Chapter 6: Multiple Access Techniques

Overview
 Interference modeling
 Multiplexing

2
Several Transmitters?
 What happens when there are
several transmitter/receiver pairs?
 Transmissions of “wrong” transmitter
disturb receiver! Interference
 Effects of interference are comparable to narrowband noise
 Extend signal to noise ratio to signal to noise & interference ratio
(SINR)
 In dB:

 SINR can be (approximately) used in SNR! BER formulas

! Receiver should only hear single transmitter
! Wireless medium needs to be multiplexed

3
Overview
 Interference modeling
 Multiplexing
 4 dimensions: Spatial, time, & frequency multiplexing
 Time division multiplexing
 Frequency division multiplexing
 Time and frequency multiplexing
 Spread spectrum techniques
 Code division multiplexing

4
Multiplexing
 Multiplexing: Allowing many (mobile) users to share a
given resources
 For high quality communications, this must be done without
severe degradation in the performance of the system
 Focus is for Wireless Communication

5
Multiplexing
channels ki
 Multiplexing in 4 dimensions
k1 k2 k3 k4 k5 k6
 Space (si)
 Time (t) c
 Frequency (f) t c
 Code (c) t
s1
 Goal: multiple use f
s2
of a shared medium f
c
t
 Important: Guard spaces needed!
 In all dimensions s3
f
 Types
 Static vs. dynamic multiplexing

6
Multiplexing – Time, Frequency, and Code

7
Multiplexing – Spatial

8
Multiplexing – Spatial …

9
Multiplexing – Wireless Systems

Cellular Systems MA Technique

AMPS (Advanced Mobile Phone system) FDMA / FDD
GSM (Global System for Mobile) TDMA / FDD
US DC (U. S Digital Cellular) TDMA / FDD
JDC (Japanese Digital Cellular) TDMA / FDD
DECT ( Digital European Cordless Telephone ) FDMA / FDD
IS – 95 (U.S Narrowband Spread Spectrum ) CDMA / FDD
W-CDMA (3GPP) CDMA/FDD
CDMA/TDD
Cdma2000(3GPP2) CDMA/FDD
CDMA/TDD

10
Overview
 Interference modeling
 Multiplexing
 4 dimensions: Spatial, time, & frequency multiplexing
 Time division multiplexing
 Frequency division multiplexing
 Time and frequency multiplexing
 Spread spectrum techniques
 Code division multiplexing

11
Time Division Multiplexing (TDM)
 A channel gets the whole spectrum for a certain amount of
time
 Advantages:
 Only one carrier in the
medium at any time k1 k2 k3 k4 k5 k6

 Throughput high
at high utilization c
f
 Disadvantages:
 Precise
synchronization
necessary
 Long delays
t
at low utilization

12
Time Division Multiplexing …
 A single carrier frequency is shared with several users
 Each user makes use of non-overlapping timeslots
 In each time slot, only one user is allowed to either transmit
or receive
 I.e., transmission for any user is non-continuous
 Synchronous TDM
 Each user occupies a cyclically repeating time slots and static
allocation
 Asynchronous TDM
 Allow different number of time slots for separate user and statistical
multiplexing

13
Components of TDM Frame
 Preamble: Address and synchronization information
 In cellular system, for base station and subscriber identification
 Guard times: Synchronization of receivers between
different slots and frames

14
Overview
 Interference modeling
 Multiplexing
 4 dimensions: Spatial, time, & frequency multiplexing
 Time division multiplexing
 Frequency division multiplexing
 Time and frequency multiplexing
 Spread spectrum techniques
 Code division multiplexing

15
Frequency Division Multiplexing (FDM)
 Separation of the whole spectrum into smaller frequency
bands
 A user gets a certain band of the spectrum for the whole
time
 Advantages: k1 k2 k3 k4 k5 k6

 No dynamic coordination c
necessary f
 Works also for analog signals
 Disadvantages:
 Waste of bandwidth
if traffic is unevenly
distributed
 Inflexible t
 Guard spaces

16
Frequency Division Multiplexing …
 Frequency spectrum is divided into unique frequency
bands or channels
 These channels are assigned to users on demand
 Multiple users cannot share a channel

 Users are assigned a channel as a pair of frequencies

 Forward and reverse channels for Duplexing

 FDMA requires tight RF filtering to reduce adjacent channel

interference
 During the period of a call, no other user can share the
same frequency band
 If the FDM channel is not in use, then it sits idle and cannot
be used by other users to increase capacity

17
Frequency Division Multiplexing …
 Co- and adjacent channel interference!
 FDM avoids co-channel interference – direct interference
from another transmitter using the same band
 What about adjacent-channel interference?
 Interference from transmitters using a neighboring channel

