Novel adaptive VSS-NLMS algorithm for system identification

In this paper, to mitigate the tradeoff between fast convergence rate, low steady-state misadjustment and good tracking ability, a novel, easy to implement, time-varying step-size normalized least mean square (NLMS) algorithm-based transversal filters is presented in system identification applications. By utilizing the system input power and cross-correlation between the input signal and estimated error, the new variable step-size scheme can reduce the effect of the system noise on the performance without the priori knowledge of system noise power, especially for variable system noise. Experimental results in context of system identification illustrate that the propose algorithm is superior to other existing algorithms in terms of convergence speed, misadjustment and tracking ability.