Title :
Improving performances of complementary pair LMS algorithm
Author :
Bilcu, Radu Ciprian ; Kuosmanen, Pauli ; Rusu, Corneliu
Author_Institution :
Signal Process. Lab., Tampere Univ. of Technol., Tampere, Finland
Abstract :
In this paper we discuses how to improve the behavior of the CP-LMS algorithm that was proposed by Min-Soo Park and Woo-Jin Song [5]. The difference between the new algorithm and CP-LMS algorithm is that the new algorithm uses two LMS algorithms that operate in parallel, one with large and fixed step and another with variable step. The coefficients of the novel variable step algorithm are re-initialized by the coefficients of the algorithm with fixed step whenever possible. In the meantime the step-size of the proposed algorithm is increased or decreased if the algorithm is respectively far or near to the optimum.
Keywords :
adaptive filters; least mean squares methods; CP-LMS algorithm; adaptive filtering; algorithm coefficients; complementary pair LMS algorithm; large-fixed step algorithm; least mean square; performance improvement; step-size; variable step algorithm; Adaptive filters; Convergence; Filtering algorithms; Least squares approximations; Mathematical model; Signal processing algorithms; Steady-state;
Conference_Titel :
Signal Processing Conference, 2000 10th European
Conference_Location :
Tampere
Print_ISBN :
978-952-1504-43-3