 FDM requires filter at receiver to “strip off” received power

in frequencies outside the assigned channel
 Unfortunately, filters are not perfect
 Power from channels “close by” can leak in
! Needs big guard spaces or ability to deal with errors

18
Frequency Division Multiplexing – Properties
 The bandwidth of FDM channels is narrow
 About 30 KHz since it supports only one call/ carrier
 ISI is low since the symbol time is large compared to
average delay spread
 No equalization is required

 FDM systems are simple than TDM systems, but modern

DSP is changing this factor
 FDM systems have higher cost
 Cell site system due to single call/carrier
 Costly band pass filters to eliminate spurious radiation
 Duplexers in both T/R increase subscriber costs

19
Time and Frequency Multiplexing
 Combination of both methods
 A channel gets a certain frequency band for a certain
amount of time
 Example: GSM
 Advantages:
 Protection against frequency k1 k2 k3 k4 k5 k6
selective interference
 Higher data rates compared to c
code multiplex f

 But: precise
coordination
required
t

20
Overview
 Interference modeling
 Multiplexing
 4 dimensions: Spatial, time, & frequency multiplexing
 Time division multiplexing
 Frequency division multiplexing
 Time and frequency multiplexing
 Spread spectrum techniques
 Frequency hopping spread spectrum
 Direct sequence spread spectrum
 Code division multiplexing

21
Errors and Narrowband Transmissions
 Modulation uses as much bandwidth H to transmit a data
stream of rate R as required by Nyquist/Shannon theorem
 Actually, much more because of imperfections everywhere

 Problem: When channel undergoes time-selective fading (i.e.,

Tc << Ts), received signal becomes weak, error rate high

 Observation: Fading is not

uniform in frequency!
 While one frequency range
is bad, another might be ok!
 Exploit frequency selectivity
to overcome time selectivity
in a single frequency
! Frequency diversity

22
Spread Spectrum Techniques
 Idea: “Spread” signal to a wider bandwidth than is actually
necessary to avoid short, frequency-selective distortions
 One form of frequency diversity exploitation
 Can be channel fading, but also narrowband interference

power interference spread power signal

signal
spread
detection at interference
receiver

f f
 Main implementation options
1. Frequency Hopping
2. Direct Sequence (is the basis for CDMA)
 Common drawbacks: Strict time synchronization & power
control needed (see later)
23
Frequency Hopping Spread Spectrum (FHSS)
 Discrete changes of carrier frequency
 Sequence of frequency changes determined via pseudo random
number sequence
 Additional “hopping” increases the required bandwidth
 Two versions
 Fast Hopping: several frequency hops per user bit
 Slow Hopping: several user bits per hop
 Advantages
 Fading and interference limited to short period
 Simple implementation
 Uses only small portion of spectrum at any time
 Disadvantages
 Not as robust as Direct Sequence

24
Frequency Hopping Spread Spectrum …

tb

user data

0 1 0 1 1 t
f
td
f3 slow
f2 hopping
(3 bits/hop)
f1

td t
f

f3 fast
f2 hopping
(3 hops/bit)
f1

tb: bit period td: dwell time

25
Frequency Hopping Spread Spectrum …
narrowband spread
signal transmit
user data signal
modulator modulator

frequency hopping
synthesizer sequence
Transmitter

narrowband
received signal
signal data
demodulator demodulator

hopping frequency
sequence synthesizer
Receiver

26
Overview
 Interference modeling
 Multiplexing
 4 dimensions: Spatial, time, & frequency multiplexing
 Time division multiplexing
 Frequency division multiplexing
 Time and frequency multiplexing
 Spread spectrum techniques
 Frequency hopping spread spectrum
 Direct sequence spread spectrum
 Code division multiplexing

27
Direct Sequence Spread Spectrum (DSSS)
 XOR of the signal with pseudo-random number (also called
chipping sequence)
 Many chips per bit (e.g., 128) result in higher bandwidth of the
signal tb
 Chipping sequence also called
user data
code or key
0 1 XOR

 Advantage tc
chipping
 Reduces time-selective sequence
fading by frequency diversity 01101010110101 =
 Disadvantage resulting
 Precise power control signal
01101011001010
necessary
tb: bit period
tc: chip period

28
DSSS Transmitter/Receiver Structure

spread
spectrum transmit
user data signal signal
X modulator

chipping radio
sequence carrier

transmitter

correlator
lowpass sampled
received filtered products sums
signal signal data
demodulator X integrator decision

radio chipping
carrier sequence

receiver

29
DSSS – Transmission
 DSSS: XOR of a “long” symbol with a chipping
sequence/code
 DSSS can also be elegantly written as pointwise product of
vectors when using +/-1 instead of 0/1
 Assign: “1” = -1, “0” = +1
 Formally: Transmitted signal, AS(t) for 0  t  T,
 T symbol time
 Represent data by Ad(t) = constant over the duration of a given symbol
 Ak(t) is the chipping key 0  t  T

 Resulting signal has more level changes per time

 Higher required bandwidth !

