CDMA: An Introduction

Βαγγέλης Δούρος
EY0619
douros06@aueb.gr

1 26/11/2007
 Motivation

z Why do we need CDMA?

z Which are its basic principles?
z What do we earn by using it?

2
 Outline

z Introduction
z CDMA Basics
z CDMA Deeper
z CDMA Problems
z CDMA Benefits
z Conclusions

3
 Why Multiple Access?

z Goal: Multiple use of a

shared medium.
z Multiplex channels in
three dimensions:
– Time (t)
– Frequency (f)
– Code (c)

4
 Frequency Division Multiple Access
(FDMA)

z Cocktail Party Analogy: People

create teams and discuss. There
is a distance among them.
z (-) Requires guard band
between channels
z (-) Waste of bandwidth if traffic is
distributed unevenly
z Example: broadcast radio

5
 Time Division Multiple Access (TDMA)

z Cocktail Party Analogy:

People have access to the
same room but each of
them waits for his turn to
speak.
z (-) Precise synchronization
necessary

6
 Time and Frequency Division Multiple
Access

z (-) Precise coordination

required
z Example: GSM

7
 Code Division Multiple Access (CDMA)

z Cocktail Party Analogy:

All people are in the same
room together. They can
all be talking the same
time!
z Example: UMTS

8
 Spread Spectrum

z PN: Pseudo-Noise code sequence Æ spread/

despread the signal.
z Modulation:
– FSK ÆFrequency hopped (FH) multiple access)
– PSK ÆDirect sequence (DS) multiple access)

D mD

9
 FHMA

z Bandwidth divides in non overlapping bands.

z Signal shifts from band to band in predefined
intervals.
z Receiver is synchronized with the transmitter
z (+) less interference
z Use: Bluetooth

10
 CDMA Basics (1)

z Each station is assigned a unique m-bit code (chip sequence)

z To send bit 1, station sends chip sequence.
z To send bit 0, station sends the complement.
z Example: 1 MHz band with 100 stations.
z FDMA
– Each station a 10kHz band
– Rate:10 kbps (Assume that you can send 1bit/Hz)
z CDMA
– Each station uses the whole 1MHz bandÆ106 cps.
– If <100 chips per bitÆ Rate >10 kbps

11
 CDMA Basics (2)

z Let or

z Compare any pair of these sequences-

vectors…
z Multiply any pair of these sequences-
vectors…
z Two chips S,T are orthogonal IFF S • T = 0
1 m −
z S • T = ∑ SiTi = 0 S •T = 0 ⇔ S •T = 0
m i =1
m m m m m
1 1 1
S • S = ∑ S S = ∑ S = ∑ (±1)
_ _
2 2
=1 S • S = ∑ Si S i = ∑ (−1) = 1 ( − m) = −1
1 1
12 m
i =1
i
m
i
mi =1
i
i =1 m i =1 m i =1 m
 CDMA Basics (3)
z Let A, B, C, D transmit correspondingly bit 1,0,1,_.

z Assume that:
– All stations are perfectly synchronous.
– All codes are pair wise orthogonal (aren’t they?).
– If two or more stations transmit simultaneously, the bipolar signals
add up linearly.
_
z Receiver “understands” S = A + B + C = (−1 + 1 − 3 + 3 + 1 − 1 − 1 + 1)
z How can the receiver “understand” what bit station C send?

13
 CDMA Basics (4)

z Let’s compute the normalized inner product S•C

1
S • C = (−1 +1 -3 +3 +1 -1 -1 +1) ⋅ (-1 +1 -1 +1 +1 +1 -1 -1)=
8
1 1
= (1 + 1 + 3 + 3 + 1 − 1 + 1 − 1) = 8 = 1
8 8

z Right!
z By accident??
− −
S • C = ( A + B + C ) • C = A • C + B• C + C • C = 0 + 0 + 1 = 1

z Remember: All codes are pair wise orthogonal!

