Grouped Quantum-Inspired Particle Swarm Optimization Algorithm for Spectrum Allocation in Heterogeneous Network

Abstract

Up to now, classical Quantum Particle Swarm Optimization Algorithm in the late period of convergence has showed some drawbacks, such as population diversity reduce, convergence speed slow down and easy to fall into local optimal solution. This paper improves the classic QPSO algorithm and proposes Grouped Quantum-inspired Particle Swarm Optimization (G-QPSO). In this algorithm, quantum particles are grouped and regrouped periodically. Synthesizing the group optimal solution and the overall optimal solution, we update the speed and position of every quantum particle. We consider each solution vector as a viable spectrum allocation scheme and select the best one to achieve maximum value of Max-Sum-Reward (MSR) or Max-Proportional-Fair (MPF). As is shown in the simulation results, compared with the Genetic Type Algorithm, traditional Particle Swarm Optimization and Color-Sensitive-Graph Coloring Algorithm, this algorithm has better performance on the convergence speed and convergence precision, and avoids falling into the local optimal solution effectively.