Interleave Division Multiple Access

(IDMA)

Lihai Liu, Raymond Leung and Li Ping

Department of Electronic and Engineering
City University of Hong Kong

Outline

Introduction
Iterative detection
Performance evaluation
Multi-user gain in fading channels
Other applications
Conclusions

Outline

Introduction
Iterative detection
Performance evaluation
Multi-user gain in fading channels
Other applications
Conclusions

Features of IDMA

low receiver cost

de-centralized (i.e., asynchronous) control,
simple treatment of ISI,
cross-cell interference mitigation,
diversity against fading,
high power efficiency
high spectral efficiency
suitable for wide or narrow band transmission
flexible rate adaptation

MAC Capacity (per Dimension)

4

S p e c tr a l e f f ic ie n c y ( b its /c h ip )

3.5
3

Optimal
Multi-user detection
is necessary to get
into this range.

2.5
2
1.5
1

Matched Filter

0.5

0
-2

6
Eb/N0 (dB)

10

12

14
5

IDMA System Model

Transmitter for user-1
d1

ENC1

x1

...

h1

Transmitter for user-k

dk

ENCk

xk

hk

r = hk x k + n
k =1

...

hK

Transmitter for user-K

dK

ENCK

xK

..

Lihai Liu, Jun Tong, and Li Ping, "Analysis and Optimization of CDMA Systems with ChipLevel Interleavers," IEEE J. Select. Areas Commun. vol. 24, no. 1, pp. 141-150, Jan. 2006.

CDMA System Model

Transmitter for user-1
d1

s1

ENC1

x1

...

h1

Transmitter for user-1

dk

sk

ENCk

xk

hk

r = hk x k + n
k =1

...

hK

Transmitter for user-K

dK

ENCK

sK

xK

Comparison of IDMA and CDMA

Transmitter for user-1
d1

x1

ENC1

Transmitter for user-1

d1

Transmitter for user-1

ENCk

h1
xk

xK

IDMA

dk

Transmitter for user-1

sk

ENCk

h1
xk

Transmitter for user-K

dK

ENCK

hk
hK

...

...

ENCK

hk
hK

Transmitter for user-K

dK

x1

...

...

dk

s1

ENC1

sK

xK

CDMA
8

A Factor Graph for a LDPC Code

A Factor Graph for a CDMA System

User 1 information bits

User 2 information bits

10

A Factor Graph for an IDMA System

User 1 information bits

User 2 information bits

11

Outline

Introduction
Iterative detection
Performance evaluation
Multi-user gain in fading channels
Other applications
Conclusions

12

IDMA System Model

user-1
d1

s1

ENC1

x1

...

h1

user-k
dk

sk

ENCk

xk

hk

r = hk x k + n
k =1

...

hK

user-K
dK

ENCK

sK

xK

13

Message Passing at Channel Nodes

User 1 information bits

User 2 information bits

14

Gaussian Approximation Detection

Path model and Gaussian approximation
K

r ( j ) = hk xk ( j ) + n( j )
k =1

= hk xk ( j )

Estimation:

e ( xk ( j ) ) =

k ( j)

Gaussian

2hk
( r ( j ) E( k ( j )) )
Var( k ( j ))

Some details:

(r ( j ) E( k ( j )) hk )2
exp(
)
Pr( xk ( j ) = +1)
2Var( k ( j ))
2hk
log
= log
=
( r ( j ) E( k ( j )) )
(r ( j ) E( k ( j )) + hk )2
Pr( xk ( j ) = 1)
Var( k ( j ))
exp(
)
2Var( k ( j ))

15

Chip-by-Chip (CBC) Detection Algorithm

K

Step 1. E ( r ( j ) ) = hk E ( xk ( j ) )

Var ( r ( j ) ) = hk Var ( xk ( j ) )

Step 2. E(k ( j)) = E( r( j)) hkE( xk ( j))

