Prev: Window and/or Filter - possibly naive question
Next: imresize in 3D
From: Nitram on 23 Nov 2009 11:06
>
>If your goal is frequency search algorithm, why bother with low level
>details in time domain?

I guess I should explain myself in more details:

Actually I want to extract synchronization information of an OFDM signal.
I am planning to use cyclic prefix correlation to get the timing
information and then wideband filtering plus a correlation of a known
sequence in frequency domain to get the frequency correlation. Therefore, I
have to introduce a timing and frequency offset (which I was hoping could
be anywhere from 0 to 50MHz). I was planning to introduce the frequency
offset by upconverting the signal to RF and downconverting it with a
slighthly different LO, creating a frequency offset.


>> Therefore, I need
>> to go to RF in order to verify my search algorithm (unless there is a
>> better way).
>
>Split your problem into a set of different problems at different scales
>and do everything in the baseband.

I am not sure if I understand correctly what are the different scales you
are thinking of. Would it be possible to elaborate?

I really appreciate your help.
From: Nitram on 23 Nov 2009 11:25

>Is the algorithm going to be applied at baseband or at RF? If it will
>be applied to the baseband signal, why not test it there? Frequency
>offset at RF = frequency offset at baseband.

True, but I would need to simulate pretty large frequency offset e.g.
cases where half the bandwidth of the baseband signal is filtered out by
the RF front-end filter. What would be a simple way to achive this in
baseband?

Thanks
From: Eric Jacobsen on 23 Nov 2009 13:17
On 11/23/2009 9:25 AM, Nitram wrote:
>> Is the algorithm going to be applied at baseband or at RF? If it will
>> be applied to the baseband signal, why not test it there? Frequency
>> offset at RF = frequency offset at baseband.
>
> True, but I would need to simulate pretty large frequency offset e.g.
> cases where half the bandwidth of the baseband signal is filtered out by
> the RF front-end filter. What would be a simple way to achive this in
> baseband?
>
> Thanks

Why not model the RF filter effect at baseband? Is there an asymmetric
distortion that you need to model?

--
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com
From: Nitram on 23 Nov 2009 15:28

>Why not model the RF filter effect at baseband? Is there an asymmetric
>distortion that you need to model?

What I would need to do is shift my baseband signal by a certain frequency
offset, pass it through a lowpass filter to emulate a receiver RF filter(in
the graph the lowpass filter is represented by a brickwall filter) and then
attempt to find the frequency of the random shift on the receiver. ^
^
| |_________
|______ | _____|<--low pass filter
| \ | | |\
| \ --> | | | \<--shifted signal
|________\_________>f |___|_____|__\______>f
<--------> <-------->
BW
The shift would be a simple complex multiplication and I guess the only
thing I need to be careful about is not to shift my signal too far so that
I get aliasing before my lowpass filter (either this or do successive
shift/lowpass procedures).

What do you think?

Thanks.

From: Eric Jacobsen on 23 Nov 2009 15:51
On 11/23/2009 1:28 PM, Nitram wrote:
>> Why not model the RF filter effect at baseband? Is there an asymmetric
>> distortion that you need to model?
>
> What I would need to do is shift my baseband signal by a certain frequency
> offset, pass it through a lowpass filter to emulate a receiver RF filter(in
> the graph the lowpass filter is represented by a brickwall filter) and then
> attempt to find the frequency of the random shift on the receiver. ^
> ^
> | |_________
> |______ | _____|<--low pass filter
> | \ | | |\
> | \ --> | | | \<--shifted signal
> |________\_________>f |___|_____|__\______>f
> <--------> <-------->
> BW
> The shift would be a simple complex multiplication and I guess the only
> thing I need to be careful about is not to shift my signal too far so that
> I get aliasing before my lowpass filter (either this or do successive
> shift/lowpass procedures).
>
> What do you think?
>
> Thanks.
>

If you're trying to simulate the effects of a frequency offset and how
it would be affected by the frequency selectivity of the RF or IF
filter, that looks appropriate to me.

--
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com
First  |  Prev  |  Next  |  Last
Pages: 1 2 3
Prev: Window and/or Filter - possibly naive question
Next: imresize in 3D