Performance evaluation of superorthogonal turbo codes in AWGN andflat Rayleigh fading channels

摘要

Turbo codes are parallel concatenated codes whose performance innthe additive white Gaussian noise (AWGN) channel has been shown to bennear the theoretical limit. In this paper, we describe a low-ratensuperorthogonal turbo code that combines the principles of low-ratenconvolutional coding and that of parallel concatenation. Due to thenbandwidth expansion, this code outperforms the ordinary turbo code bothnin AWGN and especially in fading channels. Thus, superorthogonal turboncodes are suited mainly for spread-spectrum applications. For thenpurposes of iterative decoding, we concisely describe the connectionnbetween the optimal maximum a posteriori symbol estimation andnsuboptimal soft-output decoding based on sequence estimation. Thensuboptimal decoder produces outputs that can directly be used asnadditive metrics at successive decoding iterations, without the need fornestimating channel noise variance. Simulation results in AWGN and flatnRayleigh fading channels are also presented, along with analytical uppernbounds of bit- and frame-error probabilities