Reply by Vladimir Vassilevsky November 25, 20062006-11-25

Richard_K wrote:
> It is given by "Adaptive Filter Theory" that the average time constant for > LMS algorithm is 1/(2*Step Size*average of the eigenvalue).
This is an oversimplification. It depends.
> However, the > time constant for NLMS algorithm is not given.
But of course. The question does not make sense.
>Can anyone please tell me > the average time constant expression for NLMS algorithm? Many thanks.
The NLMS is LMS which was modified to make the convergence speed independent from the absolute energy of the reference signal. For the apple to apple comparison, the properties of NLMS and LMS are identical. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by Richard_K November 25, 20062006-11-25
It is given by "Adaptive Filter Theory" that the average time constant for
LMS algorithm is 1/(2*Step Size*average of the eigenvalue). However, the
time constant for NLMS algorithm is not given. Can anyone please tell me
the average time constant expression for NLMS algorithm? Many thanks.