WNET-GT · Analysis and Management of Wireless Networks with Selfish Users

The emerging use of wireless technologies for data communication has brought to focus novel system characteristics, such as power control, time variation in the channel quality and mobility of the network users. In addition to diversity in these physical parameters, mobile users are heterogeneous with regard to their Quality of Service (QoS) criteria, differently evaluating delay, throughput and power. Distributed mechanisms for wireless medium-sharing, branching from the celebrated Aloha protocol, have gained prominence due to their relative simplicity. However, Aloha-based protocols (such as 802.11x) are not fully adjusted to the distinctive features of the wireless channel. Moreover, the parameters of these protocols are identically configured among users, not allowing for QoS differentiation. Centralized scheduling algorithms, on the other hand, have been designed in order to consider the wireless specific features and to support QoS. Yet, high-quality schedulers require substantial data and computational resources, and might become hard to implement. Our research attempts to find a middle ground between fully distributed mechanisms and centralized scheduling algorithms, by allowing users to adjust their protocol parameters in order to accommodate their individual QoS requirements. Under the above framework, users can be seen as competing for scarce network resources. Accordingly, this research will employ (non-cooperative) game-theoretic tools for the analysis of multiple access wireless networks. Based on the analysis, we shall design distributed protocols, which allow users to obtain their required QoS, and still preserve network-wide objectives such as fairness. At a higher level, we shall develop efficient management schemes, with the objective of bringing network interference to a minimum. Finally, the designed protocols and management schemes will be evaluated under current network standards.