LMS ALGORITHM WITH REUSE OF COEFFICIENTS AND ROBUSTNESS AGAINST IMPULSIVE NOISE
Leonardo C. Resende1; Newton N. Siqueira2; Rodrigo M.S. Pimenta3; Diego B. Haddad4; Mariane R. Petraglia5
1 Instituto Federal do Rio de Janeiro - Paracambi; 2 CEFET-RJ - Nova Iguaçu; 3 CEFET-RJ - Maracanã; 4 CEFET-RJ - Petrópolis; 5 COPPE/UFRJ - PEE
Baixar PDF doi:10.20906/CPS/CBA2018-1010
Resumo
This article contributes with two major improvements related to the learning performance of the least mean squares (LMS) adaptive filtering algorithm under challenging scenarios, low signal-to-noise ratio and in the presence of impulsive noise. The first problem is solved through the insertion of the coefficient reuse technique in the update equation, while the second one uses the maximum correntropy criterion to make the algorithm less sensitive to impulsive noise, which is very often present in real applications.
Palavras-chave: Adaptive filtering; LMS; Reuse of Coefficients; Maximum Correntropy