A Modified LMS Algorithm for Variable Step Size Adaptive Filters
XU Kai, JI Hong, YUE Guang-xin ( Telecommunication Engineering School, Beijing University of Posts and Telecommunication, Beijing 100876, China )
A modified LMS algorithm, which establishes another new non-linear functional relationship between m and e(n), is proposed. This functional relationship not only has the virtue of e(n) slightly changing close to zero, but also has better performance than that of the former algorithms in the low SNR circumstance. Theoretical analysis and computer simulations show that with the proposed algorithm, convergence rate and misadjustment can be improved than the former in low SNR environments. Simulations also show that the algorithm is robust against to the abrupt change of the system.