Stochastic soft backpressure algorithms for routing and scheduling in wireless ad-hoc networks

摘要

We develop a generalization of the backpressure (BP) algorithm to find routes and schedules in wireless ad-hoc networks. Different from BP that schedules links that maximize queue differentials, the proposed stochastic soft (SS)BP algorithm randomizes schedules across links and flows with significant queue differentials. We show that SSBP shares the fundamental property of BP, namely, that if given arrival rates can be supported by some routing-scheduling policy, they can be supported by SSBP. Simulations illustrate SSBP's improvement in delay performance.