Clarification on power_dbm in Sionna RT and Its Effect on CIR / Spectral Efficiency

[hssam1002]

Hi all,

I'm working on designing an AI-based precoder using the Sionna RT framework.

My current plan is as follows:

• I generate arbitrary noise with power computed as
  $noise_{dbm} = -174 + 10 \cdot \log_10 \left( 30kHz \right) + 7$
  , where 30 kHz is the subcarrier spacing and 7 dB is the noise figure.

• Then I add this noise to the received signal.

• On the x-axis of my plot, I vary the transmit power using the power_dbm parameter of the transmitter. (Or scaling the CIR to target transmit power)

• The y-axis shows the spectral efficiency, which I compute using the difference between the transmit power (power_dbm) and the noise power (noise_dbm).

However, I noticed that changing the power_dbm value of the transmitter doesn’t affect the output at all.
So, I’m wondering:
Where and how is power_dbm applied in the Sionna RT pipeline?
Is the CIR generated under the assumption of a fixed transmit power level?
And if so, what is that assumed power?

Thanks in advance for your help!

[jhoydis]

Hi @hssam1002,

The CIR is used to described how the channel transforms a transmitted signal until it reaches a receiver. The noise power as well as the transmit power are not part of the CIR. The noise is added to the received signal at the receiver and the transmit power scales the transmit signal before it is fed into the channel. Have a look at the definition of CIR in the EM Primer to see that neither noise nor transmit power impact it in any way.

It seems that you try to make simulations in which you want to change the SNR. A typical of doing this is to only change the noise power while keeping the transmit power fixed (by default this is 1). Have a look at this tutorial for an example.

[hssam1002]

Hi @jhoydis,
Thank you for your reply.

What I’m trying to do is calculate the sum-rate of a MU-MIMO system in the RT model.
My goal is to plot the sum-rate versus the transmitter power — ideally using transmitter.power_dbm for the x-axis.

My initial plan was to follow the tutorial method for setting noise power:

Noise power per subcarrier

temperature = 294 # [K]
no = BOLTZMANN_CONSTANT * temperature * subcarrier_spacing

Then, I would compute the effective SINR and use it to calculate the sum-rate.

However, based on what I saw in the following tutorial:
https://nvlabs.github.io/sionna/sys/tutorials/LinkAdaptation.html

It seems like transmitter.power_dbm only affects the system when using the built-in precoding functions.

Am I understanding this correctly?
If not, could you clarify where transmitter.power_dbm actually has an effect?

[jhoydis]

The power that you configure in Sionna RT has no impact at all on Sionna PHY. It is only used for computing RSS and SINR radio maps.

[hssam1002]

Thank you, @jhoydis !
I now clearly understand what I need to do