Soft-encoding distributed coding for parallel relay systems

In the paper, a new distributed coding scheme for parallel relay systems is proposed, in which a sender communicates to a destination that is two hops away via two (or more) parallel relays. The key idea is the exploitation of a (rate-1) soft convolutional encoder at each of the parallel relays, to collaboratively form a simple but powerful distributed analog coding scheme to achieve efficient forwarding of soft reliability messages. We detail the encoding and decoding process of the proposed soft-encoding distributed coding. As the input of the encoder would affect the overall performance, we analyze what form of messages at the relay is most appropriate to be forwarded to the destination. The range-limited log likelihood ratio (range-limited LLR) is chosen as the input. The optimality of the range-limited LLRs as the best form of relaying messages is verified by the simulation results. Our new distributed coding scheme can obviously outperform the existing ones.