Misadjustment Measurement with Normalized Weighted Noise Covariance based LMS Algorithm

Abstract

In this paper, a new refined technique called Normalized Weighted Noise Covariance based LMS (NWC-LMS) technique has been proposed to measure misadjustment of an adaptive filtering system. This technique is aimed to track the measured misadjustment properly. Several techniques had been developed during past years to analyze and calculate the measured misadjustment based on weight noise covariance matrix. However there are still significant error in misadjustment measurement to the calculated misadjustment. The performance of the proposed NWC-LMS technique is verified by computer simulations for an unknown system with additive white Gaussian noise. From the simulation results it is observed that even for larger value of step size, NWC-LMS technique can predict the misadjustment better than the conventional techniques. Around 37 % error improvement is achieved for the larger step size 0.04 with same input condition with our proposed NWC-LMS technique.