# Design filter by least squares

1. ## Design filter by least squares

The LS error methods use an error criterion that is related to the energy of
the signal or noise.
The design problem is posed by defining an error measure E as a sum (or integral) of the
squared differences between the actual and the desired frequency response over a
set of L frequency samples. This error function is defined as

Code:
E=\sum_{n = 0}^{L-1}|(H(\omega)-H(\omega)_d|^2

One of the most effective modifications of the direct LS error design methods is
to change the bands of frequencies over which the minimization is carried out.
(from the book : Digital Filter Design - Parks and Burrus).

The question : how the LS error design methods is to change the bands of frequencies ?

•

2. ## Re: Design filter by least squares

Originally Posted by Sandi2000
The question : how the LS error design methods is to change the bands of frequencies ?
What do you want to mean ?

1 members found this post helpful.

•

3. ## Re: Design filter by least squares

Originally Posted by pancho_hideboo
What do you want to mean ?
I want to understand ,''One of the most effective modifications of the direct LS error design methods is
to change the bands of frequencies over which the minimization is carried out.''

what i understood the change bands of frequencies is done by using the iteration method, am I right?

•

4. ## Re: Design filter by least squares

What dou you mean by change band ?
Do you mean weighting for band ?

1 members found this post helpful.

5. ## Re: Design filter by least squares

Originally Posted by pancho_hideboo
What dou you mean by change band ?
Do you mean weighting for band ?
I don't know .
Digital Filter Design,Parks and Burrus ,page 70,this is a reference to the information.

•

6. ## Re: Design filter by least squares

I didn't check the phrase in the book. But as far as I understand, it refers to the bandwidth-changing property of a regular digital FIR filter.
Remembering that the passband freq of FIR is related to the sampling frequency with a fixed percentage, changing the sampling frequency changes the band of the filter.

1 members found this post helpful.

--[[ ]]--