September 2004

France TelecomSlide 1

doc.: IEEE 802.11-04/903-00-0000n

Submission

Partial Proposal: Turbo Codes

Marie-Helene Hamon, Olivier Seller, John Benko France TelecomClaude Berrou ENST BretagneJacky Tousch TurboConceptBrian Edmonston iCoding

September 2004

France TelecomSlide 2

doc.: IEEE 802.11-04/903-00-0000n

Submission

Outline

Part I: Turbo Codes

Part II: Turbo Codes for 802.11n • Why TC for 802.11n?• Flexibility• Performance

September 2004

France TelecomSlide 3

doc.: IEEE 802.11-04/903-00-0000n

Submission

Outline

Part I: Turbo Codes

Part II: Turbo Codes for 802.11n • Why TC for 802.11n?• Flexibility• Performance

September 2004

France TelecomSlide 4

doc.: IEEE 802.11-04/903-00-0000n

Submission

Known applications of convolutional

turbo codes

Application turbo code termination polynomials rates

CCSDS(deep space)

binary,16-state

tail bits 23, 33, 25, 37 1/6, 1/4, 1/3, 1/2

UMTS, CDMA2000(3G Mobile)

binary,8-state

tail bits 13, 15, 17 1/4, 1/3, 1/2

DVB-RCS(Return Channel over Satellite)

duo-binary,8-state

circular 15, 13 1/3 up to 6/7

DVB-RCT(Return Channel over Terrestrial)

duo-binary,8-state

circular 15, 13 1/2, 3/4

Inmarsat(M4)

binary,16-state

no 23, 35 1/2

Eutelsat(Skyplex)

duo-binary,8-state

circular 15, 13 4/5, 6/7

IEEE 802.16(WiMAX)

duo-binary,8-state

circular 15, 13 1/2 up to 7/8

September 2004

France TelecomSlide 5

doc.: IEEE 802.11-04/903-00-0000n

Submission

Main progress in turbo coding/decoding since 1993

• Max-Log-MAP and Max*-Log-MAP algorithms

• Sliding window

• Duo-binary turbo codes

• Circular (tail-biting) encoding

• Permutations

• Parallelism

• Computation or estimation of Minimum Hamming distances (MHDs)

• Stopping criterion

• Bit-interleaved turbo coded modulation

• Simplicity

• Simplicity

• Performance and simplicity

• Performance

• Performance

• Throughput

• Maturity

• Power consumption

• Performance and simplicity

September 2004

France TelecomSlide 6

doc.: IEEE 802.11-04/903-00-0000n

Submission

k binarydata

permutation

Y1

Y2

X

B

A

Y1

Y2

permutation

k/2 binarycouples

polynomials 15, 13 (or 13, 15)

k binarydata

permutation

Y1

Y2

XB

A

Y1

Y2

permutation

k/2 binarycouples

polynomials 23, 35 (or 31, 27)

(a) (b)

(c) (d)

The TCs used in practice

September 2004

France TelecomSlide 7

doc.: IEEE 802.11-04/903-00-0000n

Submission

The turbo code proposed for all sizes, all coding rates

permutation (N)N = k/2

couplesof data

codeword

systematic part

redundancy part

1

2punctu-ring

Y

A

B

AB

Y

systematic part

redundancy part +circular (tail-biting) encoding

Very simple algorithmic permutation:i = 0, …, N-1, j = 0, ...N-1

level 1: if j mod. 2 = 0, let (A,B) = (B,A) (invert the couple)

level 2:

- if j mod. 4 = 0, then P = 0;

- if j mod. 4 = 1, then P = N/2 + P1;

- if j mod. 4 = 2, then P = P2;

- if j mod. 4 = 3, then P = N/2 + P3.

i = P0*j + P +1 mod. N

• No ROM

• Quasi-regular (no routing issue)

• Versatility

• Inherent parallelism

September 2004

France TelecomSlide 8

doc.: IEEE 802.11-04/903-00-0000n

Submission

Decoding

Max-Log-MAP algorithm

Sliding window

FER

5

5

5

510-3

10-4

10-1

10-2

Eb/N0 (dB)3 4

Full MAP Max-Log-MAP

Theoretical limit(sphere packing bound)

Gaussian,1504 bits, R = 4/5

+ inherent parallelism, easy connectivity (quasi-regular permutation)

September 2004

France TelecomSlide 9

doc.: IEEE 802.11-04/903-00-0000n

Submission

Decoding complexityUseful rate: 100 Mbps with 8 iterations

5-bit quantization (data and extrinsic)

Gates

• 164,000 @ Clock = 100 Mhz

• 82,000 @ Clock = 200 Mhz

• 54,000 @ Clock = 400 Mhz

RAM

Data input buffer

+

8.5xk for extrinsic information

+ 4000 for sliding window

(example: 72,000 bits for 1000-byte block)

No ROM

For 0.18m CMOS

Duo-binary TC decoders are already available from several providers (iCoding Tech., TurboConcept, ECC, Xilinx, Altera, …)

September 2004

France TelecomSlide 10

doc.: IEEE 802.11-04/903-00-0000n

Submission

Outline

Part I: Turbo Codes

Part II: Turbo Codes for 802.11n • Why TC for 802.11n?• Flexibility• Performance

September 2004

France TelecomSlide 11

doc.: IEEE 802.11-04/903-00-0000n

Submission

Introduction• Purpose

– Show the multiple benefits of TCs for 802.11n standard– Overview of duo-binary TCs – Comparison between TC and .11a Convolutional Code– High Flexibility– Complexity

