Channel estimation and differential detection for MPSK signaling in time-varying flat-fading communication channels

Abstract

This paper addresses the issue of capacity achieving reliable information transfer over time-varying flat-fading com- munication channels. We use M-state, M-ary symmetric finite- state Markov channels (FSMC) to model time variations of flat- fading channels. We analyze the conventional (separate) approach to channel parameter estimation and data detection by using decision-feedback and output-feedback FSMC estimators. Our analysis includes the metric function update analysis for the decision-feedback estimator and the mutual information rate penalty caused by the input signal entropy reduction, for the output-feedback estimation. Then, we consider the implementa- tion of differential detection instead of channel estimation. We show that differential detection transfers the memory of the channel process into a latent form, which does not interfere with the operation of standard ML coding for memoryless channels. Furthermore, we show that multiple-symbol differential detection practically achieves the channel information capacity with observation times only on the order of a few additional symbol intervals.