摘要
针对复杂噪声环境下的参数估计问题,提出了一种稳健的自适应序贯M估计算法(Adaptive RecursiveM-Estimation,ARME),并从理论分析和Monte Carlo实验仿真两方面分析了该算法的收敛性、渐进无偏特性和稳健性.理论分析和仿真试验表明:在高斯白噪声背景下,ARME具有与序贯最小二乘算法(Recursive Least Square,RLS)相近的性能;在有突出干扰等非高斯噪声背景下,与RLS相比,ARME的参数估计收敛速度更快,估计误差更小,而且在稳健性上大大优于RLS.
To estimate the parameters in complex noise situation effectively, a new robustness algorithm for adaptive signal process is presented named adaptive recursive M-estimation （ARME）. And its properties of convergence, asymptotic unbiasedness and robustness are analyzed from theory and Monte Carlo simulations. Results and simulations show that the ARME algorithm obtains the advantages of the recursive least square （RLS） in Gaussian white noise （WGN）.More importantly,in the no-Gaussian noise situation＇such as WGN with high interference, they show that the ARME algorithm has a better convergence rate and a less estimation deviation than those of the RLS. Besides that, the ARME is more robustness compared with the RLS.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2007年第9期1651-1655,共5页
Acta Electronica Sinica
基金
国家"863"高技术研究发展计划课题(No.2006AA703402B)
作者简介
胡谋法,男，1979年生于湖北省石首市，现为国防科技大学ATR重点实验室博士生．主要研究方向光学信息处理、目标识别等．E-mail：hu199709_200106@sina．com
