Enhancing the Delay Performance of Dynamic Backpressure Algorithms

For general multi-hop queueing networks, delay optimal network control has unfortunately been an outstanding problem. The dynamic backpressure (BP) algorithm elegantly achieves throughput optimality, but does not yield good delay performance in general. In this paper, we obtain an asymptotically delay optimal control policy, which resembles the BP algorithm in basing resource allocation and routing on a backpressure calculation, but differs from the BP algorithm in the form of the backpressure calculation employed. The difference suggests a possible reason for the unsatisfactory delay performance of the BP algorithm, i.e., the myopic nature of the BP control. Motivated by this new connection, we introduce a new class of enhanced backpressure-based algorithms which incorporate a general queue-dependent bias function into the backpressure term of the traditional BP algorithm to improve delay performance. These enhanced algorithms exploit queue state information beyond one hop. We prove the throughput optimality and characterize the utility-delay tradeoff of the enhanced algorithms. We further focus on two specific distributed algorithms within this class, which have demonstrably improved delay performance as well as acceptable implementation complexity.