人工蜂群优化算法在复数盲源分离中的应用

Abstract

Aiming at the key technologies of number estimation and ordered recovery of complex-valued signals, a method of complex blind source separation with an unknown number of sources based on artificial bee colony optimization was proposed. Firstly, we introduced an algorithm based on cross validation technology to estimate number of sources. Then, the optimally extracted vectors were determined through maximizing absolute value of kurtosis by using artificial bee colony optimization, so as to separate complex-valued signals one by one. The simulation results show that the proposed method can achieve the blind separation for source signals of any distribution in decreasing order of absolute kurtosis, and compared to other conventional algorithms, this method has preferable estimating performancee. In addition, we considered the underdetermined complex blind source separation problem of instantaneous mixtures based on kurtosis. The cost function of the extracted vector in the underdetermined mixed case was constructed by exploiting the statistics properties of complex-valued source signals, and then artificial bee colony optimization algorithm was used to maximize the function to determine the optimally extracted vectors. The underdetermined complex blind source separation was achieved through many times of extracting. The simulation result of blind separation of different types of signals validates the feasibility of the proposed method.