High performance scheduler design using multiuser diversity in high rate wireless packet networks

Abstract

The demand for high data rates in wireless communication networks is increasing in a rapid pace. Satisfying such an uprising demand is quite a difficult problem with current technology. Therefore, various researches are being conducted in each communication layer to increase wireless network capacity. One of the most notable techniques is the multiuser diversity scheduling scheme. This kind of scheduling scheme drastically increases network capacity, however, it has some drawbacks such as heavy feedback load, fairness problem, and resource allocation problem. In this thesis, we propose the high performance wireless packet scheduler which overcomes the problems of the conventional multiuser diversity scheduler. Multiuser diversity exploits the independence of the channel fading among mobile stations. This scheduling scheme grants medium access only to the mobile station having the best channel condition. This scheduling scheme increases wireless network capacity, however, it has several problems. First, mobile stations send feedback information to the base station. Hence, the feedback load linearly increases with the number of mobile stations. Moreover, the mobile station located close to the base station can monopolize network resource resulting in severe fairness problem. Second, the exact performance analysis for multiuser diversity scheduler is a very challenging task. We should consider the input traffic model, fading channel model, adaptive modulation and coding scheme, and finite queue size in the analysis. The resource allocation scheme and admission control policy is also an open problem. Supporting various QoS categories in multiuser diversity system is a big challenge because this scheduling scheme increases delay variation. The resource allocation problems should be addressed before we support various QoS categories. Allocation the adequate resource is very crucial object in operating wireless networks. In this thesis, we analyze the performance of m...