Var ( k ( j ) ) = Var ( r ( j ) ) hk Var ( xk ( j ) )

k =1

Step 3. e ( xk ( j ) ) =

k =1

2hk
( r ( j ) E( k ( j )) )
Var( k ( j ))

Notes:
(1) There is no matrix operation.
(2) E(xk(j) and Var(xk(j)) are the feedback from the decoders.
16

A Factor Graph for an IDMA System

User 1 information bits

User 2 information bits

17

Chip-by-Chip Multiuser Detection

1
1

r={r(j)} Chip-by-Chip
Processing

APP
DEC-1

APP
DEC-k

18

Complexity
6 additions and 6 multiplications per chip per iteration per
user.
Complexity (per user) is independent of user number K .
Comparison: To achieve good performance, the cost for MMSE
CDMA multi-user detection is O(K2) due to matrix operations.

19

Un-coded IDMA
Rate-1/8 repetition coding
1.E+00
8 users

1.E-01

64 users

BER

1.E-02
single-user

1.E-03
1.E-04
1.E-05
0

10 12 14 16 18 20 22 24

Average Eb/N 0 (dB)

20

Convolutional-Repeat Coded IDMA

Rate and rate repetition coding. Overall rate =1/8.
1.E+00
CDMA
6 users
matched filter

1.E-01
(b)
(a)

BER

1.E-02
1.E-03
IDMA
8 users

1.E-04

IDMA
16 users

IDMA
32 users

IDMA
64 users

1.E-05
0

capacities

10

12

14

16

18

20

22

Average Eb/N0 (dB)

21

Outline

Introduction
Iterative detection
Performance evaluation
Multi-user gain in fading channels
Other applications
Conclusions

22

SNR Evolution for an IDMA Detector

ESE

{E(x1(j))}

{e(xk(j))}
{E(xk(j))}

k1

APP
DEC-1

r={r(j)}

11

Iterative
detector

{e(x1(j))}

APP
DEC-k

SNR1
r={r(j)}

ESE

SNRk
Variancek

f()

Evolution
process

Variance1

f()

23

IDMA Performance Evaluation

Received signal:

r ( j ) = pk hk xk ( j ) + n( j )

Gaussian

k =1

=
SNR evolution:

SNRk( new)

pk hk xk ( j ) + pi hi xi ( j ) + n( j )
ik

pk | hk |2
=
pi | hi |2 fi SNRi(old ) + 2
ik

Initialization:

SNR k( 0 ) = 0, k

24

SNR Evolution for an IDMA Decoder

The iterative detector can be characterized by the following SNR
evolution process:
Evolution
formula

( new )
k

SNR

| hk |2 pk
=
| hi |2 pi fi (SNRi(old ) ) + 2
ik

SNR1

ESE

SNRk
Variancek

k1

f()

This is much simpler and faster than simulation.

f()

r={r(j)}

Evolution
process

Variance1

11

25

Un-coded IDMA
Rate-1/8 repetition coding
1.E+00
8 users

1.E-01

64 users

BER

1.E-02
single-user

1.E-03
1.E-04
1.E-05
0

10 12 14 16 18 20 22 24

Average Eb/N0 (dB)

26

Outline

Introduction
Iterative detection
Performance evaluation
Multi-user gain in fading channels
Other applications
Conclusions

27

Multi-User Gain in Fading Channels

Sum-rate = 8 bits/chip, Pout = 0.01

Average transmitted power (dB)

35
TDMA capacity

30

about
12dB

25

channel
capacity
Capacity

20

15

3
4
5
The number of users K

8
28

Multi-User Gain in Fading Channels

Sum-rate = 8 bits/chip, Pout = 0.01

Average transmitted power (dB)

35

Simulated
TDMA,
AssumedTDMA
performance
TDMA capacity

30

25

IDMA
IDMA,Simulated
Convolutional
coding

20

channel
capacity
Capacity

15

3
4
5
The number of users K

about
12dB
about
12dB

8
29

IDMA-OFDM
IDMA

IFFT

channel

FFT

CBC

Basic Principle:
Randomly distribute information on the sub-carriers
on frequency- time.
Multi-layer structure using superposition coding.
Iterative detection.
This is proposed by Zhou et al and by Mahafeno et al. It has been
considered as an option in 3GPP LTE.
30

