MAC Scheduling With Low Overheads by Learning Neighborhood Contention Patterns

Aggregate traffic loads and topology in multihop wireless networks may vary slowly, permitting MAC protocols to “learn” how to spatially coordinate and adapt contention patterns. Such an approach could reduce contention, leading to better throughput. To that end, we propose a family of MAC scheduling algorithms and demonstrate general conditions, which, if satisfied, ensure lattice rate optimality (i.e., achieving any rate-point on a uniform discrete lattice within the throughput region). This general framework enables the design of MAC protocols that meet various objectives and conditions. In this paper, as instances of such a lattice-rate-optimal family, we propose distributed, synchronous contention-based scheduling algorithms that: 1) are lattice-rate-optimal under both the signal-to-interference-plus-noise ratio (SINR)-based and graph-based interference models; 2) do not require node location information; and 3) only require three-stage RTS/CTS message exchanges for contention signaling. Thus, the protocols are amenable to simple implementation and may be robust to network dynamics such as topology and load changes. Finally, we propose a heuristic, which also belongs to the proposed lattice-rate-optimal family of protocols and achieves faster convergence, leading to a better transient throughput.