Approximate maximum likelihood decoding for a coded system with pulse jamming

In a convolutionally encoded system with worst case pulse jamming, a coherent receiver using a soft decision decoder may give very poor performance. This occurs when the jammer chooses its duty cycle to maximize the probability of error and the decoder uses the common correlation metric, which is the maximum likelihood metric for a uniform channel. The maximum likelihood metric for a two state channel is derivable but is too complex to be used in a practical system. In this paper a simplified approximation to the maximum likelihood metric for a two state channel is introduced, and the performance of a system using this metric is analyzed using the Chernoff bound. It is shown that a system using this metric gives better performance in the presence of worst case pulse janmming than either a system using the common correlation metric, a system using hard decision decoding, or a system using a clamped version of the common correlation metric.