Code design for the relay channel
Doctor of Philosophy thesis
The design of wireless communication networks is based on the premise that networks are collections of reliable point-to-point links between communicating nodes. Such a communication model, although simple, is inefficient because wireless propagation is not point-to-point in nature. It is more efficient for nodes to share their resources by cooperating at the symbol level to forward each other's physical layer packets collectively to the destination. The improved paradigm is known as cooperative communication. My work develops practical coding techniques that approach the theoretical limits of cooperation predicted by information theory. Cooperative strategies, often called relay protocols, describe the processing performed by the information forwarding (relaying) nodes. My research proposes implementable schemes for two relay protocols - decode-and-forward and estimate-and-forward. In each case, the starting point is an optimal but non-constructive information theoretic random coding scheme, which motivates a practical code construction. Novel code design princi ples and some surprising insights emerge from this work of research. The performance of the each scheme developed here is found to approach theoretical limits. For decode-and-forward relaying, we propose dual-rate low-density parity-check (LDPC) codes. In our designs, the source transmission is decoded with the help of side information in the form of additional parity bits from the relay. The key challenge is to discover codes that simultaneously perform well for the source-relay and the source-destination links. The asymptotic noise thresholds of the discovered relay code profiles are a fraction of a decibel away from the achievable lower bound for decode-and-forward relaying. With random component LDPC codes, the overall relay coding scheme performs within 1.2 dB of the theoretical limit. In estimate-and-forward relaying, the key challenge is to form a quantized estimate of the source transmission from the received signal at the relay. We illustrate with an example that the existing approach of distortion minimization at the relay is suboptimal. We derive an improved quantizer design criterion based on rate-constrained mutual information maximization between the source transmission and the quantizer output, using which, we obtain performance less than 0.9 dB from the achievable rate at a BER of 10-4.
Engineering, Electronics and Electrical