GSM channel filter

Hi, 

Suppose I have to extract a single 200kHz channel from the GSM band and
I'd do it using a non-linear phase response filter. Could someone give me
an idea or a reference on how to evaluate the impact of the phase
distortion? I mean how the GMSK demodulation process will be affected. Any
advice or reference regarding this would be appreciated.

thank you, 

friedman

friedman wrote:
> Hi, 
> 
> Suppose I have to extract a single 200kHz channel from the GSM band and
> I'd do it using a non-linear phase response filter. Could someone give me
> an idea or a reference on how to evaluate the impact of the phase
> distortion?

Take the difference between the original and the altered spectrums. 
Think of this difference as if it is a noise which is added to the 
original signal. Compute the equivalent SNR impairment due to that noise.

> I mean how the GMSK demodulation process will be affected. Any
> advice or reference regarding this would be appreciated.

If you have the amplitude limiter after the filter (which is the usual 
case), this becomes the tough nonlinear problem. If you don't, then the 
filter can be accounted for.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

Thank you for the answer Vladimir,

>
>friedman wrote:
>> Hi, 
>> 
>> Suppose I have to extract a single 200kHz channel from the GSM band
and
>> I'd do it using a non-linear phase response filter. Could someone give
me
>> an idea or a reference on how to evaluate the impact of the phase
>> distortion?
>
>Take the difference between the original and the altered spectrums. 
>Think of this difference as if it is a noise which is added to the 
>original signal. Compute the equivalent SNR impairment due to that
noise.
>


This would mean to be able to implement a sufficient long FIR filter for
generating my reference, which could be a problem. I was wondering if
anyone was aware of some analytical results or, better,  methods to
investigate this.


>> I mean how the GMSK demodulation process will be affected. Any
>> advice or reference regarding this would be appreciated.
>
>If you have the amplitude limiter after the filter (which is the usual 
>case), this becomes the tough nonlinear problem. If you don't, then the 
>filter can be accounted for.
>

Are you aware of any other "UNusual" case regarding the receiver structure
in a typical SDR platform? Do you know any "didactic" (maybe even open?)
SDR implementation (with software channel separation) suited for analyzing
GSM signal?


Thanks,
    Friedman

friedman wrote:

> Thank you for the answer Vladimir,
> 
> 
>>friedman wrote:
>>
>>>Hi, 
>>>
>>>Suppose I have to extract a single 200kHz channel from the GSM band
> 
> and
> 
>>>I'd do it using a non-linear phase response filter. Could someone give
> 
> me
> 
>>>an idea or a reference on how to evaluate the impact of the phase
>>>distortion?
>>
>>Take the difference between the original and the altered spectrums. 
>>Think of this difference as if it is a noise which is added to the 
>>original signal. Compute the equivalent SNR impairment due to that
> 
> noise.
> 
> 
> 
> This would mean to be able to implement a sufficient long FIR filter for
> generating my reference, which could be a problem.

If this is a problem for you, stop here.

> I was wondering if
> anyone was aware of some analytical results or, better,  methods to
> investigate this.

Investigate exactly what? The dependence of what from which in the case 
of what? There were tons of books and publications about GSM in 80x.

>>>I mean how the GMSK demodulation process will be affected. Any
>>>advice or reference regarding this would be appreciated.
>>
>>If you have the amplitude limiter after the filter (which is the usual 
>>case), this becomes the tough nonlinear problem. If you don't, then the 
>>filter can be accounted for.
>> 
> Are you aware of any other "UNusual" case regarding the receiver structure
> in a typical SDR platform? 

A trivial discriminator followed by DFE is usual. Of course, there could 
be better approaches depending on what is available, however the first

> Do you know any "didactic" (maybe even open?)
> SDR implementation (with software channel separation)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Bad idea. Drowning in the intermodulation products.

> suited for analyzing
> GSM signal?

DYI. It is quite simple.



Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com