The research of λVS_BLMS algorithm in digital hearing aid

The Least Mean Square (LMS)algorithm is the most basic adaptive echo cancellation algorithm. However, the fixed step size of LMS does not solve the contradiction between the convergence rate and the steady state error which are all used for judging the performance of algorithm. This paper references the Sigmoid function and creates the nonlinear relation between estimated error e(n) and variable step size step(n) . Furthermore, the paper also introduces the error coefficient λ_L and deviation degree |λ_L| to modify the step size step(n), which references the “forgetting factor” in the RLS algorithm. In order to satisfy the requirements about the real-time and the low power consumption in digital hearing aid, the paper also integrates into the Blocked LMS (BLMS) algorithm and proposes the new algorithm λVS_BLMS. The simulation verifies that the λVS_BLMS posses the better performances and the smaller computation.