Throughput analysis of user identification for wireless local area networks

Multi-user diversity can be exploited in a wireless communication network to significantly increase system throughput through allocating the channel to the user with the best instantaneous channel condition. However, the overhead incurred in polling channel state information (CSI) in a large network can outweigh the gain of multi-user diversity. In our previous work, a user identification approach (UIDA) was proposed to reduce the system overhead. In this paper, we present a theoretical upper bound of the throughput of UIDA over general channels. Computer simulations show that the expected throughput of UIDA matches the theoretical upper bound closely over frequency-selective fading channels. Also, we apply UIDA in 802.11a systems. Simulation results show that it achieves better throughput performance than that of Medium Access Diversity (MAD) and Multi-user Diversity with Capture (MDC) in the literature, when there are a reasonably large number of mobile stations.