Iterative decoding combined with physical-layer network coding on impulsive noise channels

Abstract

This thesis investigates the performance of a two-way wireless relay channel (TWRC) employing physical layer network coding (PNC) combined with binary and non-binary error-correcting codes on additive impulsive noise channels. This is a research topic that has received little attention in the research community, but promises to offer very interesting results as well as improved performance over other schemes. The binary channel coding schemes include convolutional codes, turbo codes and trellis bitinterleaved coded modulation with iterative decoding (BICM-ID). Convolutional codes and turbo codes defined in finite fields are also covered due to non-binary channel coding schemes, which is a sparse research area. The impulsive noise channel is based on the well-known Gaussian Mixture Model, which has a mixture constant denoted by α. The performance of PNC combined with the different coding schemes are evaluated with simulation results and verified through the derivation of union bounds for the theoretical bit-error rate (BER). The analyses of the binary iterative codes are presented in the form of extrinsic information transfer (ExIT) charts, which show the behaviour of the iterative decoding algorithms at the relay of a TWRC employing PNC and also the signal-to-noise ratios (SNRs) when the performance converges. It is observed that the non-binary coding schemes outperform the binary coding schemes at low SNRs and then converge at higher SNRs. The coding gain at low SNRs become more significant as the level of impulsiveness increases. It is also observed that the error floor due to the impulsive noise is consistently lower for non-binary codes. There is still great scope for further research into non-binary codes and PNC on different channels, but the results in this thesis have shown that these codes can achieve significant coding gains over binary codes for wireless networks employing PNC, particularly when the channels are harsh.