Particle Swarm Optimization (PSO) of power allocation in cognitive radio systems with interference constraints

Cognitive radio is used for enhancement of spectrum efficiency. Although many works have been accomplished on the power allocation of cognitive radio, limited efforts have considered evolutionary algorithms. In this paper, we study this problem in the cognitive radio networks where interference constraints are defined for protection of quality of service (QoS) for both primary and secondary users. Utilities defined as functions of the signal-to-interference-plus-noise ratio (SINR) are matched for each secondary user which meets Nash's axioms. In general, the region of utilities that meets the constraints is non-convex. It is possible to make simplifications, generate a convex region, and then use common convex optimization approaches to obtain a solution. However, Particle Swarm Optimization (PSO) does not need such simplifications and thus its results are superior to those of the convex optimization methods. PSO is an evolutionary algorithm based on social intelligence, utilized in many optimization problems. PSO is a global optimizations algorithm that does not require the objective function be differentiable as required in classic optimization methods.