The focus of this project is on optimal resource management and control of distributed cognitive radio (CR) networks, with the goal of harvesting the benefits of dynamic spectrum sharing that include improved spatial reuse (i.e., higher network throughput), programmable connectivity, and increased network availability. The research agenda includes centralized analytical formulations that aim at optimizing the operation of the bottom three layers as well as distributed routing and medium access protocols that implement the outcomes of such optimization. Depending on the channel dynamics (e.g., magnitude of the channel coherence time relative to the optimization window), optimizations based on deterministic-control formulations (for slowly varying channels) as well as stochastic-control formulations (for fast varying channels) are considered. In both cases, the underlying setup allows for multi-channel, multi-path routing at either the packet or the session levels. Distributed optimizations are also considered for peer-to-peer (ad hoc) hybrid networks, consisting of cognitive and "legacy" nodes. For this scenario, game-theoretic pricing policies are used to manage the interactions between CR and non-CR nodes for the purpose of achieving optimal routing performance. To validate the findings of this research, the proposed protocols will be implemented on a prototype CR/SDR platform. The proposed optimization framework provides a general methodology that can be applied to significantly improve the performance of resource constrained networks. Given the widespread use of wireless devices, this research is expected to have a strong impact on the performance of many wireless applications, including sensor networks, mesh networks, military radios, wireless LANs, etc.
