How much feedback is multi-user diversity really worth?

Abstract

Wireless scheduling algorithms can extract multi-user diversity (MUDiv) via prioritizing the users with best current channel conditions. One drawback of MUDiv is the required feedback carrying the instantaneous channel rates from from all active subscribers to the access point/base station. This paper shows that this feedback load is, for the most part, unjustified. To alleviate this problem, we propose a technique allowing to dramatically reduce the feedback (by up to 90%) needs while preserving the essential of the scheme performance. We provide a theoretical analysis of the feedback load as function of the system's ergodic and outage capacity for both the traditional MUDiv scheme and the new scheme.