Network-coded SIR-based distributed coding scheme: A new soft estimate modelling and performance analysis

In this paper, we consider a multiple access relay channel (MARC) that operates in the decode and forward mode with network-coded soft information relaying (NC-SIR). We show that the combination of channel coding and NC-SIR in the MARC can be seen as a distributed coding scheme (DCS). In order to mitigate the error-propagation at the relay, we propose a new method for modelling the log-likelihood-ratio (LLR)-based soft estimated symbols at the output of the BCJR encoder. The calculation of the received LLRs at the destination is derived according to the proposed model. Moreover, a virtual one-hop link, the equivalent to the sources–relay–destination channels, is provided in terms of signal-to-noise ratio and related statistic. This allows us to give a tractable analysis on the bit error rate (BER) performance. The expression of the pairwise error probability under fully interleaved Rayleigh fading channels and an upper bound on the BER are derived analytically. Several simulation results show that the NC-SIR-based DCS using the proposed model outperforms those using other well-known models in terms of BER performance. Moreover, simulation results are presented to demonstrate the accuracy of the proposed bound for different relay positions between source and destination nodes.