Trellis turbo-codes in flat Rayleigh fading with diversity

This paper compares the performance of two schemes for joint channel estimation and turbo-decoding of high-rate trellis turbocodes in flat Rayleigh fading, where antenna diversity is available at the receiver. The first method relies on iterative quantized phase estimation, and the second on optimum filtering of pilot and coded symbols. For Doppler rates of practical interest, simulations indicate that optimum pilot filtering is superior to approximating the channel phase with a quantized Markov model. The optimum filtering approach also has lower complexity. For an absolute measure of performance, proximity to channel capacity is also discussed. Iterative joint channel estimation and turbo-decoding with either method is demonstrated to achieve almost all the capacity gain due to receiver antenna diversity.