A study of turbo codes for multilevel modulations in Gaussian and mobile channels

Similar documents
Space Time Equalization-space time codes System Model for STCM

Impact of Interference Model on Capacity in CDMA Cellular Networks. Robert Akl, D.Sc. Asad Parvez University of North Texas

Error Probability of RS Code Over Wireless Channel

Rejection of PSK Interference in DS-SS/PSK System Using Adaptive Transversal Filter with Conditional Response Recalculation

Bit-interleaved Rectangular Parity-Check Coded Modulation with Iterative Demodulation In a Two-Node Distributed Array

Parameter Free Iterative Decoding Metrics for Non-Coherent Orthogonal Modulation

DC-FREE TURBO CODING SCHEME FOR GPRS SYSTEM

Performance Analysis of Multi User MIMO System with Block-Diagonalization Precoding Scheme

Adaptive Modulation for Multiple Antenna Channels

Calculation of the received voltage due to the radiation from multiple co-frequency sources

Digital Transmission

Markov Chain Monte Carlo Detection for Underwater Acoustic Channels

Bit Error Probability of Cooperative Diversity for M-ary QAM OFDM-based system with Best Relay Selection

arxiv: v1 [cs.it] 30 Sep 2008

THE USE OF CONVOLUTIONAL CODE FOR NARROWBAND INTERFERENCE SUPPRESSION IN OFDM-DVBT SYSTEM

PRACTICAL, COMPUTATION EFFICIENT HIGH-ORDER NEURAL NETWORK FOR ROTATION AND SHIFT INVARIANT PATTERN RECOGNITION. Evgeny Artyomov and Orly Yadid-Pecht

DVB-T/H Digital Television Transmission and its Simulation over Ricean and Rayleigh Fading Channels

Power Minimization Under Constant Throughput Constraint in Wireless Networks with Beamforming

A thesis presented to. the faculty of. the Russ College of Engineering and Technology of Ohio University. In partial fulfillment

The Throughput of Hybrid-ARQ in Block Fading under Modulation Constraints

Performance Analysis of Power Line Communication Using DS-CDMA Technique with Adaptive Laguerre Filters

Multicarrier Modulation

A NEW TRANSMISSION STRATEGY FOR SCALABLE MULTIMEDIA DATA ON OFDM SYSTEMS

Evaluation of Downlink Performance of a Multiple-Cell, Rake Receiver Assisted CDMA Mobile System

High Speed, Low Power And Area Efficient Carry-Select Adder

NATIONAL RADIO ASTRONOMY OBSERVATORY Green Bank, West Virginia SPECTRAL PROCESSOR MEMO NO. 25. MEMORANDUM February 13, 1985

The Performance Improvement of BASK System for Giga-Bit MODEM Using the Fuzzy System

Comparative Analysis of Reuse 1 and 3 in Cellular Network Based On SIR Distribution and Rate

Approximating User Distributions in WCDMA Networks Using 2-D Gaussian

Revision of Lecture Twenty-One

Phasor Representation of Sinusoidal Signals

Index Terms Adaptive modulation, Adaptive FEC, Packet Error Rate, Performance.

Iterative Detection and Decoding (IDD) MIMO-OFDM HARQ Algorithm with Antenna Scheduling

BER Performances of a Differential OFDM System in Fading Channels

Exponential Effective SIR Metric for LTE Downlink

Performance of Reverse Link CDMA in a Multi-cell Environment with Moving Cells*

Optimizing Transmission Lengths for Limited Feedback with Non-Binary LDPC Examples

Reduced Cluster Search ML Decoding for QO-STBC Systems

IEE Electronics Letters, vol 34, no 17, August 1998, pp ESTIMATING STARTING POINT OF CONDUCTION OF CMOS GATES

Dynamic Optimization. Assignment 1. Sasanka Nagavalli January 29, 2013 Robotics Institute Carnegie Mellon University

antenna antenna (4.139)

On the Feasibility of Receive Collaboration in Wireless Sensor Networks

AN IMPROVED BIT LOADING TECHNIQUE FOR ENHANCED ENERGY EFFICIENCY IN NEXT GENERATION VOICE/VIDEO APPLICATIONS

OFDM AF Variable Gain Relay System for the Next Generation Mobile Cellular Networks

