deconvolution in time?
- From: Julian Stoev <stoev@xxxxxxxxxxxx>
- Date: Fri, 04 Nov 2005 15:17:02 +0900
Hello!
I have a output signal y(k) and a plant P(z). The signal y(k) contains some noise and I know the PSD of the noise. But lets assume that for now that the noise is qhite and the system looks like this:
Y(z)=U(z)P(z)+E(z)
u(k) y(k)
---------->[P(z)]-->[+]--->
|
e(k) |
-------|I want to restore the signal u(k) from y(k). I can do this off-line in freq domain, no problem, but I need this on-line. I think I need something related to Wiener filters.
The deconvolution P^(-1)(z) may be unstable, which is the biggest problem for me. I am not worried that it may be no casual.
I read that there are Kalman filter based approaches to this, but I can't find some practical explanation I can understand in reasonable amount of time.
Can somebody give me some reading directions - books, papers. Better papers, because I can download most of them in PDF.
Thanks a lot!
--JS .
- Follow-Ups:
- Re: deconvolution in time?
- From: abariska
- Re: deconvolution in time?
- From: Real_McCoy
- Re: deconvolution in time?
- From: Real_McCoy
- Re: deconvolution in time?
- From: Christian Langen
- Re: deconvolution in time?
- Prev by Date: Re: help me set timer1 and time3 of tms320f2812 at the same time
- Next by Date: Re: The Miracle of new life!
- Previous by thread: Finding error in analog qam system
- Next by thread: Re: deconvolution in time?
- Index(es):
Relevant Pages
|
