A fundamental cross-layer approach to uplink resource allocation

The literature on multiaccess (uplink) communications has traditionally treated "network-layer" issues such as source burstiness and network delay apart from "physical-layer" issues such as channel modelling, coding, and detection. We establish a unified cross-layer framework for optimal resource allocation over multiaccess fading channels where packets arrive randomly to transmitters. Using optimal coding assumptions, we first show that the region of all stable arrival rates is the same as the information-theoretic capacity region of [D. Tse and S. Hanly, 1998]. Next, we show that a longest-queue-highest-possible-rate (LQHPR) allocation strategy stabilizes the multiaccess queueing system whenever stability is possible. Finally, we show that under symmetric conditions, the LQHPR policy also minimizes the average packet delay. Such a policy can be interpreted in the coding context as adaptive successive decoding. Since optimal coding assumptions are used, the throughput (delay) performance of the LQHPR policy provides a fundamental upper (lower) bound to the performance of all reliable multiaccess coding schemes. especially when the measurement noise is low.