30
DSSS Transmission – Example
 Example: Sender A sends Ad = 1, chipping key Ak =
010011
 Rewrite, only one value per chip is shown:
Ad = (-1, -1, -1, -1, -1, -1); Ak = (+1, -1, +1, +1, -1, -1)
 Compute transmitted signal As = Ad * Ak (pointwise product)
 AS = Ad * Ak = (-1, -1, -1, -1, -1, -1) * (+1, -1, +1, +1, -1, -1)
= (-1, +1, -1, -1, +1, +1)

31
DSSS Transmission on Signal Level – 2nd example
data A
1 0 1 Ad

key A
key
0 1 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0 0 Ak
sequence A
data  key 1 0 1 0 1 1 0 1 0 1 0 0 1 0 1 0 1 1

signal A
As

Remark: Real systems use much longer keys resulting in a

larger distance between single code words in code space

32
DSSS – Receiver
 How does a receiver convert spreaded chip sequence back
to bits?
 Receiver (correlation receiver or matched filter)
 Needs to know sender’s key Ak, symbol duration T
 Has received sequence AR from A, perfect phase synchronization
 Computes scalar product

 Example:
 AR = (-1, +1, -1, -1, +1, +1), Ak = (+1, -1, +1, +1, -1, -1)
 AR ¢ Ak = -1 + -1 + -1 + -1 + -1 + -1 = -6

33
DSSS at Receiver – Error-free Case
 How to use Ae to decide which bit Ad has been transmitted?
 In error-free case, AR = AS (up to attenuation, here treated
as constant)
 Then, scalar product will be

 Observation: Since Ak(t) = +/- 1, Ak(t)2 = 1 !

 Thus:

! Decision rule for receiver which data had been transmitted

34
DSSS at Receiver – Noise?
 Noise/interference can flip some chips
 When computing AR ¢ Ak, no longer +/- n will result!
 Example:
 AR = (-1, +1, +1, -1, -1, +1) <> AS !, Ak = (+1, -1, +1, +1, -1, -1)
 AR ¢ Ak = -1 + -1 + 1 + -1 + 1 + -1 = -2 > -6

 But: Results will be closer to zero

 Formally: Write AR(i) = AS(i) + N(i) + I(i), where
 N(i) is Gaussian noise, I(i) is narrowband distortion
 Compute
s0T AR(t) Ak(t) dt = s0T Ad(t) Ak(t)2 dt + s0T N(t) Ak(t) dt + s0T I(t) Ak(t) dt
 It holds: E[N ¢ Ak] = 0; similarly, E[I ¢ Ak] is small
 Note: data is multiplied with Ak(t)2, interference only with Ak(t) !

 Decision rule: Decide for a 0 if AR¢ Ak > 0; for a 1 if < 0;

report error if = 0
35
DSSS in multipath environment
 Consider a 2 path environment
 Received signal AR(t) consists of two components, one received
over the direct path A0(t) and one over a path with delay  A(t)
! AR(t) = A0(t) + A(t) = AS(t) + AS(t-) = Ad(t) Ak(t) + Ad(t - ) Ak(t-)
(neglecting path loss, noise)
 Receiver computes scalar product AR ¢ Ak:
 AR ¢ Ak = s0T Ad(t)Ak(t) Ak(t) + Ad(t-)Ak(t-) Ak(t)dt
= s0T Ad(t) 1 dt + s0T Ad(t-) Ak(t-) Ak(t) dt
= Ad(0) + Ad(0) s0T Ak(t-)Ak(t) dt since Ad(t)=const

 s0T Ak(t-)Ak(t) dt is autocorrelation R() of chipping

sequence
 Look for chipping sequences with small R() for  > Tthreshold
 E.g., Barker sequence

36
Summary: Spread Spectrum Technologies
 Basic idea: Combat time-selective fading by exploiting
frequency diversity
 “Even if the channel at some given frequency is currently bad, the
chance that it is ok at some other frequency is high”
 Therefore: spread the signal to use a larger frequency band

 Typical implementation: Direct sequence spread spectrum

 Works (essentially) because
 Signal is spreaded and despreaded (= multiplied with spreading
sequence twice)
 Interference only despreaded (=multiplied once)
 Handles noise, interference, multipath environment

37
Overview
 Interference modeling
 Multiplexing
 4 dimensions: Spatial, time, & frequency multiplexing
 Time division multiplexing
 Frequency division multiplexing
 Time and frequency multiplexing
 Spread spectrum techniques
 Frequency hopping spread spectrum
 Direct sequence spread spectrum
 Code division multiplexing

