摘要
为解决盲源分离算法中收敛速度和稳定性的折中问题,基于最优步长的思想,提出了一种新的步长自适应的自然梯度盲分离算法.在自然梯度盲分离算法的基础上,对步长进行自适应迭代,步长偏移量的选取原则是使得下一次迭代时的步长最优,或者说目标函数最小.仿真结果表明,提出的算法相对固定步长自然梯度算法,其收敛速度提高了1倍以上,而系统的稳定性能基本不变.
The convergence speed and stabiliy of the blind source separation（BSS） algorithm are subjected to people′s extensive concern.And the key factor that affect them is the selection of iteration step-size.To obtain the tradeoff between convergence speed and stablility of the BSS algorithm,a novel step-size adaptive blind source separation algorithm is proposed in the paper which is based on the idea of optimal step-size.Based on the natural gradient BSS algorithm,the algorithm carry on adaptive iteration to step-size.And the selective criterion of the step-size offset is to make the step-size of the next iteration optimum or the objective function minimum.The algorithm simulation result shows that the proposed algorithm increases the convergence speed more than one times relative to natural gradient algorithm with fixed step-size and the stable performance is not deteriorated with the enhancing of the convergence speed.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第10期18-20,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
通信抗干扰国家重点实验室基金资助项目
关键词
通信信号
盲源分离
收敛速度
步长自适应
communication signal
blind source separation
convergence speed
step-size adaptive
