In this paper we describe a MAP sequence estimator for frequency-selective
fast-fading channels. The estimator incorporates prior channel knowledge
using a Gauss-Markov model of the FIR channel by marginalizing over
the resulting channel coefficient joint probability density function.
A computationally efficient PSP algorithm is also described, and its
performance illustrated. With simulations, we consider
the accuracy to which symbols can be estimate as influenced by the channel
fading rate, and we study the sensitivity of the Gauss-Markov channel
model based MAP estimator on inaccuracies in assumed model parameters.