To: Professor Avitabile Date: February 4, 2003 From: Mechanical Student Subject: Experiment #1 Numerical Methods Using Excel

DESIGN OF OPTIMIZED FIXED-POINT WCDMA RECEIVER

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 5, NO. 3, MARCH Transactions Letters

Multipath Propagation. Outline. What is OFDM? (OFDM) for Broadband Communications and. Orthogonal Frequency Division Multiplexing

Resource Allocation Optimization for Device-to- Device Communication Underlaying Cellular Networks

EFFICIENT FIELD PROGRAMMABLE GATE ARRAY IMPLEMENTATION OF A CONVOLUTIONAL TURBO CODE FOR LONG TERM EVOLUTION SYSTEMS

A Low-Complexity Soft QAM De-Mapper Based on First-Order Linear Approximation

On the Optimal Solution for BER Performance Improvement in Dual-Hop OFDM Relay Systems

How to Scale Up the Spectral Efficiency of Multi-way Massive MIMO Relaying?

NOVEL ITERATIVE TECHNIQUES FOR RADAR TARGET DISCRIMINATION

An Efficient Method for PAPR Reduction of OFDM Signal with Low Complexity

Performance Study of OFDMA vs. OFDM/SDMA

Uncertainty in measurements of power and energy on power networks

Side-Match Vector Quantizers Using Neural Network Based Variance Predictor for Image Coding

MIMO-OFDM Systems. Team Telecommunication and Computer Networks, FSSM, University Cadi Ayyad, P.O. Box 2390, Marrakech, Morocco.

DESIGN OF OPTIMIZED FIXED-POINT WCDMA RECEIVER

PAPER Effect of Joint Detection on System Throughput in Distributed Antenna Network

Effect of Time-Interleaved Analog-to-Digital Converter Mismatches on OFDM Performance

Performance of Modified Iterative Decoding Algorithm for Multilevel Codes in Adaptive OFDM System

Hierarchical Generalized Cantor Set Modulation

Joint Power Control and Scheduling for Two-Cell Energy Efficient Broadcasting with Network Coding

TECHNICAL NOTE TERMINATION FOR POINT- TO-POINT SYSTEMS TN TERMINATON FOR POINT-TO-POINT SYSTEMS. Zo = L C. ω - angular frequency = 2πf

SC-FDMA and OFDMA: An Efficient Wireless Image Transmission Schemes

64-QAM Communication System using Three-beam Spatial Power Combining Technology

Throughput Maximization by Adaptive Threshold Adjustment for AMC Systems

A NSGA-II algorithm to solve a bi-objective optimization of the redundancy allocation problem for series-parallel systems

The Detection Algorithms Performance in BLAST Enhanced IEEE a WLAN Standard on Measured Channels. University of Bristol

Spatial Multiplexing applied to Turbo Coded Multi-Carrier CDMA

RESOURCE CONTROL FOR HYBRID CODE AND TIME DIVISION SCHEDULING

Channel Alternation and Rotation in Narrow Beam Trisector Cellular Systems

Multiuser Detection in Multiple Input Multiple Output Orthogonal Frequency Division Multiplexing Systems by Blind Signal Separation Techniques

Resource Allocation for Throughput Enhancement in Cellular Shared Relay Networks

Novel Sampling Clock Offset Estimation for DVB-T OFDM

Walsh Function Based Synthesis Method of PWM Pattern for Full-Bridge Inverter

Chaotic Filter Bank for Computer Cryptography

Resource Control for Elastic Traffic in CDMA Networks

A Current Differential Line Protection Using a Synchronous Reference Frame Approach

A MODIFIED DIFFERENTIAL EVOLUTION ALGORITHM IN SPARSE LINEAR ANTENNA ARRAY SYNTHESIS

A Recursive Approach to Compute Bit Error Rate in Underwater Channels with Multiple Paths

Ergodic Capacity of Block-Fading Gaussian Broadcast and Multi-access Channels for Single-User-Selection and Constant-Power

A Spreading Sequence Allocation Procedure for MC-CDMA Transmission Systems

DIGITAL multi-tone (DMT) modulation, also known as

AN EFFICIENT ITERATIVE DFT-BASED CHANNEL ESTIMATION FOR MIMO-OFDM SYSTEMS ON MULTIPATH CHANNELS

A Comparison of Two Equivalent Real Formulations for Complex-Valued Linear Systems Part 2: Results

