Measurement-based Opportunistic Feedback and Scheduling for Wireless Systems

We study opportunistic feedback of channel state and scheduling of users transmissions at a wireless access point. Our focus is on a regime where users have unknown possibly slowly-varying channel characteristics and it is desirable to limit the resources expended to exchanging channel state opportunistic. The paper argues that opportunistically schedul-ing the user whose current rate is in the highest quantile with respect to an empirical distribution has substantial advantages. Indeed by comparing to other measurement-based opportunistic scheduling schemes it systematically achieves high ‘opportunism ’ and thus throughput in fair and robust manner. Further the throughput penalty resulting from esti-mating users distributions is limited as long as the number of independent samples grows linearly with the number of active users. Second, a maximum quantile based scheduling scheme is consistent the use of a simple opportunistic feedback strategy that can be used to substantially reduce the overheads associated with deciding which user should in fact be scheduled at any point in time. Using both analytical and simulation results we will show that the benefits can be substantial for a variety of practical wireless system scenarios.