An Analysis of Delay-Constrained Opportunistic Scheduling for Cellular Wireless Systems

Base station schedulers in 3G and evolving 4G cellular systems use knowledge of the time-varying channel conditions of mobile users to exploit the multiuser diversity inherent in wireless networks. Although such opportunistic schedulers significantly improve the system throughput by scheduling users when their channel conditions are most favorable, they could degrade the user experience as a result of unfair resource allocation and increased variability in the scheduled rate and delay. The growing need to provide service differentiation between delay-sensitive multimedia traffic and non real-time data traffic over packet switched air-interfaces underscores the need for these schedulers to incorporate delay constraints. In this work, we focus primarily on the trade-off between the realization of multiuser diversity gain and the provision of delay guarantees. Our main contribution is an analytical characterization of the distributions of the delay and rate offered by an opportunistic scheduler. The scheduling metric used in the algorithm combines the rate requested by the user and scheduling delay in a general form. Our analysis of a wireless system with a finite number of users in discrete time is strongly supported by system simulations of a time-slotted cellular downlink shared by multiple mobile users with independent, fading channels. We also compute closed form expressions for the scheduler statistics using a continuous approximation. The results in this paper can be used to evaluate system performance and provision resources to support Quality of Service (QoS) guarantees in broadband wireless networks.