SDMA with Sionna

[BaptisteSambon]

Hi everyone,

I was wondering how it would be possible to do spatial multiplexing (SDMA) with Sionna to allow users to access the same time-frequency resources but still be able to separate them in the spatial domain.

I guess one would have to use a multi-user MIMO system and play with the StreamManagement class, but I am not familiar enough with Sionna to see how to actually implement this.

Many thanks in advance for your help ;-)

[faycalaa]

Hi @BaptisteSambon,

I recommend checking out the tutorials. In particular, there's a tutorial focused on setting up multi-user MIMO simulations.

[BaptisteSambon]

Hi @faycalaa,

Thank you very much for your response and your help!

I have looked at this tutorial but I have a few misunderstandings about this one and about understanding the results.

Indeed, I tried to simulate a communication between 1 user and 1 BS in uplink but with the number of antennas equal to 4 and 8 for the UT and the BS respectively. The following parameters were then modified compared to the initial tutorial in the Keras model (at the end of the tutorial):

    self._num_bs_ant = 8
    self._num_ut = 1
    self._num_ut_ant = 4

  self._ut_array = AntennaArray(
                                   num_rows=1,
                                   num_cols=int(self._num_ut_ant/2),
                                   polarization="single",
                                   polarization_type="V",
                                   antenna_pattern="omni",
                                   carrier_frequency=self._carrier_frequency)

I wanted to understand how the transmitter could perform spatial multiplexing to send multiple independent streams of information in parallel with a classic MIMO structure and how it could benefit from diversity to obtain a diversity gain. With the diversity-multiplexing trade-off, the diversity gain should decreases as the multiplexing gain increases. I tried to observe that in Sionna with the scenario described above by changing the "self._num_streams_per_tx" parameter. As I understand it, this parameter corresponds to the number of parallel streams that can be sent, and therefore to the multiplexing gain.

However, when I simulate 1 stream per tx or 2 by changing the “self._num_streams_per_tx” parameter to 1 and 2 respectively, I'm surprised to observe that the BER values obtained are lower when the number of streams equals 2 (see image below). This doesn't make sense with my reasoning, since with a single stream of information and multiple antennas at the receiving and transmitting end, the diversity gain should be greater and therefore the BER values obtained lower.

So I am wondering what does the “self._num_streams_per_tx” parameter correspond to ? Does it corresponds to the number of different versions of the same information received at the BS ? But in this case, how could multiplexing be performed ?

Many thanks in advance for your response and thanks again for your help.