Information combining based ARQ in relay networks

In this paper, we present a novel centralized ARQ technique based on the lower bound on information combining. Contrary to conventional approaches where relays are selected according to the end-to-end quality of the link or decoding status of the relay, we select the minimum number of relays which suffice to allow the destination to successfully decode. Generally, it is believed that if a relay is unable to decode successfully it is useless, here we show that a relay with failures is also useful. We demonstrate that a system considering relays with successful decoding as well as with failures delivers much better performance than a system only considering successful relays for forwarding.