Stewart Crozier

Distance upper bounds and true minimum distance results for turbo-codes designed with DRP interleavers

This paper presents a number of distance upper bounds for Turbo-codes designed with dithered relative prime (drp) interleavers. These bounds help determine how much dither should be applied and which input data patterns to test when designing high-distance interleavers with different lengths. Using a block length of only 640 data bits, true minimum distances of 30, 42 and 53 have been achieved for rate 1/3 Turbo-codes with 4, 8 and 16-state constituent codes, respectively. Simulated error rate results for a block length of 1504 data bits (188 bytes,mpeg packet) show excellent flare performance.RésuméCet article présente un certain nombre de limites supérieures de distance pour des turbo-codes à entrelacements «réguliers à faibles variations» (dithered relative primedrp). Ces limites aident à choisir quelles petites variations doivent être appliquées et quels motifs de données d’entrée sont à tester pour définir des entrelacements permettant d’atteindre de grandes distances avec différentes tailles de blocs. Pour un bloc de 640 bits seulement, des distances minimales de 30, 42 et 53 ont pu être atteintes sur des turbocodes de rendement 1/3 à 4, 8 et 16 états respectivement. Des simulations de taux d’erreur pour des blocs de 1504 bits (188 octets, trame mpeg) montrent d’excellentes performances asymptotiques.

Improving the flare performance of turbo codes using error detection and event flipping

The error rate performance of turbo codes tends to flare at high SNRs due to poor distance properties. It is shown that the flare performance can be significantly improved by using an outer error detection code and a low-complexity event flipping cleanup approach. Using a rate 1/2, single-binary, 4-state turbo code with a 2048-bit data block, the packet error rate at 1.8 dB is lowered from 10-4 to about 10-7, with only a small increase in peak decoder complexity. The approach also requires significantly fewer overhead bits than algebraic cleanup codes with hard decoding.

Convergence performance and EXIT analysis of 4-state partially-systematic Turbo codes

The error-rate performance of punctured turbo codes may be significantly improved by puncturing data symbols as well as parity symbols, yielding partially-systematic turbo codes (PSTCs). In this paper, it is shown that relatively simple 4-state PSTCs are able to achieve convergence (i.e., waterfall) performance that is 0.26 dB from the binary channel capacity limit at rate 1/2 and 0.18 dB from the limit at rate 8/9. It is also demonstrated that extrinsic information transfer (EXIT) charts do not accurately predict convergence performance for 4-state PSTCs.

Implementation of a simple delay-multiply-average technique for frequency estimation on a fixed point DSP

Describes the implementation of an efficient algorithm for frequency estimation. The algorithm uses a bank of simple delay-multiply-average operations with an increasing delay in each branch. The branch outputs are combined to obtain the frequency estimate. This approach is capable of performance that is close to the Cramer-Rao lower bound. The implementation on a fixed point DSP is described for use in a burst acquisition application. A comparison is made to FFT based techniques.<<ETX>>

Adaptive coding and modulation for return satellite links using binary turbo coding

A turbo code that is based on an 8-state binary constituent code is considered in this paper for use in an adaptive coding and modulation system for a return satellite link. The paper presents the results of finding good puncture-constrained dithered relative prime interleavers along with data puncture patterns to provide the required performance. A tradeoff between waterfall and flare performance can be made by selecting the best puncture masks for the required error rate. The performance of the turbo code with a packet size of 1504 data bits is presented for the following modulations: BPSK, QPSK, 8 PSK and 16 APSK. A comparison with other coding strategies shows that this code is an attractive option for use in a return satellite link application using adaptive coding and modulation.

Forward error correction for an enhanced SARSAT distress message

This paper presents a proposed enhancement to the current SARSAT distress message that is backward-compatible with the existing local user terminals (LUT) (i.e., ground stations) receivers. Additional coding is appended to the current message to allow for better performance for the new LUTs, soft-decision decoding is used in the evaluation of the current and enhanced format. Performance results on the AWGN channel are presented. It is shown that near maximum-likelihood performance is achieved for the new code format.