Wideband signal estimation using Karhunen-Loeve filters

Abstract

Karhunen-Loeve (K-L) transform is known as an optimum transform in the mean square sense and it has been used in many areas such as in signal and image processing applications. In this paper, the K-L transform is used in the design of Wiener filters for wideband signal estimation. These filters are formulated based on nonlinear smoothing algorithms which produce circulant, Toeplitz and diagonal impulse response matrices. The performance of these filters is investigated based on the mean-squared-error (MSE) and percent distortion. Computer simulations show that the filters perform well in estimating the signal embedded in white noise.<>