一种新的通信信号盲分离算法-EASIBSA算法

Novel Algorithm for Communication Signal Blind Separation—EASIBSA Algorithm

针对通信信号盲源分离的收敛速度问题,在EASI算法的基础上,提出了一种基于参数自适应优化思想的步长自适应盲源分离算法—EASIBSA算法。该算法将EASI算法中的学习速率与目标函数联系起来,采用随机梯度的方法自适应地进行调节,避免了最优学习速率的选择问题,同时提高了算法的收敛速度。实验仿真结果表明,EASIBSA算法在不影响分离效果的前提下,将收敛速度提高了近2倍。

Aim at the problem of the communication signal blind source separation,a novel blind separation algorithm based on step size adaptation(EASIBSA) to increase the convergence rate of the separation algorithm was present by using the idea of parameter adaptation optimization,which was based on Equivariant Adaptive Separation via Independent(EASI)algorithm.This algorithm related the learning rate with the object function and adapted it by using the stochastic gradient method,which avoided the selection of optim...