Belief propagation over frequency selective fading channels

Abstract

In this paper, we propose an iterative belief propagation (BP) based equalizer for frequency selective fading channels as an alternative to the typically used maximum a-posteriori (MAP), Viterbi, or soft output Viterbi algorithm (SOVA) based equalizers. The proposed equalizer has a parallel structure unlike the trellis based equalizers and it is less complicated than the MAP equalizer. Via simulations we show that, over frequency selective fading channels, a BP based equalizer achieves the optimal maximum a-posteriori performance, even in the presence of the length 4 cycles. We also discuss the joint decoding of an outer error correcting code and the equalization of the frequency selective fading channel using a joint factor graph and a BP based equalizer.