deconvolution using known set of filters

i have lot of filters impulse responses and signal 'x' that i got by
convolving original signal 's' with one of the filter impulse
response.Also,the signal 'x' contains noise. Now i want to recover the
original signal 's'.
1)Which techniques are most feasible and effective to solve this problem
2)which technique is more real time

On Jun 27, 6:47 am, "riz" <rizwan....@gmail.com> wrote:
> i have lot of filters impulse responses and signal 'x' that i got by
> convolving original signal 's' with one of the filter impulse
> response.Also,the signal 'x' contains noise. Now i want to recover the
> original signal 's'.
> 1)Which techniques are most feasible and effective to solve this problem
> 2)which technique is more real time

It's hard to answer these kinds of questions in much generality,
and you don't describe your problem in any particularity.

It's usually possible to do deconvolution in the frequency domain,
by setting an appropriate cutoff of which frequency points are
unstable because either the filter is near zero or the input is near
zero.

As for real-time, some people have successfully used adaptive filter
type of settings.

Julius

>On Jun 27, 6:47 am, "riz" <rizwan....@gmail.com> wrote:
>> i have lot of filters impulse responses and signal 'x' that i got by
>> convolving original signal 's' with one of the filter impulse
>> response.Also,the signal 'x' contains noise. Now i want to recover the
>> original signal 's'.
>> 1)Which techniques are most feasible and effective to solve this
problem
>> 2)which technique is more real time
>
>It's hard to answer these kinds of questions in much generality,
>and you don't describe your problem in any particularity.
>
>It's usually possible to do deconvolution in the frequency domain,
>by setting an appropriate cutoff of which frequency points are
>unstable because either the filter is near zero or the input is near
>zero.
>
>As for real-time, some people have successfully used adaptive filter
>type of settings.
>
>Julius
>
>
Can we apply Wiener filter to solve this kind of problem?

On Jun 27, 9:21 am, "riz" <rizwan....@gmail.com> wrote:

> Can we apply Wiener filter to solve this kind of problem?

Signals and systems are interchangeable.  Suppose that
you have an input x(t), LTI system h(t), and the output y(t).
Then, for the appropriate cross-correlation and autocorrelations,
  rxy(t) = rxx(t) * h(t).
You have to satisfy a few conditions to be able to apply
Wiener filter-type solutions, but it's possible.

Of course, I don't guarantee that it will work ;-).

Julius