A soft decision metric for coherently detected, differentially encoded MPSK systems

This paper defines the maximum likelihood (ML) metric for soft decision decoding of forward error correction (FEC) codes on DEPSK systems. Analysis over an AWGN channel shows that the ML soft decision metric provides approximately a 2.0 dB improvement over hard decision FEC decoders for DEPSK systems at a BER of 10^-6, which is consistent with results obtained for the coherent case. Unlike coherent systems, the ML soft decision metric for DEPSK systems is computationally burdensome. Hence, an easily implemented suboptimal soft decision metric that induces negligible loss to BER performance is proposed