摘要

Based on weighted signal subspace fitting and fractional lower order covariance matrix, a fractional lower order covariance - weighted signal subspace fitting (FLOC-WSSF) algorithm is proposed. The proposed algorithm can reduce impaction of impulsive noise, while significantly improving the performance of the original WSSF algorithms. In order to fit the proposed direction finding algorithm of FLOC-WSSF, a cultural shuffled frog leaping (CSFL) algorithm is proposed and applied to objective function of direction finding. In the proposed CSFL algorithm, a cultural mechanism is used for the better candidates in the population to progress and shuffled frog leaping for the worst candidates in the population to move towards the best candidates. The merit of the algorithm lies in the fact that it avoids stagnation and has a very fast convergence speed. Monte-Carlo simulations have proved that the proposed CSFL-FLOC-WSSF method has some good performance such as high-precision solution and the capability to find coherent signal sources.

全文