Analysis of generalized retransmission backoff policies for slotted-ALOHA multiaccess channels

Backoff techniques, of which the well-known exponential backoff is a special case, are based on adaptation of average retransmission delay as a function of the number of collisions experienced by each message accessing the contention channel. An analytical model for slotted-ALOHA channels using general backoff functions has been developed and used to assess the performance advantages offered by either exponential backoff or alternative policies motivated by heuristic considerations. It is demonstrated that for a finite-source traffic model, the use of appropriate backoff policies can result in significant improvements in stable throughput-delay characteristics relative to nonadaptive systems. It is also shown that while several alternative classes of backoff functions with optimized parameters do not differ substantially in performance, exponential backoff may be outperformed by alternative policies such as the mu -law or step function.<>