摘要
给出了一种服从AR过程的随机信号,利用维纳滤波器通过观测方程来测量该信号,研究了维纳滤波器的阶数、随机信号采样点、噪声方差及滤波的均方误差之间的联系。仿真结果表明:在噪声方差一定的条件下,阶数与采样点的一半较接近时,阶数越大,均方误差越小,其总体均方误差小于阶数与采样点差距较大时的均方误差,但是滤波器可能存在畸变现象;阶数一定的条件下,噪声方差较小时,均方误差性能对噪声方差比较敏感,当噪声方差超过一定值时,均方误差稳定趋于某个区间值。
An autoregressive （AR） model of random signal is given to be measured by the observation equation of Wiener filters. This paper discusses the relationships among the order numbers of Wiener filter, the sampling numbers of random signal, the noise covariance and the mean square error of the Wiener filter algorithm in detail. The simulating results show that the increase of order numbers can not ensure to reduce the mean square error with improper choice of the parameters, and noise covariance is more sensitive to mean square error of the Wiener filter algorithm with low noise covariance. Also mean square error will converge to a stationary scope as noise covarlance goes to a setting value.
出处
《中国新通信》
2008年第11期47-50,共4页
China New Telecommunications
基金
国家自然科学基金(40776022)
国家863计划专项经费(2006AA09Z108)资助
关键词
维纳滤波器
阶数
采样点
均方误差
Wiener filter, filter order number, sampling number, mean square error