An Efficient Energy Adaptive Hybrid Error Correction Technique for Underwater Wireless Sensor Networks

A NOVEL PREAMBLE DESIGN FOR CHANNEL ESTIMATION IN MIMO- OFDM SYSTEMS RESULTING IN ENHANCED THROUGHPUT

Chapter 13. Filters Introduction Ideal Filter

THE GENERATION OF 400 MW RF PULSES AT X-BAND USING RESONANT DELAY LINES *

Approximate Joint MAP Detection of Co-Channel Signals

An Efficient Secured Turbo Codes for Long Term Evolution System Enhancement

Microelectronic Circuits

Adaptive Technique for CI/MC-CDMA System using Combined Strategy of Genetic Algorithms and Neural Network

LOCAL DECODING OF WALSH CODES TO REDUCE CDMA DESPREADING COMPUTATION

Comparison of Coding Schemes over FRBS Aided AOFDM Systems

LETTER An MMSE-Nulling Partial-PIC Receiver for Multiuser Downlink MIMO MC-CDMA Systems

Distributed Resource Allocation and Scheduling in OFDMA Wireless Networks

Transcription:

A study of turbo codes for multlevel modulatons n Gaussan and moble channels Lamne Sylla and Paul Forter (sylla, forter)@gel.ulaval.ca Department of Electrcal and Computer Engneerng Laval Unversty, Ste-Foy, Quebec Canada G1K 7P4 (418) 656-3555, (418) 656-3159 fax ABSTRACT - In ths paper, we present smulaton results for a system combnng turbo codes and multlevel modulaton. Ths scheme s shown to be an excellent soluton when t comes to transmsson over bandlmted channels. Its performance n moble rado channels can be mproved usng OFDM. 1. Introducton Snce the frst publcatons regardng turbo codes, varous studes have been carred out. Most of these studes llustrated performances or dealt wth the optmzaton of the structure of turbo codes usng bnary modulaton. The requrement for hgher nformaton transmsson rates on bandlmted channels have led some researchers to consder turbo codes wth spectrally effcent modulatons. In [2], an applcaton of turbo codes to trells coded modulaton (TCM) called parallel concatenated trells coded modulaton s presented. At the encoder ths scheme uses two recursve systematc convolutonal coders (RSC) of rate R = b ( b 1), b beng even, and two nterleavers for a throughput of b bts/sec/hz 2 2b 2 wth -QAM. The decodng s performed teratvely by the symbol MAP decoder followed by a bt relablty calculaton module. In [3], a turbo encoder employng two Ungerboeck codes n ther recursve form as component codes has been used n combnaton wth TCM. Ths turbo TCM (TTCM) uses an nterleaver operatng over symbols of 2 bts for an 8 PSK modulaton. The varables passed between symbol MAP decoders are vectors of four loglkelhood ratos (LLRs), one for each possble nformaton group value. These two models have n common a major modfcaton on the classcal turbo encoder to ft the TCM model at the expense of an ncrease n complexty at the decoder. Ths paper presents some results obtaned wth turbo codes and multlevel modulatons n Gaussan and moble rado channels by applyng the so called pragmatc approach [1]. Secton 2 deals wth the encoder structure and the sgnal mappng scheme. Secton 3 presents the channels models used n the smulatons and the Secton 4 presents the decodng operaton. The last two sectons are devoted to the smulatons results and some concludng remarks. 2. Codng and modulaton schemes The encoder s made up of two RSCs wth constrant length k = 3 and polynomal generators (5,7). The RSCs are separated by an unform block nterleaver. For short frame transmsson applcatons, the nterleaver was set to a sze of 192 bts whch s compatble wth the IS-95 CDMA cellular standard. Another block sze of 1022 bts has been used. Ths sze, after addng 2 bts for termnatng the RSCs, match a COFDM symbol wth bnary modulaton. The output of the 1/3 turbo encoder s punctured to obtan a 1/2 codng rate. The system presented here s dfferent from the TCM approach n the sense that t uses a codng rate R = 1/2 for a 16 pont constellaton. The output sequence of the encoder (u 1,c 1,u 2,c 2,...) s modulated by two 4-ASK n quadrature resultng n a 16-QAM system where the I and Q components are ndependently coded (usng Gray code). The ASK symbols {A k, B k } are then transmtted through the channel. 3. Channel models Generally, for purpose of smulaton, two approaches are used to model a moble rado channel. The frst one deals wth the multpath phenomenon and the second one deals, n addton to multpath, wth frequency selectvty [7]. A Raylegh flat fadng channel has been used to smulate the multpath phenomenon. In ths case, the receved symbol, (x k, y k ), s gven by x k = α k A k y k = β k B k (1) where (α k, β k ) are two ndependent Raylegh dstrbuted random varables and ( k, q k ) are the addtve whte Gaussan nose components wth varance σ 2. The frequency selectvty has been ntroduced by means of ntersymbol nterference (ISI). In an ar-to-ground k q k 0-7803-4314-X/98/$10.00 1998 IEEE

