Hallo Alexander,
On Jan 4, 2010, at 12:31 PM, Alexander Chemeris wrote:
> Hi Karsten,
>
> As a part of our support to OpenBTS project we plan to create a cheap
> hardware for it, i.e. SDR module suitable for running BTS. We plan to
> overcome some of USRP1/2 limitations while specializing hardware for
> GSM and we'll be glad if our hardware will help this project too. But
> I seriously doubt that it will be possible to capture both up-link and
> down-link with one unit - usual applications do not require that wide
> bandwidth and creation of a specialized hardware is expensive. To get
> kinda cheap hardware for airprobe you need to produce at least 1000
> units, and I doubt there are that much interest here. With more
> universal
> and thus narrow-band units you will always have to build arrays of
> them.
> But I don't see this as a problem as long as you synchronize them.
> E.g. read GnuRadio MIMO docs to understand how to synchronize
> several USRP units.
Very good to hear that you are creating an open front-end specifically
designed for GSM work. There is definitely a market.
Requiring two devices for debugging up- and downlink at once is no
problem at all as the data can easily be correlated later.
> Also it seems to me that you can't decode full GSM-900 uplink (or
> downlink) on a usual CPU. As you mentioned, you can employ FPGA
> (likely FPGA array) OR you can try to do this with OpenCL on a capable
> 3D-accelerator - may be a good student project. Though I should admit
> I'm not a DSP or FPGA expert and may be wrong about OpenCL
> abilities and that many FPGAs will be needed instead of one.
I agree that a PC can easily decode several channels. The problem is
more getting the raw data to the PC though the USRP1's USB link. Some
people on the list are working on clocking a USRP through a phone so
that much less oversampling and much lower data rates become possible.
This option might be reserved for the USRP-2, though.
Thanks for sharing your ideas. Looking forward to your SDR module.
Cheers,
-Karsten
> On Sun, Jan 3, 2010 at 12:14, Karsten Nohl <[email protected]> wrote:
>> In appears that the USRP-1 is limited in two dimensions, one of which
>> would be required for a full sniffer:
>> First, the USB link does not support for a whole band to be
>> transfered
>> to the PC in raw form. Second, the FPGA seems too small to support
>> decoding of the channels before sending to the PC. I'd be happy to be
>> proven wrong on the latter one by some ingenious FPGA programmer.
>>
>> The current tool of choice, USRP-2, has a faster link (GbE) and a
>> larger FPGA. I second your call for cheaper hardware as two USRP-2s
>> are too expensive for most researchers. I assume the right order of
>> doing things is: 1. Implement a sniffer on the most available
>> hardware to understand its requirements; then 2. construct a fit-for-
>> purpose hardware with just enough resources. I'd be surprised if we
>> found a scaled-down radio peripheral that already matches our needs.
>> The SSRP for example seems to share the bottlenecks of the USRP-1.
>>
>> Cheers,
>>
>> -Karsten
>>
>> On Jan 3, 2010, at 11:25 AM, Clemens Gruber wrote:
>>
>>> Yes for either .., or.. but if we want to capture both up- and
>>> downlink
>>> at the same time, there has to be a setup of 2 USRP2s, am I wrong?
>>> With the USRP1 it should, due to the 2 RX slots, be possible to
>>> capture
>>> both directions..
>>> I would really appreciate a cheaper variant like the one called
>>> SSRP..
>>> students as I am, do not have much money.. (and there are many of us
>>> out
>>> there who would like to join the active development but cannot
>>> afford
>>> the hardware)
>>>
>>> On Sun, 2010-01-03 at 11:04 +0100, Karsten Nohl wrote:
>>>> P.S. As a technical note: I believe a USRP-2 can captured a whole
>>>> GSM
>>>> allocation in either the upink or downlink direction.
>>>>
>>>>
>>>> On Jan 3, 2010, at 10:35 AM, Gregory Maxwell wrote:
>>>>
>>>>> On Sun, Jan 3, 2010 at 4:23 AM, Fabio Pietrosanti (naif)
>>>>> <[email protected]> wrote:
>>>>>> This thread is making me understanding that there's not enough
>>>>>> information.
>>>>>> I read on airprobe website:
>>>>>> https://svn.berlin.ccc.de/projects/airprobe/wiki/hardware
>>>>>> But i read here that 1 USRP2 along with 2 daughterboards are
>>>>>> required.
>>>>>> Can we confirm that the equipment needed is 2 DBSRX
>>>>>> daughterboards
>>>>>> along
>>>>>> with antennas?
>>>>>
>>>>> You're spouting nonsense: USRP2 is a single daughterboard beast.
>>>>> The
>>>>> page you're linking to only mentions USRP2 in passing.
>>>>>
>>>>> If these details are causing you confusion for you then you need
>>>>> to do
>>>>> more background reading to understand the underlying technology.
>>>>> There are plenty of details out there, including information on
>>>>> the
>>>>> limitations of the USRP1 bandpass in dealing with externally
>>>>> controlled channel allocation.
>>>>>
>>>>> These tools are not (currently) suitable for casual use, you
>>>>> need to
>>>>> invest a considerable amount of effort into learning the
>>>>> technology.
>>>>> Because a point-and-click tool will never be possible (due to the
>>>>> required hardware) it isn't at all clear to me that it would be
>>>>> worthwhile for anyone to invest time into making it too much
>>>>> easier.
>>>>> _______________________________________________
>>>>> A51 mailing list
>>>>> [email protected]
>>>>> http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
>>>>
>>>> _______________________________________________
>>>> A51 mailing list
>>>> [email protected]
>>>> http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
>>>
>>> _______________________________________________
>>> A51 mailing list
>>> [email protected]
>>> http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
>>
>> _______________________________________________
>> A51 mailing list
>> [email protected]
>> http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
>>
>
>
>
> --
> Regards,
> Alexander Chemeris.
_______________________________________________
A51 mailing list
[email protected]
http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
