A Deep Study on Layered Multi-Relay Non-Orthogonal Amplify-Forward Networks

The paper introduces a cross-layer design for transmission of Successive Refinement (SR) source code inter-played with non-orthogonal layered code, deployed over a half-duplex multi-relay Non-orthogonal Amplify-Forward (NAF) network. Assuming the Channel State Information (CSI) is not available at the source node, firstly the achievable layered Diversity-Multiplexing Tradeoff (DMT) curve is derived. Then, by taking Distortion Exponent (DE) as the figure of merit, several achievable lower bounds are proved and the optimal expected distortion performance under high Signal to Noise Ratio (SNR) approximation is explicitly obtained. It is shown that the proposed coding can achieve the Multi-Input Single-Output (MISO) upper bound under certain regions of bandwidth ratios, by which the optimal performance in these regions can be explicitly characterized. Further the non-orthogonal layered coding scheme is extended to a multi-hop MIMO Decode-Forward (DF) relay network and set of DE lower bounds is derived. Simulation results confirm the effectiveness of the proposed scheme under general SNR values and compared with its counterparts.