Sponsor: National Science Foundation

Title: Collaborative Research: Opportunistic Scheduling for Multimedia Wireless Systems

Investigator(s): Edwin K. Chong

Abstract
This collaborative project focuses on network-level solutions for exploiting the time-varying nature of the wireless environment to increase the overall performance of the system taking into account quality of service (QoS) and fairness for Multimedia Wireless Systems. The proposed algorithms are classified "opportunistic" since they consider channel conditions (and related definitions of per-user utility). The following research activities will be particularly pursued: opportunistic scheduling with and without explicit delay requirements as well as opportunistic scheduling in the frequency domain. The objective related to opportunistic scheduling without explicit delay requirements is to develop scheduling algorithms that exploit the time-varying nature of the wireless channel to optimize network performance under different fairness requirements. In particular, the channel estimation problem will be explored for which fast methodologies to effectively track the system parameters through stochastic approximation types of techniques will be developed. For opportunistic scheduling with explicit delay requirements, the goal will be to develop scheduling mechanisms for real-time applications such as voice and video where the objective is not only to schedule users based on their channel conditions and fairness constraints, but also based on their deadlines. In particular, the problems of opportunistic scheduling without individual QoS constraints and problems of opportunistic scheduling with individual QoS requirements will be studied. For that purpose, the off-line optimal scheduling solutions that provide benchmarks in evaluating other schemes will be developed. On the other hand, the on-line algorithms that take into account network traffic statistics will be developed and proven on practicability. Finally, the activities related to opportunistic scheduling in the frequency domain will consider taking advantage of both time-domain and frequency-domain diversity in wireless systems. In particular, the problems will be explored where the objective will be to maximize the system performance where power could be an explicit constraint or a penalty function. Other types of problems that will be explored here will deal with minimizing the transmitted power subject to performance constraints.