symmetry about T/8
– The sequence has even symmetry about T/4
– The sequence has odd symmetry about T/2
30
 Walsh Functions (contd.)

 The above symmetry properties can be generalized

 For e.g. 13 in binary notation can be written as:
(1101) = (j1 j2 j3 j4)
– j1 = 1  symmetry is odd at axis T/16
– j2 = 1  symmetry is odd at axis T/8
– j3 = 0  symmetry is even at axis T/4
– j4 = 1  symmetry is odd at axis T/2

 The sequence may now be written down, starting

with 0, according to the above symmetry properties
as :
0101101010100101 31
 Walsh Functions on the Forward Link

 IS-95 forward link uses orthogonal multiplexing of

the pilot, sync, paging and traffic channels by
exploiting the orthogonality of the set of Walsh
functions of order 64.

32
Walsh Functions on the Forward Link
(contd.)

Example of Walsh Function

orthogonal multiplexing, N=8

33
Walsh Functions on the Forward Link
(contd…)

Total multiplexed signal for the N=8

34
 Walsh Functions on the Forward Link
(contd…)

Multiplying Stot(t) by different Walsh functions for channel information recovery

35
Walsh Functions on the Forward Link
(contd…)

 It is essential that there is perfect

synchronization at the receiver, for the
orthogonal multiplexing system to work.

 Hence in IS-95 they are resynchronized at

every even second of time.

36
IS-95 CDMA

 Direct Sequence Spread Spectrum Signaling on

Reverse and Forward Links
 Each channel occupies 1.25 MHz
Reverse CH Forward CH

45 MHz
847.74 MHz 892.74 MHz
 Fixed chip rate 1.2288 Mcps
37
Spreading Codes in IS-95

 Orthogonal Walsh Codes

– To separate channels from one another on forward
link
– Used for 64-ary orthogonal modulation on reverse
link.
 PN Codes
– Decimated version of long PN codes for
scrambling on forward link
– Long PN codes to identify users on reverse link
– Short PN codes have different code phases for
different base stations

38
Forward Link Modulation

Power Control Wi
bits

M
Convolutional Block 19.2 kbps U
Forward Encoder & Interleaver + X +
Traffic repetition
Channel
9.6 kbps
4.8 kbps
2.4 kbps I-PN
Long Code Decimator
1.2 kbps
Generator
x Seq

x
Q-PN Seq
39
Forward Link Modulation (contd…)

I-PN Seq
Pilot channel W0
x
all 0s (no data) x Q-PN Seq
x

I-PN Seq
Sync channel W3
2 x
Convolutional 4.8 kbps
Block x
Encoder &
1.2 kbps Repetitor Interleaver
x

Q-PN Seq 40
Forward Link Modulation (contd…)

Paging Channel I-PN Seq

19.2 ksps x
Convolutional
Block
Encoder & +
Interleaver
9.6 kbps Repetitor
4.8 kbps x

Long PN Q-PN Seq

code Decimator

1.2288 Mcps

41
Reverse Link Modulation

 The signal is spread by the short PN code

modulation (since it is clocked at the same
rate)

 Zero offset code phases of the short PN code

are used for all mobiles

 The long code PN sequence has a user

distinct phase offset.
42
Reverse Link Modulation

42 stage 1.2288 Mcps

Access Channel Long PN code
28.8
4.8 Convolutional (64,6) 307.2 kcps
ksps Block 28.8 x
kbps Coder & Walsh
Repetitor Interleaver ksps
modulator

I-PN code n=15 Cos 2πfct

1.2288 Mcps
x Filter x

+
Delay
x Filter
½ chip x
1.2288 Mcps Sin 2πfct
Q- PN code n=15 43
 Traffic Channel

28.8 28.8
Convolutional ksps ksps W(64.6)
Encoder & Block
Interleaver Walsh
Repetitor 307 kcps
9.6 kbps Modulator
4.8 kbps
2.4 kbps
1.2 kbps x X
Data 1.2288 Filter
Burst + Mcps I-PN code cos 2πfct
Randomizer +
½ chip
frame x delay Filter x
Long PN
data rate code
Q-PN code sin 2πfct

