A Time-Varying Convergence Parameter for the LMS Algorithm in the Presence of White Gaussian Noise
ABSTRACT:
A novel approach for the least-mean-square (LMS) estimation algorithm is proposed. Rather than using a fixed convergence parameter μ, this approach utilizes a time-varying LMS parameter 
. This technique leads to faster convergence and provides reduced mean-squared error compared to the conventional fixed parameter LMS algorithm. The algorithm has been tested for noise reduction and estimation in narrow-band FM signals corrupted by additive white Gaussian noise.
. This technique leads to faster convergence and provides reduced mean-squared error compared to the conventional fixed parameter LMS algorithm. The algorithm has been tested for noise reduction and estimation in narrow-band FM signals corrupted by additive white Gaussian noise. For the LMS algorithm in a white Gaussian noise environment. A general power decaying law has been studied, however, other time-varying laws could also be applicable. The main idea is to set the convergence parameter to a large value in the initial state in order to speed up the algorithm convergence. The modified algorithm has been tested for noise reduction and estimation in linear frequency-modulated (LFM) narrowband signals corrupted by additive white Gaussian noise.
click here to download full project
