A Cross-Layer View of Optimal Scheduling

Abstract

The problem of joint scheduling and rate control for multicast traffic in wireless networks is considered under the performance objectives of sum throughput maximization and proportional fairness. Our results are also valid for the special cases of unicast and broadcast traffic. First, the problem of maximizing the sum throughput of the network is studied and an optimal scheduling and rate control policy is obtained. Given the combinatorial complexity of providing an optimal policy, a simple, polynomial-time, suboptimal alternative scheme is introduced that restricts the space of scheduling and rate control decisions to operation one at a time or all together. The optimal policy to the problem of maximizing the sum throughput of the network with respect to this restricted action space is found. Next, the objective of proportional fairness is considered. Under this restricted space of actions, the resulting scheduling and rate control policy is explicitly characterized analytically and the effects of the current channel conditions are incorporated into the scheduling decisions. Furthermore, it is shown that the policy under this restricted action space is of threshold type. Finally, our analytical results are verified through a set of numerical experiments.