communcaton, the channel s characterzed by a drect lne of sght (LOS) and a multpath component wth a relatve delay [5]. Ths delayed path creates some ISI on the receved sgnal (2) wth n representng the relatve delay of the dffuse component. Ths model fts the Rcan fadng channel. For applcatons n mcrowave LOS rado channels, a smlar model has been consdered, the Rummler channel whch s a two path model [6] (3) where a s the overall attenuaton, and the term wthn parenthess represents the nterference between two rays havng a relatve delay τ 0 and producng a mnmum ampltude at the notch frequency f 0. b s the relatve ampltude of the multpath ray. The outputs of the channel are coherently demodulated and the soft nputs of the decoder are then (X k, Y k ). 4. Decoder structure x k = A k α k n A k n y k = B k β k n B k n The decodng of ths multlevel modulaton system s much smplfed by use of the pragmatc approach [1]. The turbo decoder, mplemented wth the SOVA [4] algorthm, and optmzed for bnary modulaton, can then be used. Wth ths approach, the decoder must be preceded by a module that computes the LLR of each bt contaned n an ASK symbol. The LLR can be expressed as follows: Λ( u ) k, (4) wth u k,1 = u k and u k,2 = c k and (a 1,, a 0, ) are the realzatons of symbol A k condtonal to u k, = 1 and u k, = 0 respectvely. A good approxmaton of ths expresson s gven by k q k ( )τ 0 Hf () a 1 b e j2π f f 0 = ( ) = σ 2 ------ 2 2 1 --------- 2σ 2 ( X a ) 2 exp k 0, = 1 log----------------------------------------------------------------------------- 2 1 --------- 2σ 2 ( X a ) 2 exp k 1, = 1 (5) Although ths approxmaton can lead to good performance, the expresson of Λ(u k ) n (5) s an overestmaton of ts real value. A better approxmaton can be obtaned by defnng the bt-mappng n one dmenson of the constellaton wth unt energy as follows (6) where 2d s the dstance between two ponts. Ths equaton can be decomposed nto two terms. Λ(u k ) can now be approxmated by the frst term on the second part of ths equaton. Fgure 1 show three curves of Λ(u k ) versus X k. The sold lne represents the true expresson n (4), the * lne represents the frst approxmaton n (5) and the o lne the last approxmaton. The approxmaton n (7) s closer to ts real value and as t wll be shown later, leads to a better performance. 5. Smulatons results Λ( c ) = X 2 k k Λ( u k ) = X k A = 2d ( 1 2u ) k k, 1 d ( 1 2u k, 2 ) sgn ( 1 2u k, 1 ) Λ( u k, 1 ) = 2dX k σ 2 --------- 1 2σ 2 4d 2 2dX k cosh (7) ------ log-------------------------------------------------------------------- 2 --------- 1 2σ 2 4d 2 2dX k cosh The smulatons have been carred out usng the Sgnal Processng WorkSystem (SPW) package. The BER was computed after 3 teratons of the decoder. Fgure 2 shows for a Gaussan channel 3 BER curves. These curves demonstrate that turbo codes combned wth 16-QAM (TC-QAM), even wth short frame sze and non random block nterleaver, gve a better performance than conventonal TCM wth the same throughput. The LLRs were computed as n (5). The codng gan over the TCM can be ncreased by usng the second approxmaton n (7). In fact, by usng ths equaton to compute the LLR, the performance of the TC-QAM can be mproved as s shown n Fgure 3.

