Optimal Strategies for Cooperative MAC-Layer Retransmission in Wireless Networks

The concept of cooperative retransmission in wireless networks has attracted considerable research attention. The basic idea is that when a receiver cannot decode a frame, the retransmission is handled not by its original source but rather by a neighbour that overheard the transmission successfully, and may have a better channel to the destination. However, the majority of existing literature tackles the issue from the physical layer perspective, with either a single cooperating neighbour, or a multiple-neighbour setting where the receiver is capable of combining and decoding the signal from several simultaneous retransmissions. In this paper, we consider the case of multiple cooperating neighbours from a MAC-layer perspective. Thus, we assume a receiver that can only decode one transmission at a time, while multiple simultaneous retransmissions (by several neighbours that had overheard the frame successfully) will cause a collision. As a result, each neighbour with a successfully overheard copy of the frame faces a tradeoff between helping with a cooperative retransmission and possibly causing a collision. Accordingly, we pose the optimization problem of finding a distributed randomized strategy for the cooperating neighbours, which assigns a certain retransmission probability to every neighbour in each time slot, so as to minimize the expected latency until successful reception. We analyse the performance achieved by two approaches: one where the original source is silent while the neighbours conduct their cooperative retransmissions, and another where both the source and the neighbours may have a nonzero retransmission probability simultaneously. We show that the latter approach offers a significant performance improvement over the former one, as well as either traditional retransmission or two-hop routing to the destination.