Optimal Joint Viterbi Detector Decoder (JVDD) over AWGN/ISI channel

Communication channels today use a state-of-the art iterative detector/decoder system on its receiver end to detect and decode the transmitted bits. This iterative detection system is comprised of a soft output detector, either the soft output Viterbi algorithm (SOVA) or the Bahl, Cocke, Jelinek and Raviv (BCJR) algorithm, and the Sum Product Algorithm (SPA) is used in the decoder. Although iterations of the soft information between these detector and decoder blocks gives rise to good performance over an inter-symbol-interference (ISI)/additive white Gaussian noise (AWGN) channel when the codeword length (CWL) is large, the iterative detector is sub-optimal. This suboptimality originates from the SPA algorithm that itself is suboptimal whenever there are cycles in the factor graph, in particular, when there are short cycles. Any practical code will have cycles in its factor graph. A second source of suboptimality is the iterative process itself. There exist iterations both within the SPA decoder and between the decoder and the detector. In this work, the authors propose a novel detection/decoding algorithm coined the Joint Viterbi Detector Decoder (JVDD) that functionally replaces the iterative detector/decoder in the channel. Unlike the iterative detector/decoder, the proposed algorithm performs detection and decoding on a single structure and is optimal over an ISI/AWGN channel when there are sufficient computational resources. In this work we describe the JVDD algorithm and perform preliminary analysis on its performance and complexity under various conditions.