The performance of convolutional codes on fading channels

Abstract

A rather complete performance analysis of the uncoded system as well some diversity transmission systems is presented first. Fairly general performance bounds on Viterbi decoding for the case of zero-memory (fully interleaved) fading channels are then derived. Useful techniques for predicting code performance on a slow fading channel is demonstrated. The efficacy of simple block interleaving as a means of combatting channel memory is thoroughly explored. The performance of list-of-ℓ as well as uniform receiver quantization is investigated.

The primary concern of this work has been with deep space communications. However, the analysis is quite general and the resultp are expected to be of much wider applications.

Several soft-decision decoding algorithms for block codes are considered. For comparable decoder complexity, short constraint length convolutional codes are shown to be superior to block codes and the difference in performance increases with fading severity.

The impact of phase tracking errors on the overall system performance is examined. We consider phase tracking systems which employ the transmission of pilot tones as well as those which derive the phase estimate from the information-bearing signal. A hybrid phase tracker, which makes use of a low-energy pilot carrier to resolve the phase ambiguity of the reference phase recovered from the modulated signal, is shown to be an attractive scheme. The question of optimum allocation of power between the synchronization and the information signals is discussed

An empirical correspondence between the Rician and the lognormal channel parameters is developed. It is shown that under this correspondence, the coded system performance is relatively insensitive to the underlying probability distribution of the fading provided that full interleaving is employed.

The performance of short constraint length convolutional codes used in conjunction with the Viterbi algorithm on both the Rician and the lognormal fading channels is examined. The modulation strategies studied include coherent binary phase-shift keying (BPSK), differential phase-shift keying (DPSK), noncoherent binary and multipIe frequency-shift keying (BFSK and tWSK). The quantity of interest is the bit error probability performance as a function of the signal-to-noise ratio, Eb/N₀ parameterized by a few meaningful channel parameters. Extensive computer simulation is used to support and supplement the analysis.