Near-Optimal Spectrum Allocation for Cognitive Radio Networks

This paper investigates network-wide spectrum allocation based on the cross-entropy (CE) method in cognitive radio network (CRN). A few spatial spectrum allocation techniques in CRN have been proposed in the literature. However, the optimum spectrum allocation requires an exhaustive search over all combinations of available channels and constraints on secondary users, whose complexity increases exponentially with the number of users and channels. Simulation results show that the proposed modified CE-based scheme is an efficient method to greatly reduce the complexity while still reaching optimal network utility. Most of the solutions generated by the modified CE algorithm are equal to the optimal values.