44
Power Control in CDMA

 CDMA goal is to maximize the number of

simultaneous users
 Capacity is maximized by maintaining the
signal to interference ratio at the minimum
acceptable
 Power transmitted by mobile station must be
therefore controlled
• Transmit power enough to achieve target
BER: no less no more
45
Two factors important for power
control
 Propagation loss
• due to propagation loss, power variations up to
80 dB
• a high dynamic range of power control required

 Channel Fading
• average rate of fade is one fade per second per
mile hour of mobile speed
• power attenuated by more than 30 dB
• power control must track the fade
46
Power Control on Forward Link and
Reverse Link

 On Forward Link
• to send just enough power to reach users at the
cell edge

 On Reverse Link
• to overcome the ‘near-far’ problem in DS-CDMA

47
Types of Power Control

 Open Loop Power Control (on FL)

• Channel state on the FL estimated by the mobile
– measuring the signal strength of the pilot channel
• RL transmit power made inversely proportional
to FL power measured
• Mobile Power = Constant – Received power
(dBm) (dBm) (dBm)
• Works well if FL and RL are highly correlated
– slowly varying distance and propagation losses
– not true for fast Rayleigh Fading.

48
Closed Loop Power Control (on RL)

 Measurement of signal strength on FL as a rough

estimate
 Base station measures the received power on RL
 Measured signal strength compared with the
target Eb/No (power control threshold)
 Power control command is generated
• asking mobile to increase/decrease
 Must be done at fast enough a rate (approx 10
times the max Doppler spread) to track multi-path
fading
49
Outer Loop Power Control

 Frame error rate (FER)is measured

 Power control threshold is adjusted at the

base station

50
Power Control in IS-95A

 At 900 MHz and 120 km/hr mobile speed

Doppler shift =100Hz
 In IS 95-A closed loop power control is
operated at 800 Hz update rate
 Power control bits are inserted (‘punctured’)
into the interleaved and encoded traffic data
stream
 Power control step size is +/- 1 dB
 Power control bit errors do not affect
performance much
51
Diversity Techniques in CDMA

Rationale for Diversity:-

if ‘p’ is the probability that a given path in a

multi-path environment is below a detection
threshold, then the probability is ‘pL’ that all ‘L’
paths in an L-path multi-path situation are
below the threshold

52
Diversity Techniques

 Frequency Diversity
– transmission of signal on two frequencies spaced further apart
than the coherence bandwidth
– inherent in spread spectrum system if the chip rate is greater
than the coherence bandwidth
 Time Diversity
– transmission of data at different times
– repeating the data ‘n’ times
– interleaving and error correcting codes used in IS-95
 Space Diversity
– Multi-path tracking (Path Diversity)
– Transmission space diversity
• Signal can be emitted from multiple antennas at a single
cell site 53
Diversity Combining

 Selection Diversity (SD)

 Equal Gain Diversity (EGC)
 Maximal Ratio Combining (MRC)
 MRC is an optimal form of diversity
 RAKE receiver in IS-95 is a form of MRC

54
Selection Diversity Combining

Diversity Receiver
z1
Ch #1 #1

Diversity
z2
Receiver
Ch #2 User data
#2
…..

…..
Receiver
zL
Diversity
Ch #L #L

• Channel with the highest SNR is chosen

• (L-1) channel outputs are ignored
55
 Equal Gain Combining (EGC)
n1(t)
Diversity Receiver z1
Ch #1
+ #1
n2(t) Combiner
Transmitted Receiver z2
Diversity + #2
Signal
Ch #2 L

nL(t)
 Z
l 1

Diversity Receiver zL
+ #L
Ch #L

• Symbol decision statistics are combined with equal gains

to obtain overall decision statistics. 56
Maximal Ratio Combining(MRC)
 Similar to EGC – decision statistics are summed or
combined
 In EGC – each channel is multiplied by equal gain
 In MRC – each channel is multiplied by gain
proportional to the square root of SNR of the channel
g i  SNR i
 This gives optimal combining
L
Output SNR   ( SNR)
i 1
i

