Delay-optimal rate allocation in multiaccess communications: a cross-layer view

Abstract

The literature on multiaccess communications has traditionally treated "network-layer" issues such as source burstiness, network delay, and buffer overflow, apart from "physical-layer" issues such as channel modeling, coding, and detection. The recent work of Telatar and Gallager [I. E. Telatar and R. Gallager, Combining Queueing Theory with Information Theory for Multiaccess, August 1995] [I.E. Telatar, Multiple Access Information Theory and Job Scheduling, 1995] have sought to bridge this unfortunate division. We extend this line of inquiry by examining a multiaccess communication scenario where users' packets arrive randomly into separate queues and transmission rates are allocated from the information-theoretic multiaccess capacity region based on the respective users' queue states. In the symmetric case, a longer-queue-higher rate (LQHR) allocation strategy is shown to minimize the average system delay of packets. Such a policy can be interpreted in the coding context as adaptive successive decoding. The delay performance of the LQHR policy provides a fundamental lower bound to the performance for multiaccess coding schemes which seek to meet any given level of decoding error probability.