Search

The underdetermined blind source separation (BSS), which based on sparse representation, is discussed in this paper; moreover, some difficulties (or real assumptions) that were left out of consideration before are aimed. For instance, the number of sources, , is unknown, large-scale, or time-variant; the mixing matrix is ill-conditioned. For the proposed algorithm, in order to detect a time-variant mixing matrix, short-time Fourier transform is employed to segment received mixtures. Because is unknown, our algorithm use more estimates to find out the mixing vectors by particle swarm optimizer (PSO); and then, surplus estimates are removed by two proposed processes. However, the estimated accuracy of PSO will affect the correctness of extracting mixing vectors. Consequently, an improved PSO version called the cluster guide PSO (CGPSO) is further proposed according to the character of sparse representation. In simulations, several real assumptions that were less discussed before will be tested. Some representative BSS algorithms and PSO versions are compared with the CGPSO-based algorithm. The advantages of the proposed algorithm are demonstrated by simulation results.

The particle swarm optimizer (PSO) is a population-based optimization technique that can be applied to a wide range of problems. This paper presents a variation on the traditional PSO algorithm, called the efficient population utilization strategy for PSO (EPUS-PSO), adopting a population manager to significantly improve the efficiency of PSO. This is achieved by using variable particles in swarms to enhance the searching ability and drive particles more efficiently. Moreover, sharing principals are constructed to stop particles from falling into the local minimum and make the global optimal solution easier found by particles. Experiments were conducted on unimodal and multimodal test functions such as Quadric, Griewanks, Rastrigin, Ackley, and Weierstrass, with and without coordinate rotation. The results show good performance of the EPUS-PSO in solving most benchmark problems as compared to other recent variants of the PSO.

Blind source separation (BSS) is a technique used to recover a set of source signals without prior information on the transformation matrix or the probability distributions of the source signals. In previous works on BSS, the choice of the learning rate would result in a competition between stability and speed of convergence. In this paper, a particle swarm optimization (PSO)-based learning rate adjustment method is proposed for BSS, and a simple decision-making method is introduced for how the learning rate should be applied in the current time slot. In the experiments, samples of four and ten source signals were mixed and separated and the results were compared with other related approaches. The proposed approach exhibits rapid convergence, and produces more efficient and more stable independent component analysis algorithms, than other related approaches.