Wireless networks are not only ubiquitous and of great economic importance to modern society, but also the source of many fundamental scientific challenges faced by modern communication networks. Radio resource control and optimization in wireless networks has received a lot of attention from both the research community and industries in the past few years. Distributed (iterative) algorithms, which are scalable and robust, are critical for large scale wireless networks consisting of randomly placed transmitter-receiver pairs (e.g., wireless ad hoc networks). One of the outstanding issues of the existing works on distributed resource control and optimization in wireless networks is that, the convergence behavior and optimality of the iterative algorithms are usually established under the...[ Read more ]

Wireless networks are not only ubiquitous and of great economic importance to modern society, but also the source of many fundamental scientific challenges faced by modern communication networks. Radio resource control and optimization in wireless networks has received a lot of attention from both the research community and industries in the past few years. Distributed (iterative) algorithms, which are scalable and robust, are critical for large scale wireless networks consisting of randomly placed transmitter-receiver pairs (e.g., wireless ad hoc networks). One of the outstanding issues of the existing works on distributed resource control and optimization in wireless networks is that, the convergence behavior and optimality of the iterative algorithms are usually established under the assumption that the convergence time of the iterative algorithms is shorter than the channel coherence time (i.e., the quasi-static channel model). In other words, the wireless channels are treated as constant when analyzing the behavior of the distributed resource control algorithms. However, as the iterative resource allocation algorithms take quite a lot steps to converge for large scale wireless networks and wireless channel is time-varying by nature, it is not realistic at all to use the quasi-static channel model to study the behavior of the iterative algorithms. In this thesis, we shall investigate the convergence behavior and tracking performance of the distributed resource allocation algorithms (e.g., the distributed scaled gradient projection algorithm) in wireless networks under time-varying fading channels. Specifically, we shall utilize the randomly switched system modeling to analyze the convergence behavior of the distributed algorithms and develop a novel dynamic scaling matrix adaptation to optimize the tracking performance of the distributed algorithms under time-varying fading channels (e.g., the FSMC model). As an illustration, two application examples shall be discussed in detail. The framework developed in this thesis can be applied to various scenarios of resource control and optimization in wireless networks.