DSPRelated.com
Forums

matched filter before interpolator

Started by Pave...@gmail.com February 7, 2008
Hello.
Who knows how matched filter can be realized if sample rate is not
integer multiple of symbol rate?
In book of Meyr "Digital Communications Recievers" matched filter is
placed before controlled interpolation. What is the impulse response
of fiter matched for example with rectangular transmit filter  or
raised cosine filter in that case? In Matlab I can calculate filter
impulse response only if sample rate is integer multiple of symbol
rate.
Thanks.
Pavel Schukin.

Pavel.Schukin@gmail.com wrote:

> Hello. > Who knows how matched filter can be realized if sample rate is not > integer multiple of symbol rate?
I know. But I won't tell you.
> In book of Meyr "Digital Communications Recievers" matched filter is > placed before controlled interpolation. What is the impulse response > of fiter matched for example with rectangular transmit filter or > raised cosine filter in that case? In Matlab I can calculate filter > impulse response only if sample rate is integer multiple of symbol > rate.
Just another illustration of the mental harm done by MatLab. VLV
">> Hello.
>> Who knows how matched filter can be realized if sample rate is not >> integer multiple of symbol rate? > > I know. But I won't tell you. >
VLV, you can at least point him in the right direction and I'm curious too:)
On 7 Feb, 17:43, "Pavel.Schu...@gmail.com" <Pavel.Schu...@gmail.com>
wrote:
> What is the impulse response > of fiter matched for example with rectangular transmit filter &#4294967295;or > raised cosine filter in that case?
Window functions in a matched filter? That's a new one. How does that work? Rune
On 7 Feb, 17:43, "Pavel.Schu...@gmail.com" <Pavel.Schu...@gmail.com>
wrote:
> What is the impulse response > of fiter matched for example with rectangular transmit filter =A0or > raised cosine filter in that case?
Window functions in a matched filter? That's a new one. How does that work? Rune
On Feb 19, 4:15 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 7 Feb, 17:43, "Pavel.Schu...@gmail.com" <Pavel.Schu...@gmail.com> > wrote: > > > What is the impulse response > > of fiter matched for example with rectangular transmit filter or > > raised cosine filter in that case? > > Window functions in a matched filter? That's a new one. > How does that work? > > Rune
If the signal was generated by windowing, as all finite signals are, don't you try to match the window? Dale B. Dalrymple http://dbdimages.com
On 19 Feb, 18:10, dbd <d...@ieee.org> wrote:
> On Feb 19, 4:15 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > On 7 Feb, 17:43, "Pavel.Schu...@gmail.com" <Pavel.Schu...@gmail.com> > > wrote: > > > > &#4294967295;What is the impulse response > > > of fiter matched for example with rectangular transmit filter &#4294967295;or > > > raised cosine filter in that case? > > > Window functions in a matched filter? That's a new one. > > How does that work? > > > Rune > > If the signal was generated by windowing, as all finite signals are, > don't you try to match the window?
Not explicitly. The way I know the matched filter, you try and match the impulse response. How the impulse is found is a separate issue. If a non-rectangular window function was used somewhere it would be part of the impulse response already and not need explicit handling. Rune
On Thu, 7 Feb 2008 08:43:27 -0800 (PST), "Pavel.Schukin@gmail.com"
<Pavel.Schukin@gmail.com> wrote:

>Hello. >Who knows how matched filter can be realized if sample rate is not >integer multiple of symbol rate?
Resampling filters, like Farrow filters, are often used.
>In book of Meyr "Digital Communications Recievers" matched filter is >placed before controlled interpolation. What is the impulse response >of fiter matched for example with rectangular transmit filter or >raised cosine filter in that case?
An integrate-and-dump filter in the receiver matches rectangular pulses from the transmitter.
> In Matlab I can calculate filter >impulse response only if sample rate is integer multiple of symbol >rate. >Thanks. >Pavel Schukin.
Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org
On Feb 19, 9:29 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 19 Feb, 18:10, dbd <d...@ieee.org> wrote: > >
> > On Feb 19, 4:15 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > On 7 Feb, 17:43, "Pavel.Schu...@gmail.com" <Pavel.Schu...@gmail.com> > > > wrote: > > > > > What is the impulse response > > > > of fiter matched for example with rectangular transmit filter or > > > > raised cosine filter in that case?
> > > Window functions in a matched filter? That's a new one. > > > How does that work?
> > > Rune > > > If the signal was generated by windowing, as all finite signals are, > > don't you try to match the window? > > Not explicitly. The way I know the matched filter, you > try and match the impulse response. How the impulse is > found is a separate issue. If a non-rectangular window > function was used somewhere it would be part of the > impulse response already and not need explicit handling. > > Rune
Rune You are the only one to suggest any explicit handling of windowing. The OP mentioned it as part of the transmit function which produces the impulse response. It was my point that windows are implicit in all finite responses, a fact that is better dealt with by understanding the consequences than by a knee-jerk reaction on seeing the word: windowing. The confusion about windowing comes from thinking that it is something we can do, or not, explicitly after the impulse response is determined. Dale B. Dalrymple
On 19 Feb, 21:52, dbd <d...@ieee.org> wrote:
> On Feb 19, 9:29 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > > > > On 19 Feb, 18:10, dbd <d...@ieee.org> wrote: > > > > On Feb 19, 4:15 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > > On 7 Feb, 17:43, "Pavel.Schu...@gmail.com" <Pavel.Schu...@gmail.com> > > > > wrote: > > > > > > &#4294967295;What is the impulse response > > > > > of fiter matched for example with rectangular transmit filter &#4294967295;or > > > > > raised cosine filter in that case? > > > > Window functions in a matched filter? That's a new one. > > > > How does that work? > > > > Rune > > > > If the signal was generated by windowing, as all finite signals are, > > > don't you try to match the window? > > > Not explicitly. The way I know the matched filter, you > > try and match the impulse response. How the impulse is > > found is a separate issue. If a non-rectangular window > > function was used somewhere it would be part of the > > impulse response already and not need explicit handling. > > > Rune > > Rune > > &#4294967295;You are the only one to suggest any explicit handling of windowing.
You are right. I didn't understand (and still don't) why and how the terms 'rectangular' or 'raised cosine' appear in a discussion about matched filters. Are these impulse repsonses or transfer functions? Rune