Abstract
A new normalized subband adaptive filter based on the minimum error entropy criterion (MEE-NSAF) is proposed for identifying a highly noisy system. The MEE-NSAF utilizes a kernel function and a number of past errors in adaptation, whereas the classical NSAF relies only on the current error signal. Moreover, the stability of the MEE-NSAF is analyzed. To further improve the performance of the MEE-NSAF under the sparse impulse responses, an improved proportionate MEE-NSAF (MEE-IPNSAF) algorithm is proposed. Simulation results show that the proposed algorithms can achieve improved performance as compared with the conventional NSAF when noise gets severe.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have