论文部分内容阅读
就均方误差含义下的最佳角度来说,能从已知输入统计特性导得的 Wiener 滤波器是最佳线性滤波器。当这些统计特性是未知的时候,则采用最小均方误差(LMS)自适应滤波器就能使过滤问题获得极好的结果。梯度下降算法引入了一个正比于自适应滤波器长度的超量均方误差。在自适应基阵处理器中,LMS滤波器能构成一个从指定的接收波束中部分地去除旁瓣干扰源的噪声抵消器。本文推导自适应横向滤波器的最佳长度,从而使残余干扰信号能量加上由 LMS 算法引入的超量均方误差为最小。本文对窄带和宽带两种干扰信号均将进行探讨。
The Wiener filter, which can be derived from known input statistics, is the best linear filter in terms of the best fit under mean square error. When these statistical properties are unknown, filtering problems can be excellently achieved using a least mean square error (LMS) adaptive filter. The gradient descent algorithm introduces an excess mean square error that is proportional to the length of the adaptive filter. In adaptive array processors, the LMS filter forms a noise canceller that partially removes the sidelobe interferers from a given receive beam. In this paper, we derive the optimal length of the adaptive transversal filter so that the residual interference signal energy plus the excess mean square error introduced by the LMS algorithm is minimized. In this paper, both narrowband and wideband interference signals will be explored.