Announcing Sionna v0.17

[jhoydis]

Ray tracing

• Every scene object has a now a velocity vector which is used to compute per-path Doppler shifts. These are used by Paths.apply_doppler()to compute time evolution of channel impulse responses.
• Every scene object has now a position and orientation property that can be modified after a scene is loaded.
• A new tutorial notebook “Mobility in Sionna RT” demonstrates various ways how the new features can be used to simulate mobility.

Misc

• Fixes typos in the 5G NR PUSCH Tutorial (Hello World Example in 5G NR PUSCH Tutorial does not take channel into account. #384)
• Fixes typos in the Neural Receiver Tutorial (Fixed 2 typos on the Neural_Receiver.ipynb notebook #406, contribution by @pablosreyero)
• Fixes typos in Tutorial Notebooks 1 and 2 (Fix typos in tutorial notebooks 1 and 2 #355, contribution by @LateNightIceCream)
• Fixes a bug in the Demapper for very high SNR (mapping: allow demapping infinite/high SNR symbols #327, contribution by @japm48)
• Fixes an issue in sample_positions() for coverage maps (Compatibility between Tensorflow 2.10-2.12 and Sionna 0.16.2 #376)
• Removes warnings related to casting from tf.float to tf.complex (Tensorflow warning #348)
• Fixes a bug in theta_phi_from_unit_vec() so that gradients are always well defined

[fanyimeng0]

Hello, I am trying to run the example in Mobility in Sionna RT, but I find that I always fail to run it when loading objects in the scene. I have encountered the same problem on both Windows using pip installation and Debian using docker installation. I make sure that my Sionna has been updated to version 0.17.0. How can I solve this problem?

[jhoydis]

I have found the bug. It will be corrected in the next release.
In the meantime, you can make the following change in these lines:

bbox_min = mi_to_tf_tensor(mi.Point3f(self._mi_shape.bbox().min), rdtype)
bbox_max = mi_to_tf_tensor(mi.Point3f(self._mi_shape.bbox().max), rdtype)

[robpegurri]

Thanks Jakob! I did update the lines as you suggested.

Again referring to the example, I do confirm the lines are finally working:

car_2 = scene.get("car_2")
print("Position: ", car_2.position.numpy())
print("Orientation: ", car_2.orientation.numpy())

By the way, the following code block does not:

# Move the car 10m along the y-axis
car_2.position += [0, 10, 0]

# And rotate it by 90 degree around the z-axis
car_2.orientation = [np.pi/2, 0, 0]

if colab_compat:
    scene.render(camera="cam", num_samples=512)
    raise ExitCell
scene.preview(); # Do not open a new preview, the preview above will update

The output is the following, hope it helps!

Traceback (most recent call last):
  File "sionna_mobility_example.py", line 69, in <module>
    car_2.position += [0, 10, 0]
  File "/home/roberto/.local/lib/python3.8/site-packages/sionna/rt/scene_object.py", line 235, in position
    solver_paths.wedges_origin.scatter_nd_update(wedges_ind, wedges_origin)
  File "/home/roberto/.local/lib/python3.8/site-packages/tensorflow/python/ops/resource_variable_ops.py", line 1424, in scatter_nd_update
    gen_state_ops.resource_scatter_nd_update(
  File "/home/roberto/.local/lib/python3.8/site-packages/tensorflow/python/ops/gen_state_ops.py", line 740, in resource_scatter_nd_update
    _ops.raise_from_not_ok_status(e, name)
  File "/home/roberto/.local/lib/python3.8/site-packages/tensorflow/python/framework/ops.py", line 6656, in raise_from_not_ok_status
    raise core._status_to_exception(e) from None  # pylint: disable=protected-access
tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __wrapped__ResourceScatterNdUpdate_device_/job:localhost/replica:0/task:0/device:CPU:0}} Dimensions [0,1) of indices[shape=[18,1]] = 18 must match dimensions [0,1) of updates[shape=[1,18,3]] = 1 [Op:ResourceScatterNdUpdate] name: 

[jhoydis]

Indeed, it seems that the fix causes another issue. We'll look into it.

[jhoydis]

You can try:

bbox_min = tf.cast(self._mi_shape.bbox().min, rdtype)
bbox_max = tf.cast(self._mi_shape.bbox().max, rdtype)

My previous code created a dummy batch dimension.

[robpegurri]

Everything works perfectly now, thanks!

[abrxao]

Hello, i'm very exicited with this new feature and want to integrate Sumo simulation traffics to Sionna RT. I searched some content to do this, but not succesfully. Can i do this integration? If yes, have some documentation to do this?

[sabusari]

Hi,
Regarding v0.18, further to the sample parameters in the RIS tutorial, is the implementation frequency-agnostic? Does it scale to mmWave and THz bands with appropriate configuration? Is the RIS also compatible with multiple antennas at the TX and RX?

PS: Asked here because it appears the v0.18 discussion forum is locked.

[jhoydis]

Hi @sabusari,

Yes, you can scale it to any frequency. The unit cell spacing is always $\lambda/2$. And, yes, you can have multiple antennas at the TX and RX.

[sabusari]

Great, thanks!