DSPRelated.com
Forums

short term estimate

Started by d.pal March 19, 2007
I could not understand how a low pass IIR filter like below will represent
a short term power/energy estimate..

y(i+1)=(1-a)y(i)+ a|y(i)|

can anybody tell me anything about this?

Hi,


This is a non-linear system, are you sure that equation is right
(where did you get it from)?

I am not sure for non-linear system (maybe someone else here can
help). But for linear systems you can obtain a short trem
power/energy estimate with a low pass filter by mulitplying your
signal by a complex exponential which can shift any part of the
spectrum to the low pass region of the filter. You spectral estimate
for that band is then the power output of the filter (normalized by
the power of the impulse response of the filter).

One of the problems with non-linear systems in general is that they
can generate harmonic distortions which can bias the power estimate...



On Mar 19, 8:45 am, "d.pal" <pal_debabrata...@rediffmail.com> wrote:
> I could not understand how a low pass IIR filter like below will represent > a short term power/energy estimate.. > > y(i+1)=(1-a)y(i)+ a|y(i)| > > can anybody tell me anything about this?
Hi,

This is a non-linear system, are you sure that equation is right
(where did you get it from)?


Hi,

This is a non-linear system, are you sure that equation is right
(where did you get it from)?


I am not sure for non-linear system (maybe someone else here can
help). But for linear systems you can obtain a short trem
power/energy estimate with a low pass filter by mulitplying your
signal (y(i)) by a complex exponential which can shift any part of
the
spectrum to the low pass region of the filter. You spectral estimate
for that band is then the power output of the filter (normalized by
the power of the impulse response of the filter).


One of the problems with non-linear systems in general is that they
can generate harmonic distortions which can bias the power
estimate...

Hi,

This is a non-linear system, are you sure that equation is right
(where did you get it from)?


I am not sure for non-linear system (maybe someone else here can
help). But for linear systems you can obtain a short trem
power/energy estimate with a low pass filter by mulitplying your
signal (y)  by a complex exponential which can shift any part of the
spectrum to the low pass region of the filter. You spectral estimate
for that band is then the power output of the filter (normalized by
the power of the impulse response of the filter).


One of the problems with non-linear systems in general is that they
can generate harmonic distortions which can bias the power
estimate...

On Mar 20, 1:45 am, "d.pal" <pal_debabrata...@rediffmail.com> wrote:
> I could not understand how a low pass IIR filter like below will represent > a short term power/energy estimate.. > > y(i+1)=(1-a)y(i)+ a|y(i)| > > can anybody tell me anything about this?
It should be as above except the modulus is replaced by a square ie ay(i)^2. This is a form of variance estimate which is average power if the dc level is zero. N.