Abstract
Premature convergence has been recognized as one of the major drawbacks of particle swarm optimization (PSO) algorithms. In particular, the lack of diversity in PSO performance is an essential cause that commonly results in high susceptibility to prematurely converge to local optima especially in complex multimodal problems with high dimensionality. This paper presents a new PSO operational strategy based on gravity concept to address the aforementioned drawback and it is named as gravity-based particle swarm optimizer (GPSO). In addition, GPSO is further modified by adopting the cooperation concept of the conventional cooperative particle swarm optimizer (CPSO) to develop an extended version of GPSO called cooperative gravity-based particle swarm optimizer (CGPSO). Simulation results manifest that CGPSO performs satisfactorily on unimodal functions while it generally performs better on multimodal functions than GPSO and other conventional PSO variants. Finally, the proposed GPSO and CGPSO are applied into the problem of optimizing the detection performance of soft decision fusion for cooperative spectrum sensing in cognitive radio networks. For this problem, computer simulations show that the proposed CGPSO outperforms all other PSO variants in terms of quality of solutions whereas GPSO is found to be the best when the computational cost is taken into account.