• Properties of Turbo Codes (TCs)

– Rely on soft iterative decoding to achieve high coding gains– Good performance, near channel capacity for long blocks– Easy adaptation in the standard frame

• (easy block size adaptation to the MAC layer)– Well controlled hardware development and complexity– TC advantages led to recent adoption in standards

September 2004

France TelecomSlide 12

doc.: IEEE 802.11-04/903-00-0000n

Submission

Duo-Binary Turbo Code

s1 s3s2A

B

W

systematic part

redundancy partY

permutation (k/2)

N = k/2 couplesof data

codeword

systematic part

redundancy part

1

2

puncturing

Y1 or 2W1 or 2

A

B

September 2004

France TelecomSlide 13

doc.: IEEE 802.11-04/903-00-0000n

Submission

Duo-Binary Turbo Code

• Duo-binary input:– Reduction of Latency & Complexity (compared to UMTS TCs)– Complexity per decoded bit is 35 % lower than binary UMTS TCs.– Better convergence in the iterative decoding process

• Circular Recursive Systematic Codes– Constituent codes– No trellis termination overhead!

• Original permuter scheme– Larger minimum distance– Better asymptotic performance

September 2004

France TelecomSlide 14

doc.: IEEE 802.11-04/903-00-0000n

Submission

# of Iterations vs. Performance

The number of iterations can

be adjusted for better

performance – complexity trade-off

September 2004

France TelecomSlide 15

doc.: IEEE 802.11-04/903-00-0000n

Submission

Simulation Environment

• Both Turbo Codes and 802.11a CCs simulated

• Simulation chain based on 802.11a PHY model – SISO configuration– CC59 and CC67 followed– Simulated Channels: AWGN, models B, D, E– No PHY impairments– Packet size of 1000 bytes.– Minimum of 100 packet errors

• Assume perfect channel estimation & synchronization

• Turbo Code settings:– 8-state Duo-Binary Convolutional Turbo Codes– Max-Log-MAP decoding– 8 iterations

September 2004

France TelecomSlide 16

doc.: IEEE 802.11-04/903-00-0000n

Submission

Performance: AWGN

3.5-4 dB gain over

802.11a CC

September 2004

France TelecomSlide 17

doc.: IEEE 802.11-04/903-00-0000n

Submission

Performance: model B

~3 dB gain over 802.11a

CC

September 2004

France TelecomSlide 18

doc.: IEEE 802.11-04/903-00-0000n

Submission

Performance: model D

~3 dB gain over 802.11a

CC

September 2004

France TelecomSlide 19

doc.: IEEE 802.11-04/903-00-0000n

Submission

Performance: model E

~3 dB gain over 802.11a

CC

September 2004

France TelecomSlide 20

doc.: IEEE 802.11-04/903-00-0000n

Submission

Flexibility

• All Coding Rates possible (no limitations)• Same encoder/decoder for:

– any coding rate via simple puncturing adaptation– different block sizes via adjusting permutation parameters

• 4 parameters are used per block size to define an interleaver

• Higher PHY data rates enabled with TCs:– High coding gains over 802.11a CC ( =>lower PER)– More efficient transmission modes enabled more often.

• Combination with higher-order constellations

• Better system efficiency– ARQ algorithm used less frequently

September 2004

France TelecomSlide 21

doc.: IEEE 802.11-04/903-00-0000n

Submission

Conclusions• Mature, stable, well established and implemented

• Multiple Patents, but well defined licensing– All other advanced FECs also have patents

• Complexity:– Show 35% decrease in complexity per decoded bit over UMTS TCs– Performance is slightly better than UMTS TCs

• Significant performance gain over .11a CC:– 3.5 - 4 dB on AWGN channel– 3 dB on 802.11n channel models

September 2004

France TelecomSlide 22

doc.: IEEE 802.11-04/903-00-0000n

Submission

References

• [1] IEEE 802.11-04/003, "Turbo Codes for 802.11n", France Telecom R&D, ENST Bretagne, iCoding Technology, TurboConcept, January 2004.

• [2] IEEE 802.11-04/243, "Turbo Codes for 802.11n", France Telecom R&D,iCoding Technology, May 2004.

• [3] IEEE 802-04/256, "PCCC Turbo Codes for IEEE 802.11n", IMEC, March 2004.• [4] C. Berrou, A. Glavieux, P. Thitimajshima, "Near Shannon limit error-correcting

coding and decoding: Turbo Codes", ICC93, vol. 2, pp. 1064-1070, May 93.• [5] C. Berrou, "The ten-year-old turbo codes are entering into service", IEEE

Communications Magazine, vol. 41, pp. 110-116, August 03.• [6] C. Berrou, M. Jezequel, C. Douillard, S. Kerouedan, "The advantages of non-binary

turbo codes", Proc IEEE ITW 2001, pp. 61-63, Sept. 01.• [7] TS25.212 : 3rd Generation Partnership Project (3GPP) ; Technical Specification

Group (TSG) ; Radio Access Network (RAN) ; Working Group 1 (WG1); "Multiplexing and channel coding (FDD)". October 1999.

• [8] EN 301 790 : Digital Video Broadcasting (DVB) "Interaction channel or satellite distribution systems". December 2000.

• [9] EN 301 958 : Digital Video Broadcasting (DVB) "Specification of interaction channel for digital terrestrial TV including multiple access OFDM". March 2002.