Performance analysis of multiuser selection diversity

We study the performance of multiuser selection diversity based on absolute signal-to-noise ratio (SNR)-based scheduling and normalized SNR-base scheduling. Our performance criteria include system capacity as well as access probability, average access time, and average access rate of individual users. Closed-form expressions are obtained for these performance measures when the users are subject to independent but not necessarily identically distributed Rayleigh fading and when these users move at possibly different speeds. These results confirm that the system capacity for both scheduling schemes increases as the number of users in the system grows. These results also show that while the disparateness in the average channel conditions (i.e. short-term average SNR) among users is beneficial to the system capacity when the absolute SNR-based scheduling is used, this increase comes at the expense of a certain unfairness to weak (in average) users. On the other hand, the normalized SNR-based scheduling is fair in the sense that all users have the same chance to access the channel in spite of the disparateness in their average channel strength. However this fairness comes at the cost of a system capacity penalty.