On the Properties of the System Mismatch Covariance Matrix in the LMS Adaptive Algorithm

Ahstract-The system mismatch covariance matrix is of major interest in some of the optimization approaches of the least-mean-square type adaptive algorithms. Some examples are the ones based on the Kalman filtering theory, or the variable step-size algorithms based on the minimization of the mean square system mismatch. All of them require information about this specific matrix. The usual assumption is to approximate this matrix with a scaled unity one. This paper analyzes the validity conditions of this assumption. Simulation results support the theoretical findings.

[1]  Chang Hee Lee,et al.  Mean-Square Deviation Analysis of Affine Projection Algorithm , 2011, IEEE Transactions on Signal Processing.

[2]  Jae-Woo Lee,et al.  A Theory on the Convergence Behavior of the Affine Projection Algorithm , 2011, IEEE Transactions on Signal Processing.

[3]  Gene H. Golub,et al.  Matrix computations , 1983 .

[4]  Jacob Benesty,et al.  An optimized NLMS algorithm for system identification , 2016, Signal Process..

[5]  Jacob Benesty,et al.  Study of the General Kalman Filter for Echo Cancellation , 2013, IEEE Transactions on Audio, Speech, and Language Processing.

[6]  Eweda Eweda Convergence analysis of adaptive filtering algorithms with singular data covariance matrix , 2001, IEEE Trans. Signal Process..

[7]  Azzedine Zerguine,et al.  Convergence and steady-state analysis of a variable step-size NLMS algorithm , 2003, Signal Process..

[8]  Ali H. Sayed,et al.  Adaptive Filters , 2008 .