For the slow Raylegh flat fadng channel, the results wth QPSK modulaton appear on Fgure 4. A BER of 10-5 at a SNR around 7.5 db s acheved by the TC. On the channels wth ISI, we ntended to compare the turbo COFDM (T-COFDM) wth the TC-QAM. On the Rce model, the multpath component was delayed by an amount of 10 µsec. relatve to the LOS path and the carrer to multpath rato C/M was set to 10 db. A frame sze of 1022 bts was used. The T-COFDM has a codng gan of 1db over the TC-QAM at a BER of 10-4 (Fgure 5). LLR - Eq. 4 * Eq. 5 o Eq. 7 On the Rummler channel the relatve delay between the two paths was set to 16 µsec. and the attenuaton factor, b, set to 0.2. In ths case also, the T-COFDM outperforms the TC-QAM; the BER curve s shown n Fgure 6. 6. Concluson In ths paper we presented a study of turbo codes combned wth multlevel modulatons. The system descrbed here could be an alternatve to the conventonal TCM and s less complex than the PCTCM and TTCM. In frequency selectve channels, a combnaton of TC wth OFDM results n a power and spectrally effcent system. Fgure 1: Λ(u k ) versus X k. 7. References 1. S. Goff, A. Glaveux, and C. Berrou, Turbo-codes and hgh spectral effcency modulaton, Proc., IEEE Int. Conf. on Commun., pp. 645-9, May 1994. 2. S. Benedetto, D. Dvsalar, G. Montors, and F. Pollara, Parallel concatenated trells coded modulaton, Proc., IEEE Int. Conf. on Commun., (May), pp. 974-8, 1996. 3. P. Robertson and T. Worz, Coded modulaton scheme employng turbo codes, Electroncs Letters, vol. 31, pp.1546-7, Aug. 31st 1995. 4. J. Hagenauer and L. Papke, Decodng turbo codes wth the soft-output Vterb algorthm (SOVA), Proc., IEEE Int. Symp. on Inform. Theory, p. 164, June 1994. 5. J. Proaks, Dgtal Communcatons, thrd edton. New York: McGraw-Hll, Inc.,1995. 6. W. D. Rummler et al., Multpath fadng channels models for mcrowave dgtal rado, IEEE Communcaton Magazne, vol. 24, Nov. 1996. 7. W. C. Lee, Moble Communcaton Engneerng, McGraw-Hll, 1982. Fgure 2: Comparson of TC-QAM wth TCM.

Fgure 3: Comparson of the results obtaned by computng the LLR usng dfferent approxmatons. Fgure 5: BER curves of T-COFDM and TC-QAM on a Rcan channel wth ISI. Fgure 4: BER curves of bnary TC and uncoded QPSK on a Raylegh channel. Fgure 6: BER curves of T-COFDM and TC-QAM on a Rummler channel.

