Title :
An inequality by which to adjust the LMS algorithm step-size
Author :
Jiangnan Chen ; Priemer, R.
Author_Institution :
Stoelting Co., Wood Dale, IL, USA
Abstract :
This paper gives a method for differentiating between the transient and steady states in the LMS algorithm using inequalities among a set of output errors. Based on these inequalities, it also gives an algorithm for automatic step size adjustment to both improve the convergence rate and reduce steady-state parameter estimation errors.<>
Keywords :
convergence of numerical methods; error analysis; least mean squares methods; parameter estimation; signal processing; transients; LMS algorithm step-size; automatic step size adjustment; convergence rate; inequalities; output errors; parameter estimation errors; signal processing; steady state; transient state; Convergence; Least squares approximation; Parameter estimation; Steady-state;
Journal_Title :
Communications, IEEE Transactions on