 Requires knowledge of SNR of each channel as well

as phase of the diversity signal 57
MRC

n1(t)
g1
Diversity r1
x
Ch (α1Φ1) +
n2(t)
g2 Combiner
r2 User
Diversity x
+ L Data
Ch (α2Φ2)
nL(t) 
gL l 1
Diversity rL
+ x
Ch (αLΦL)

58
RAKE Receiver Concept
 Multi-path diversity channels
 Problem
– to isolate various multi-path signals
– How to do this ?
 If the maximal delay spread (due to multi-path) is Tm
seconds and if the chip rate 1  W  1
Tc Tm

then individual multi-path signal components can be

isolated
 Amplitudes and phases of the multi-path components
are found by correlating the received waveform with
delayed versions of the signal
 Multi-path with delays less than 1/Tc can’t be resolved
59
RAKE Receiver Concept

m(t )  C (t ) cos( w0 t )
c (t ) is a PN Sequence
 E{c(t ) cos( w 0 t )c(t   ) cos( w 0 t   )}
Rm ( )  E{m(t )m(t   )}

 E{c(t )c(t   )}E{cos w 0 t cos( w 0 t   )}

1 Rc ( )
 R c ( ) cos( w 0  )
2 Rc (0)
| |
R c ( )  R c (0)[1  ] |  |T c
Tc
1
w -Tc Tc
Tc 60
Rake Receiver in IS-95

 Rake Receiver is used in Mobile receiver for

combining
– Multi-path components
– Signal from different base stations (resolve multi-path signals
and different base station signals)
 3 Parallel Demodulator (RAKE Fingers)
– For tracking and isolating particular multi-path components (up
to 3 different multi-path signals on FL)
 1 Searcher
– Searches and estimates signal strength of
• multi-path pilot signals from same cell site
• pilot signals from other cell sites
– Does hypothesis testing and provides coarse timing estimation

61
 Rake Receiver (contd…)
–Search receiver indicates where in time the strongest replicas
of the signal can be found

Rake on FL

Searcher 3-Parallel Diversity

Receiver Demod- Combiner
ulator

(Mobile Station Rake Receiver)

62
Handoff in CDMA System

 Soft Handoff
– Mobile commences Communication with a new BS
without interrupting communication with old BS
– same frequency assignment between old and new
BS
– provides different site selection diversity
 Softer Handoff
– Handoff between sectors in a cell
 CDMA to CDMA hard handoff
– Mobile transmits between two base stations with
different frequency assignment

63
Soft Handoff- A unique feature of
CDMA Mobile
Advantages
 Contact with new base station is made before the call
is switched
 Diversity combining is used between multiple cell
sites
– additional resistance to fading
 If the new cell is loaded to capacity, handoff can still
be performed for a small increase in BER
 Neither the mobile nor the base station is required to
change frequency

64
Soft Handoff Architecture

MSC
R

BSC BSC
R old link

BTS BTS R BTS BTS

R- handoff request energy measurements

sent to the old cell are made at the mobile
65
Rate Receiver on Reverse Link

 Base station receiver uses two antennas for

space diversity reception

 4 – parallel demodulators

 Since no pilot signal is present, non coherent

maximal ratio combining

66
Rate Receiver on RL (contd…)

Path 1 Path 2 Path 3 Path 4

Non coherent MRC
Soft
decoder
67
Rake Receiver on Forward Link

Direct Reflection
path

Optimal Coherent
Combining 68
Base station Diversity on Reverse Link
during soft handoff

MSC

Cell site 1 Cell site 2

Non coherent MRC Non coherent MRC
Hard decision Hard decision

antenna 1 antenna 2 antenna 1 antenna 2

69
Eb/Io
Base A
margin exceeds

T_ADD
Base B

T_DROP

B_Active Time
B added to candidate list Drop timer Drop timer
starts resets
Signal levels during Handoff Drop timer
expires