Blind marginal maximum a posteriori sequence estimation for finite impulse response channels

In this paper, we propose an original full Bayesian approach for blind equalization of discrete signals degraded by a finite impulse response (FIR) channel and observed in additive noise. The sequence of discrete symbols is estimated according to a marginal maximum a posteriori (MMAP) criterion, the other unknown parameters are regarded as random nuisance parameters and integrated out analytically. An efficient off-lime algorithm to maximize the marginal posterior distribution is proposed. Using Monte Carlo simulations, the performance of this algorithm is evaluated in terms of bit error rate (BER)