Harnessing the Channel-Capture Phenomenon in Slotted Aloha for Achieving the Optimal Tradeoff Between Throughput and Short-Term Fairness

The channel-capture phenomenon has long been observed and deemed undesirable in random access networks. When it occurs, one node would monopolize transmission for an extended period despite the contention from others, leading to high network throughput but serious short-term unfairness among nodes. Harnessing the throughput gains without compromising the fairness requirements requires a comprehensive understanding of the throughput-fairness tradeoff caused by the channel-capture phenomenon, which unfortunately cannot be accounted for by the existing analytical models. The problem is rooted in a key assumption of state-independent probability of successful transmission of Head-Of-Line (HOL) packets that is adopted in the existing models, but no longer holds when the channel-capture phenomenon occurs. In this paper, by incorporating capture states into the HOL-packet model for slotted Aloha, the probability of successful transmission of HOL packets is shown to be crucially dependent on whether they are in a capture or non-capture state. The analysis reveals that with the simplest collision receiver, the maximum network throughput of slotted Aloha can reach 1 as long as the number of capture states is no smaller than 2. The short-term fairness, on the other hand, would be worsened with more capture states. The optimal tradeoff between network throughput and short-term fairness is characterized, and shown to be achieved by the proposed Double-Capture Backoff.