Iterative decoding of BDPSK modulated LDPC codes using two-symbol-interval observations

This paper studies the detection and decoding of LDPC encoded, interleaved, BDPSK-modulated symbols on the correlated flat Rayleigh fading channel using two-symbol-interval observations. The log-likelihood metric is derived for sum-product decoders. For each hypothesized code bit, the metric depends on two consecutive received signals and only on two other variables: the received SNR per bit and the fade rate. We simulate the performance of the system with this metric, and compare it with that of other metrics that do not take into account the fade rate. Numerical results are given for different codes. In practice, the fade rate may not be precisely known at the receiver, so we also examine how the system performance is affected if the actual fade rate differs from the assumed fade rate.