Skip to Main Content
The paper studies the effects of utilizing multiuser diversity in realistic settings with limited feedback and uncertain traffic loads. A simple optimal method for setting rate quantization levels is derived and it is shown that an added benefit of multiuser diversity is that channel feedback can be heavily quantized without performance loss. In the limit as the number of users goes to infinity, even the most severe rate quantization (2 levels) does not reduce throughput at all from the case with unlimited rate resolution. Some important implications for the joint utilization of multiuser diversity and multi-antenna techniques are addressed. Further, a throughput-optimal scheduling policy for uncertain and bursty traffic loads is derived. An approximate policy is given with attractive computational complexity.