List-output priority-first sequential decoding for physical-layer network coding

Abstract

This paper considers the decoding problem for convolutionally coded physical-layer network coding (PNC) over the two-way relay channel. The block error rate (BLER) performance and decoding complexity are two important design criteria. However, most existing decoding algorithms for the relay have either unaffordably high complexity or relatively poor BLER performance. The main contribution of this work is to introduce a novel decoding algorithm based on the idea of sequential decoding and syndrome trellis decoding for the relay. Our algorithm can achieve the same BLER performance as the full-state Viterbi algorithm (FSVA) and can be readily extended to generate list outputs. If a non-trivial list size (i.e., bigger than one) is adopted, the resultant list output decoding algorithm outperforms FSVA on block error rate performance. The simulation result shows that using a 64-state zero-tailed convolutional code, the average decoding complexity (in terms of the number of branch computations) of the proposed algorithm is only a few percents of that of the FSVA at a BLER below 0.001 (i.e. E b /N 0 above 4 dB).