d*******d
发帖数: 3382 | 1 I want to generate the time domain waveform samples
of square-root raised cosine pulse, which is used as the
transmitter pulse.
I find the function rcosflt() and use it as:
g=rcosflt([0;0;0;1;0;0;0],1,4,'fir/sqrt',r)
where r is the rolloff factor, and I set 4 samples per symbol.
change 4 to larger number and you will get more continuous
waveform.
When I set r=0.2, the waveform will goes to zero suddenly
after t>3T or 4T. I think sth is wrong, as the pulse is
freq domain limited, should be unlimi | l********h
发帖数: 16 | 2 The reason is that the rcosflt() function provided by MATLAB is
in TIME domain. The shape will be close the shape of a sinc() function.
If you test for rolloff factor alpha from 0 to 1, you can see the difference.
You have to do a FFT transformation in order to see frequency wave,
which was the one you desired.
Hope this helps.
【在 d*******d 的大作中提到】
: I want to generate the time domain waveform samples
: of square-root raised cosine pulse, which is used as the
: transmitter pulse.
: I find the function rcosflt() and use it as:
: g=rcosflt([0;0;0;1;0;0;0],1,4,'fir/sqrt',r)
: where r is the rolloff factor, and I set 4 samples per symbol.
: change 4 to larger number and you will get more continuous
: waveform.
: When I set r=0.2, the waveform will goes to zero suddenly
: after t>3T or 4T. I think sth is wrong, as the pulse is
| d*******d
发帖数: 3382 | 3
thats exactly the problem I am having.
This is not exact according to math, as it should be band limited.
Will real system like this? no real system will be infinite in time domain.
I am studying the effect of strong interference on signal,
the small diff will cause big diff in output SNR.
【在 l********h 的大作中提到】
: The reason is that the rcosflt() function provided by MATLAB is
: in TIME domain. The shape will be close the shape of a sinc() function.
: If you test for rolloff factor alpha from 0 to 1, you can see the difference.
: You have to do a FFT transformation in order to see frequency wave,
: which was the one you desired.
: Hope this helps.
| l********h
发帖数: 16 | 4 OK. some learning points, hope it helps.
"The impulse response decays much faster at the zero-crossings (
approximately as 1/t3 for t>>Ts) when compared to the "brick-wall" filter
(alpha=0). The rapid time rolloff allows it to be TRUNCATED in time with
little deviation in performance by theory" (credits to Pappaport)
"Increasing rolloff factor decreases the sensitivity to timing jitter, but
increases the occupied bandwidth."(credits to Pappaport)
"The time domain filter is NONCAUSAL. It must be
【在 d*******d 的大作中提到】
:
: thats exactly the problem I am having.
: This is not exact according to math, as it should be band limited.
: Will real system like this? no real system will be infinite in time domain.
: I am studying the effect of strong interference on signal,
: the small diff will cause big diff in output SNR.
| l********h
发帖数: 16 | 5
4Ts
The answer is YES. Forget about the function in the MATLAB and design
your own raised cosine rolloff filter. The output would definitely what
you wanted, very smooth, without "jump".
shape.
【在 d*******d 的大作中提到】
:
: thats exactly the problem I am having.
: This is not exact according to math, as it should be band limited.
: Will real system like this? no real system will be infinite in time domain.
: I am studying the effect of strong interference on signal,
: the small diff will cause big diff in output SNR.
|
|