A study of turbo codes for multlevel modulatons n Gaussan and moble channels Lamne Sylla and Paul Forter (sylla, forter)@gel.ulaval.ca Department of Electrcal and Computer Engneerng Laval Unversty, Ste-Foy, Quebec Canada G1K 7P4 (418) 656-3555, (418) 656-3159 fax ABSTRACT - In ths paper, we present smulaton results for a system combnng turbo codes and multlevel modulaton. Ths scheme s shown to be an excellent soluton when t comes to transmsson over bandlmted channels. Its performance n moble rado channels can be mproved usng OFDM. 1. Introducton Snce the frst publcatons regardng turbo codes, varous studes have been carred out. Most of these studes llustrated performances or dealt wth the optmzaton of the structure of turbo codes usng bnary modulaton. The requrement for hgher nformaton transmsson rates on bandlmted channels have led some researchers to consder turbo codes wth spectrally effcent modulatons. In [2], an applcaton of turbo codes to trells coded modulaton (TCM) called parallel concatenated trells coded modulaton s presented. At the encoder ths scheme uses two recursve systematc convolutonal coders (RSC) of rate R = b ( b 1), b beng even, and two nterleavers for a throughput of b bts/sec/hz 2 2b 2 wth -QAM. The decodng s performed teratvely by the symbol MAP decoder followed by a bt relablty calculaton module. In [3], a turbo encoder employng two Ungerboeck codes n ther recursve form as component codes has been used n combnaton wth TCM. Ths turbo TCM (TTCM) uses an nterleaver operatng over symbols of 2 bts for an 8 PSK modulaton. The varables passed between symbol MAP decoders are vectors of four loglkelhood ratos (LLRs), one for each possble nformaton group value. These two models have n common a major modfcaton on the classcal turbo encoder to ft the TCM model at the expense of an ncrease n complexty at the decoder. Ths paper presents some results obtaned wth turbo codes and multlevel modulatons n Gaussan and moble rado channels by applyng the so called pragmatc approach [1]. Secton 2 deals wth the encoder structure and the sgnal mappng scheme. Secton 3 presents the channels models used n the smulatons and the Secton 4 presents the decodng operaton. The last two sectons are devoted to the smulatons results and some concludng remarks. 2. Codng and modulaton schemes The encoder s made up of two RSCs wth constrant length k = 3 and polynomal generators (5,7). The RSCs are separated by an unform block nterleaver. For short frame transmsson applcatons, the nterleaver was set to a sze of 192 bts whch s compatble wth the IS-95 CDMA cellular standard. Another block sze of 1022 bts has been used. Ths sze, after addng 2 bts for termnatng the RSCs, match a COFDM symbol wth bnary modulaton. The output of the 1/3 turbo encoder s punctured to obtan a 1/2 codng rate. The system presented here s dfferent from the TCM approach n the sense that t uses a codng rate R = 1/2 for a 16 pont constellaton. The output sequence of the encoder (u 1,c 1,u 2,c 2,...) s modulated by two 4-ASK n quadrature resultng n a 16-QAM system where the I and Q components are ndependently coded (usng Gray code). The ASK symbols {A k, B k } are then transmtted through the channel. 3. Channel models Generally, for purpose of smulaton, two approaches are used to model a moble rado channel. The frst one deals wth the multpath phenomenon and the second one deals, n addton to multpath, wth frequency selectvty [7]. A Raylegh flat fadng channel has been used to smulate the multpath phenomenon. In ths case, the receved symbol, (x k, y k ), s gven by x k = α k A k y k = β k B k (1) where (α k, β k ) are two ndependent Raylegh dstrbuted random varables and ( k, q k ) are the addtve whte Gaussan nose components wth varance σ 2. The frequency selectvty has been ntroduced by means of ntersymbol nterference (ISI). In an ar-to-ground k q k 0-7803-4314-X/98/$10.00 1998 IEEE

communcaton, the channel s characterzed by a drect lne of sght (LOS) and a multpath component wth a relatve delay [5]. Ths delayed path creates some ISI on the receved sgnal (2) wth n representng the relatve delay of the dffuse component. Ths model fts the Rcan fadng channel. For applcatons n mcrowave LOS rado channels, a smlar model has been consdered, the Rummler channel whch s a two path model [6] (3) where a s the overall attenuaton, and the term wthn parenthess represents the nterference between two rays havng a relatve delay τ 0 and producng a mnmum ampltude at the notch frequency f 0. b s the relatve ampltude of the multpath ray. The outputs of the channel are coherently demodulated and the soft nputs of the decoder are then (X k, Y k ). 4. Decoder structure x k = A k α k n A k n y k = B k β k n B k n The decodng of ths multlevel modulaton system s much smplfed by use of the pragmatc approach [1]. The turbo decoder, mplemented wth the SOVA [4] algorthm, and optmzed for bnary modulaton, can then be used. Wth ths approach, the decoder must be preceded by a module that computes the LLR of each bt contaned n an ASK symbol. The LLR can be expressed as follows: Λ( u ) k, (4) wth u k,1 = u k and u k,2 = c k and (a 1,, a 0, ) are the realzatons of symbol A k condtonal to u k, = 1 and u k, = 0 respectvely. A good approxmaton of ths expresson s gven by k q k ( )τ 0 Hf () a 1 b e j2π f f 0 = ( ) = σ 2 ------ 2 2 1 --------- 2σ 2 ( X a ) 2 exp k 0, = 1 log----------------------------------------------------------------------------- 2 1 --------- 2σ 2 ( X a ) 2 exp k 1, = 1 (5) Although ths approxmaton can lead to good performance, the expresson of Λ(u k ) n (5) s an overestmaton of ts real value. A better approxmaton can be obtaned by defnng the bt-mappng n one dmenson of the constellaton wth unt energy as follows (6) where 2d s the dstance between two ponts. Ths equaton can be decomposed nto two terms. Λ(u k ) can now be approxmated by the frst term on the second part of ths equaton. Fgure 1 show three curves of Λ(u k ) versus X k. The sold lne represents the true expresson n (4), the * lne represents the frst approxmaton n (5) and the o lne the last approxmaton. The approxmaton n (7) s closer to ts real value and as t wll be shown later, leads to a better performance. 5. Smulatons results Λ( c ) = X 2 k k Λ( u k ) = X k A = 2d ( 1 2u ) k k, 1 d ( 1 2u k, 2 ) sgn ( 1 2u k, 1 ) Λ( u k, 1 ) = 2dX k σ 2 --------- 1 2σ 2 4d 2 2dX k cosh (7) ------ log-------------------------------------------------------------------- 2 --------- 1 2σ 2 4d 2 2dX k cosh The smulatons have been carred out usng the Sgnal Processng WorkSystem (SPW) package. The BER was computed after 3 teratons of the decoder. Fgure 2 shows for a Gaussan channel 3 BER curves. These curves demonstrate that turbo codes combned wth 16-QAM (TC-QAM), even wth short frame sze and non random block nterleaver, gve a better performance than conventonal TCM wth the same throughput. The LLRs were computed as n (5). The codng gan over the TCM can be ncreased by usng the second approxmaton n (7). In fact, by usng ths equaton to compute the LLR, the performance of the TC-QAM can be mproved as s shown n Fgure 3.

