Optimal decentralized control in the random access multipacket channel

A decentralized control algorithm is sought that maximizes the stability region of the infinite-user slotted multipacket channel and is easily implementable. To this end, the perfect state information case in which the stations can use the instantaneous value of the backlog to compute the retransmission probability is studied first. The vest throughput possible for a decentralized control protocol is obtained, as well as an algorithm that achieves it. These results are then applied to derive a control scheme when the backlog is unknown, which is the case of practical relevance. This scheme, based on a binary feedback, is shown to be optimal, given some restrictions on the channel multipacket reception capability