Otimização Bioinspirada Aplicada a Separação Cega de Fontes no Contexto Post-Nonlinear

This article presents a proposal for the application of the Particle Swarm Optimization (PSO) algorithm in Blind Source Separation problems in the Post-Nonlinear context. The convergence properties of different swarm topologies are analyzed: global, square and ring. Nonlinearity removal is obtained by estimating the inverse function through Taylor series and the linear step is solved using FastICA. The mutual information is used as a cost function during the optimization process. The results show that the algorithm is able to recover the sources satisfactorily and that the square topology presents the best performance.