>It sounds like you need a raw IQ sampled binary file at about 5 Msps.
>Is that correct?

Thanks a lot for your hints!

If my math (sampling theorem) is not wrong, because I need a bandwidth
of 5 MHz (assuming only downstream transmission, given I will not use
upstream), it would require 5 Msps in case each sample is a complex number
or 10 Msps in case the samples are real-valued.

>There are a variety of methods to achieve this depending on your
>requirements. You could generate the multiplexed signal live with up
>to 3 adjacent ARFCN channels (800 kHz separation) with a B200 and
>osmo-trx with multi-ARFCN enabled.

That is a good starting point. But I would need to increase the number
of 200 kHz to reach a total of 5 MHz.

>But, it sounds like you just need a test file, so you could
>post-process an existing single ARFCN signal file in Octave or Matlab
>to carry multiple versions of the same GSM signal. That may or may not
>be sufficient depending on your needs.

Yes, exactly: just a test file. I am still studying the GSM PHY and I
suspect if I simply position the ARFCNs in different frequencies I
will end up with an invalid GSM signal (that is, it will not be properly
decoded by the Keysight VSA equipment that I have to interface to).

Marcus Dias

UFPA - Lasse - LaPS

2017-05-09 15:41 GMT-03:00 Tom Tsou <tom@tsou.cc>:

Hi Marcus,

On Tue, May 9, 2017 at 9:53 AM, Marcus Dias <vmarcusv@gmail.com> wrote:
> Most BTS code I have, generate only one or two 200 kHz channels, but I need
> to test a receiver using several (multiplexed) 200 kHz channels, composing
> approximately 5 MHz of total bandwidth. This should be a "valid" multiplex
> signal (correct and "decodable" control channel). Can anyone pointout where
> should I start to put together the software to generate a binary file with
> such signal? I guess I can inspect the code of a BTS that supports two TRX
> (two 200 kHz channels) but I would like to double check whether there is a
> more direct approach.

It sounds like you need a raw IQ sampled binary file at about 5 Msps.
Is that correct?

There are a variety of methods to achieve this depending on your
requirements. You could generate the multiplexed signal live with up
to 3 adjacent ARFCN channels (800 kHz separation) with a B200 and
osmo-trx with multi-ARFCN enabled.

But, it sounds like you just need a test file, so you could
post-process an existing single ARFCN signal file in Octave or Matlab
to carry multiple versions of the same GSM signal. That may or may not
be sufficient depending on your needs.

-TT