A modified bees colony algorithm for separating overlapping communication signals in time and frequency domain

To separating the overlapping communication signals in time and frequency domain, wireless communication system constantly put forward higher demands for the performance of separation algorithm (convergence rate, separation precision) in order to achieve better separating effect. Current algorithm cannot satisfy these new demands. Recently Ebrahimzadeh proposed to use bees colony algorithm to separate signals for better performance of convergence rate and separate precision. This paper studies the bees colony algorithm deeply, and in view of the problems that the algorithm mainly depends on the method of `casting net´ and lack clear direction in the process of optimization, which generally leads to large computational load and some blindness, this paper proposes a modified bees colony algorithm with the `directional indication´ called Gradient-BCA (GBCA). GBCA takes the `directional indication´ based on gradient into the process of seeking optimization. Simulations suggest that GBCA has the better performance of convergence rate and separation precision than BCA and classic algorithm based on gradient. GBCA is more practical in separating the overlapping communication signals in time and frequency domain in wireless communication for the demand of the better performance of convergence rate and separation precision.