This project is motivated by the lack of theoretical understanding of distributed control strategies to jointly optimize cognitive radio network functionalities at multiple layers of the networking protocol stack. When physical layer parameters are considered, the resulting optimization problem is often nonconvex, and therefore cannot be solved efficiently in general even through centralized algorithms. The objective of this project is then to look at the cross-layer design of cognitive radio multi-hop networks from a different and novel perspective. The problem of jointly optimizing the sensing parameters, multiple-access scheme, routing, and rate control is formulated as a bi-level equilibrium problem, which can be solved efficiently and in a distributed way. Successful completion of this research will provide the community with a family of novel distributed algorithms that naturally implement vertical and horizontal decompositions across the network. This would represent a major departure from well-understood decomposition algorithms for network utility maximization problems that are known to suffer from slow convergence speed.
The project will contribute to the development of novel algorithms and tools that will lay the foundation for the next generation of cognitive networking technology. The impact of the proposed technology to today's society can be as broad and as strong as the potential applications of ad hoc networks; for example, search-and-rescue operations, on-demand communications infrastructure deployment, and intelligent transportation highways and vehicles. The proposed research is integrated with an educational plan designed to train the next-generation of networking professionals, as well as promote cross-fertilization of academic research with industry needs.