OFDM-IDMA vs BICM-OFDM
OFDM-IDMA

Multi-user
gain

For details, see

OFDMA

Average transmission power

Li Ping, Qinghua Guo, and Jun Tong, The OFDM-IDMA approach to wireless communication
systems, IEEE Wireless Commun. Mag., June 2007.
31

Outline

Introduction
Iterative detection
Performance evaluation
Multi-user gain in fading channels
Other applications
Conclusions

32

IDMA
user-1
d1

ENC1

x1

...

h1

user-k
dk

ENCk

xk

hk

r = hk x k + n
k =1

...

hK

user-K
dK

ENCK

xK

..

33

Superposition Coded Modulation (SCM)

layer-1
ENC1

x1

...

layer-k
d

S/P

ENCk

xk

y = hk xk
k =1

...

layer-K
ENCK

xK

Jun Tong and Li Ping, Superposition coding with peak power limitation to appear in IEEE Trans
Inform. Theory. Available at http://www.ee.cityu.edu.hk/~liping/Research/Journal/index.html 34

Comparison of SCM and BICM (1)

1.0E+00

1.0E-01

BER

1.0E-02

1.0E-03

SCM-OFDM

1.0E-04

BICM-OFDM
(SP mapping)

1.0E-05

1.0E-06
5.0

6.0

7.0

8.0

Eb/No(dB)

SCM-OFDM vs BICM-OFDM over fading channels, with and without

clipping; R = 2 b/s, clipping ratio = 3 dB
35

Comparison of SCM and BICM (2)

1.0E+00
1.0E-01

BER

1.0E-02

BICM-OFDM
SP mapping

1.0E-03
1.0E-04
1.0E-05

SCM-OFDM

1.0E-06
8

10

11

12

Eb/No(dB)

SCM-OFDM vs BICM-OFDM over fading channels, with and without

clipping; R = 3 b/s, clipping ratio = 3 dB
36

Comparison of SCM and BICM (3)

1.0E+00

1.0E-01

BER

1.0E-02

1.0E-03

BICM-OFDM
SP mapping

1.0E-04

SCM-OFDM

1.0E-05
1.0E-06
13

14

15

16

17

18

19

20

21

22

Eb/No(dB)

SCM-OFDM vs BICM-OFDM over fading channels, with and without

clipping; R = 5 b/s, clipping ratio = 3 dB
37

IDM Space-Time Coding

layer-1
ENC1

x1

...

S/P

ENCk

xk

Antenna m

...

Antenna 1

layer-k
d

layer-K

ENCK

xK

Antenna M

Keying Wu and Li Ping, A Quasi Random Approach to Space-Time Coding IEEE Trans
Inform. Theory, pp. 1073 1085, March 2008.

38

IDM Space-Time Coding Performance

Overall rate = 4 bits/symbol

1.E-01

IDM-ST Codes
Outage capacity

FER

1.E-02
2x1

1.E-03
4x1
1.E-04
10

15

20
E b /N 0 (dB)

25

30
39

Multi-Path Channels

r = Hx+

See
Xiaojun Yuan, Qinghua Guo, Xiaodong Wang and Li Ping, Evolution analysis of low-cost iterative
equalization in coded linear systems with cyclic prefixes, IEEE J. Select. Areas Commun. Feb. 2008.
40

Application in Ad Hoc Networks

destination
source

41

Conclusions: Features of IDMA

low receiver cost

de-centralized (i.e., asynchronous) control,
simple treatment of ISI,
cross-cell interference mitigation,
diversity against fading,
power efficiency
easy for adaptive rate control
high user number and high spectral efficiency
suitable for wide or narrow band transmission

42