For the slow Raylegh flat fadng channel, the results wth QPSK modulaton appear on Fgure 4. A BER of 10-5 at a SNR around 7.5 db s acheved by the TC. On the channels wth ISI, we ntended to compare the turbo COFDM (T-COFDM) wth the TC-QAM. On the Rce model, the multpath component was delayed by an amount of 10 µsec. relatve to the LOS path and the carrer to multpath rato C/M was set to 10 db. A frame sze of 1022 bts was used. The T-COFDM has a codng gan of 1db over the TC-QAM at a BER of 10-4 (Fgure 5). LLR - Eq. 4 * Eq. 5 o Eq. 7 On the Rummler channel the relatve delay between the two paths was set to 16 µsec. and the attenuaton factor, b, set to 0.2. In ths case also, the T-COFDM outperforms the TC-QAM; the BER curve s shown n Fgure 6. 6. Concluson In ths paper we presented a study of turbo codes combned wth multlevel modulatons. The system descrbed here could be an alternatve to the conventonal TCM and s less complex than the PCTCM and TTCM. In frequency selectve channels, a combnaton of TC wth OFDM results n a power and spectrally effcent system. Fgure 1: Λ(u k ) versus X k. 7. References 1. S. Goff, A. Glaveux, and C. Berrou, Turbo-codes and hgh spectral effcency modulaton, Proc., IEEE Int. Conf. on Commun., pp. 645-9, May 1994. 2. S. Benedetto, D. Dvsalar, G. Montors, and F. Pollara, Parallel concatenated trells coded modulaton, Proc., IEEE Int. Conf. on Commun., (May), pp. 974-8, 1996. 3. P. Robertson and T. Worz, Coded modulaton scheme employng turbo codes, Electroncs Letters, vol. 31, pp.1546-7, Aug. 31st 1995. 4. J. Hagenauer and L. Papke, Decodng turbo codes wth the soft-output Vterb algorthm (SOVA), Proc., IEEE Int. Symp. on Inform. Theory, p. 164, June 1994. 5. J. Proaks, Dgtal Communcatons, thrd edton. New York: McGraw-Hll, Inc.,1995. 6. W. D. Rummler et al., Multpath fadng channels models for mcrowave dgtal rado, IEEE Communcaton Magazne, vol. 24, Nov. 1996. 7. W. C. Lee, Moble Communcaton Engneerng, McGraw-Hll, 1982. Fgure 2: Comparson of TC-QAM wth TCM.

Fgure 3: Comparson of the results obtaned by computng the LLR usng dfferent approxmatons. Fgure 5: BER curves of T-COFDM and TC-QAM on a Rcan channel wth ISI. Fgure 4: BER curves of bnary TC and uncoded QPSK on a Raylegh channel. Fgure 6: BER curves of T-COFDM and TC-QAM on a Rummler channel.