Robust Turbo Decoding in a Markov Gaussian Channel

Abstract

Strong impulse noise is widely known to adversely affect conventional receivers designed only to consider background noise. Although sophisticated receivers offer substantial performance improvements, fully exploiting the impulse statistics, which are generally not time invariant and are difficult to model accurately, can be unrealistic. Alternatively, without making assumptions regarding the underlying impulse channel model, the authors' previously developed robust decoding schemes can achieve performance equivalent to that of their optimal counterparts in impulse noise channels. In this letter, a robust turbo decoding metric is proposed to address the inherent memory in an impulse channel: a 2-D trellis is used to adapt for channel state transitions when statistics on the memory impulse channel model are lacking. The simulation results verified the robustness of the proposed decoder in harsh environments.