38
Code Division Multiplex (CDMA)
 Each channel has a unique code k1 k2 k3 k4 k5 k6

 All channels use the same spectrum c

at the same time
 Advantages:
 Bandwidth efficient
 Little coordination and synchronization f
necessary (expect code, power)
 Good protection against interference
and tapping
 Disadvantages:
 Lower user data rates t
 More complex signal regeneration
 Implemented using spread spectrum
technology
39
Spreading, FDM, and frequency selective fading
channel
quality

1 2 5 6
Narrowband channels (FDM)
3
4
frequency
narrow band guard space
signal

channel
quality
2
2
2 Spread spectrum
2
2 channels (CDM)
1

spread frequency
spectrum

40
How does CDMA work?
 How can a receiver sort out different transmissions, despite
overlapping in time and frequency?
 Transmissions need to differ in “code space”
 Different transmissions need to use different keys/chipping
sequences with certain properties

 Following example
 Senders A and B close to receiver
 Signals send in same frequency band, at same time
 A and B use different keys

 Key idea: Receiver uses key of desired transmitter to

compute scalar products

41
CDMA in theory – By example
 Sender A
 Sends Ad = 1, key Ak = 010011 (assign: „1“= -1, „0“= +1)
 Signal As = Ad ¢ Ak = (-1, +1, -1, -1, +1, +1)
 Sender B
 Sends Bd = 0, key Bk = 110101 (assign: „1“= -1, „0“= +1)
 Signal Bs = Bd ¢ Bk = (-1, -1, +1, -1, +1, -1)
 Both signals superimpose in space
 Other interference neglected (noise etc.)
 As + Bs = (-2, 0, 0, -2, +2, 0)
 Receiver wants to receive signal from sender A
 Apply code Ak bitwise (scalar product)
 Ae = (-2, 0, 0, -2, +2, 0) ¢ Ak = -2 + 0 + 0 + -2 + -2 + 0 = -6
 Result smaller than 0, therefore, original bit was „1“
 When receiving B, use B’s code
 Be = (-2, 0, 0, -2, +2, 0) ¢ Bk = +2 + 0 + 0 + 2 + 2 + 0 = +6, i.e. „0“

42
CDMA in theory – By equation
 Formally, receiver receives sum of AS and BS
 AS = Ad ¢ Ak, BS = Bd ¢ Bk
 We ignore noise and (additional) interference sources
 Receiver computes (AS + BS) ¢ Ak to receive Ad:
 Ad¢ Ak ¢ Ak + Bd ¢ Bk ¢ Ak = Ad + Bd ¢ Bk ¢ Ak
 Bk ¢ Ak should be small (Bk ¢ Ak = s0T Bk(t) Ak(t) dt )
 Ideally: scalar products of any two keys should be 0
 Then: each transmitter can be perfectly received
 So-called orthogonal keys
 If only close to 0: Quasi-orthogonal keys
 For chipping sequence of length G, there are G orthogonal
chipping sequences
! G transmitters can be perfectly separated!

43
Design of a chipping sequence family
 Choosing chipping sequences determines performance of
CDMA system
 Keys should have good autocorrelation properties
 To combat multipath fading
 Keys should have small or zero scalar products between
themselves
 To allow user separation with delayed copies from other users
 Generalization of the “small scalar products” requirement
 s 0 T Ak(t) Bk(t-) dt ¼ 0 for any keys Ak, Bk and any  > threshold

44
CDMA on signal level

signal A As

data B 1 0 0 Bd

key B
key 0 0 0 1 1 0 1 0 1 0 0 0 0 1 0 1 1 1 Bk
sequence B
1 1 1 0 0 1 1 0 1 0 0 0 0 1 0 1 1 1
data  key

Bs
signal B

As + B s

45
CDMA on signal level II
data A
1 0 1 Ad

As + Bs

Ak

(As + Bs)
* Ak

integrator
output
comparator 1 0 1
output

46
CDMA on signal level III
data B
1 0 0 Bd

A s + Bs

Bk

(As + Bs)
* Bk

integrator
output
comparator 1 0 0
output

47
CDMA on signal level IV

A s + Bs

wrong
key K

(As + Bs)
*K

integrator
output
comparator
output (0) (0) ?

48
Conclusion
 Wireless communication is made challenging by
 The time- and frequency varying nature of a wireless channel
 By interference between multiple transmitters
 By attenuation

 Consequence: Error rates can be high and highly

fluctuating
 There is no “link” between two nodes in a wireless communication
like there is in a simple graph model of a network

 Higher layer protocols will have to deal with this harsh

environment

49