Robust decoding for OFDM systems in memory impulse channels

Numerous studies have shown that a generic communication system is prone to substantial performance loss in the context of strong impulse noise, which is generally not time invariant and is hard to model accurately. A turbo coded orthogonal frequency division multiplexing system in memory impulse channels is investigated in this work. Upon reception of the transmitted data, each of the two proposed receivers aims at detecting corrupted samples associated with impulse noise using the maximum a posteriori rule-with an adaptation to the inherent memory in the impulse channel. Sequentially, the refined receiver, assuming full statistics of impulse noise, clips excessively large samples-identified at the detection stage-with a proper clipping threshold exhaustively sought offline. In contrast, an efficient receiver, which lacks access to using impulse statistics, trades off a slight performance loss (as evidenced by simulation results) for complexity reduction attributable to forgoing the clipping operation as well as memory impulse channel model estimation.