Is Our Model For Contention Resolution Wrong? Confronting The Cost Of Collisions

Randomized binary exponential backoff (BEB) is a popular algorithm for coordinating access to a shared channel. With an operational history exceeding four decades, BEB is currently an important component of several wireless standards.Despite this track record, prior theoretical results indicate that under bursty traffic (1) BEB yields poor makespan and (2) superior algorithms are possible. To date, the degree to which these findings manifest in practice has not been resolved.To address this issue, we examine one of the strongest cases against BEB: n packets that simultaneously begin contending for the wireless channel. Using Network Simulator 3, we compare against more recent algorithms that are inspired by BEB, but whose makespan guarantees are superior. Surprisingly, we discover that these newer algorithms significantly underperform.Through further investigation, we identify as the culprit a. awed but common abstraction regarding the cost of collisions. Our experimental results are complemented by analytical arguments that the number of collisions - and not solely makespan - is an important metric to optimize. We argue that these findings have implications for the design of contention-resolution algorithms.