14
 CDMA Basics (5)

z Reverse way…
z Think that each chip sequence arrives separately
z Receiver separately computes each inner product
A • C = 0 (1)
−
B• C = 0 (2)
C • C = 1 (3)
S = 0 + 0 +1 = 1
z It keeps only the non-orthogonal pair, i.e. the right bit

15
 CDMA Deeper (1)

z More advanced analysis…

z Sender:
sd (t ) = A cos(2π f c t + θ (t ))
θ (t ) ∈ {0, +π }
c(t ) ∈ {−1, +1}
s (t ) = A cos(2π f c t + θ (t ))c(t )

z Receiver:
s (t )c(t ) = A cos(2π f c t + θ (t ))c(t )c(t ) =
= A cos(2π f c t + θ (t )) = sd (t )

16
 CDMA Deeper (2)

z With orthogonal codes, we can safely decode the

coding signals.
z Noise?
z R’=R+N, N: m-digit noise vector and N=(±a ±a…±a)
z Decode … R' • S=(R+N) • S=S • S +(orthogonal codes) • S +N • S=1+0+?
z No problem if chipping codes are balanced (same ±)
R' • S=(R+N) • S=S • S +(orthogonal codes) • S +N • S=
=1+0+(a a a a)(+1+1-1-1)=1+0+0=1

17
 CDMA Deeper (3)

z How many codes can we construct with m chips?

z m (why?)
z If m=2k, Walsh-Hadamard codes can be
constructed recursively!
z The set of codes of length 1 is C = {< + >} 0
_

z For each code < c >∈ Ck we have two codes {< c c> <c c >} ∈ C k+1

z Code Tree:
C0 = {< + >}
z C1 = {< + + >, < + − >}
C2 = {< + + + + >, < + + - - >, < + - + - >, < + - - + >}

18
 CDMA Deeper (4)

z Correlation: Determines similarity between two sets of data.

– Possible values
z 1 sequences are similar
z 0 no relationship between them.
z -1 one is the mirror of the other
z Cross correlation: Compare two sequences from different
sources
z Auto correlation: Compare a sequence with itself after a time-
interval
z Walsh Codes: No cross correlation – Low auto correlation
z PN sequences: Low cross correlation – Low auto Correlation

19
 CDMA Deeper (5)

z We cannot have more than m orthogonal codes.

z Let m + k stations and m chips…
z Idea: Use PN Sequences.
R • S = S • S + (k random codes) • S+(m-1 orhogonal codes) • S =
z 1+?+0
z ?: the sum of the k random variables that are either 1 or -1.
z But PN Sequences = low cross correlation. ? should be 0.
z Experimental evaluation: For k=m=128, decoding is correct
more than 80%.

20
 CDMA Problems (1)
z All stations are received with the same power level…

z In reality… users may be received with very different powers!

z Near-far Problem
z Solutions:
– Empirical rule: Each MS transmits with the reverse power that it
receives from the BS
– Power Control!
z Open Loop
z Fast Closed Loop
21
 CDMA Problems (2)
z Bad Properties of Walsh Codes
z Perfect Synchronization of all
users required.
z Impossible…
z …is nothing! Use a long
enough known chip sequence.

z But…In a multipath channel,

delayed copies may be
received, which are not
orthogonal any longer.
z Self-Interference.

22
 CDMA Problems (3)

z So far…
z (-)tight synchronization
z (-)self-interference
z (-)Near-far problem
z (-)Higher complexity of sender/ receiver
z …
z How did Qualcomm convince people to use
this stuff??
23
 CDMA Benefits (1)

z Unlike FDMA and TDMA, CDMA does not rely on

orthogonal frequency and time slots!

24
 CDMA Benefits (2)

z In TDMA and FDMA systems

– Nothing to sendÆ time/frequency slot is wasted
– Dynamic allocation is very difficult
z In CDMA systems
– Nothing to sendÆ less interference
– Transmit ~half timesÆ doubles the capacity

25
 CDMA Benefits (3)

z FDMA-TDMA use sectors to decrease the reuse

distance
z CDMA use sectors to increase capacity (triple it)!

26
 CDMA Benefits (4)

z Why handoff?
z Types
– “Hard”
– “Soft”

27
 CDMA Benefits (5)

Hard Handoff
z Break-Before-Make
z Each MS
communicates with only
one BS each time
z (+)Reduced dropped
calls

28
 CDMA Benefits (6) Soft Handoff

z Each MS
communicates with
more than one BS each
time
z Use Signal Strength to
decide where to
connect.
z Make-Before-Break
z (+++) no dropped calls

29
 CDMA Benefits (7)

z Capacity…
z TDMA-FDMA: bandwidth limited
z CDMA: interference limited
z CDMA’s capacity is bigger.
z How?
z Long Story…

30
 Conclusions

z Back to the start…

z Why do we need CDMA?
– Introduction
z Which are its basic principles?
– CDMA Basics
– CDMA Deeper
z What do we earn by using it?
– CDMA Problems
– CDMA Benefits

31
 Ευχαριστώ!

32