| Summary: | 碩士 === 國立高雄大學 === 資訊工程學系碩士班 === 97 === In IEEE 802.11 wireless local area networks, overall throughput will be decreased if most of WSs are associated with a few APs while other APs carry only light load or are idle. This is called load unbalancing problem. Another cause of throughput degradation is performance anomaly, which refers to the phenomenon that when multiple WSs associate to one AP with different link rates, the throughput of WSs transmitting at high link rate is degraded below the level of the lower-rate throughput. Few load balancing schemes have considered the impact of performance anomaly on throughput. Consequently, overall throughput may not be improved significantly even if the load between APs is completely balanced. This thesis uses game theory to analyze various ypes of AP association schemes. Particularly, we present complete-information static repeated AP selection game and complete-information dynamic one-shot AP selection game. We have proven that there must be Nash Equilibriums in complete-information static repeated game and show how to find Nash Equilibriums in this game. In addition, we have proposed a new scheme for
complete-information dynamic one-shot game. We also compared experimentally
complete-information and incomplete-information dynamic one-shot games in terms of throughput and fairness.
|
|---|
