Throughput Optimization of Cognitive Radio Networks

In this paper, we study the problem of finding the maximum throughput achievable by secondary users while primary users are guaranteed to achieve the minimum required data rates. We assume that node positions and channel side informations are available, and a virtual central unit is in responsible for scheduling both primary and secondary users. The optimal solution provides a theoretical upper bound of throughput for cognitive radio networks. An iterative algorithm proposed in [1] was extended to solve the throughput optimization problem, which can be decomposed into sub-problems and solved iteratively. However, due to the NP-hardness of Maximum Weighted Independent Set (MWIS) problem which has to be solved at each iteration, some greedy approximation method with polynomial computation time is applied in the iterative algorithm. Finally, we derive a theoretical upper bound for the distance between the greedy solution and the optimal solution, and guarantee that the optimal solution must be within this upper bound of the greedy solution. Numerical results are given to validate the solution.