Channel Impulse Response unexpected values about sionna

[jhoydis]

Hi @DerBambus,

Could you provide a minimal code snippet that reproduces the issue you are observing? Screenshots of your code are not really helpful.

The order of the CIRs is given by their shape [...,num_rx, num_rx_ant, num_tx, num_tx_ant....].
The receivers and transmitters are numbered in the order by which they are added to the scene.
For example, the channel from alicetx to bobrx would have the index [...,1,0,0,0,....].

[DerBambus]

The following would be the minimal code to reproduce said issue:

    simulation_scene = load_scene(<scene_file_name>)
    simulation_scene.tx_array = PlanarArray(num_rows=1,
                                        num_cols=1,
                                        vertical_spacing=0.5,
                                        horizontal_spacing=0.5,
                                        pattern="tr38901",
                                        polarization="V")

    # Configure antenna array for all receivers
    simulation_scene.rx_array = PlanarArray(num_rows=1,
                                        num_cols=1,
                                        vertical_spacing=0.5,
                                        horizontal_spacing=0.5,
                                        pattern="tr38901",
                                        polarization="V")

    # Create transmitter
    alicetx = Transmitter(name="aliceTX",
                     position=sender_position,
                     orientation=[0, 0, 0])
    simulation_scene.add(alicetx)

     #TODO: rebuild simulation with two transmitters?   
    bobtx = Transmitter(name="bobTX",
                     position=receiver_position,
                     orientation=[0, 0, 0])
    simulation_scene.add(bobtx)

    # Create a receiver
    alicerx = Receiver(name="aliceRX",
                  position=sender_position,
                  orientation=[0, 0, 0])
    simulation_scene.add(alicerx)

    bobrx = Receiver(name="bobRX",
                  position=receiver_position,
                  orientation=[0, 0, 0])
    simulation_scene.add(bobrx)

    ## RX points towards TX
    bobrx.look_at(alicetx)
    alicerx.look_at(bobtx)

    simulation_scene.frequency = frequency
    simulation_scene.synthetic_array = True

    current_paths = simulation_scene.compute_paths()
    a, tau = current_paths.cir()
    frequencies = subcarrier_frequencies(config.fft_size, config.subcarrier_spacing)
    current_cir = cir_to_ofdm_channel(frequencies, a, tau)                                                                     
    current_cir = tf.reshape(current_cir, ((2, 1000))

where <scene_file_name> is the relative path to my scene xml file:

<scene version="2.1.0">

<!-- Defaults, these can be set via the command line: -Darg=value -->

	<default name="spp" value="4096"/>
	<default name="resx" value="1920"/>
	<default name="resy" value="1080"/>

<!-- Camera and Rendering Parameters -->

	<integrator type="path">
		<integer name="max_depth" value="12"/>
	</integrator>
	<sensor type="perspective">
		<string name="fov_axis" value="x"/>
		<float name="fov" value="39.597755"/>
		<float name="principal_point_offset_x" value="0.000000"/>
		<float name="principal_point_offset_y" value="-0.000000"/>
		<float name="near_clip" value="0.100000"/>
		<float name="far_clip" value="100.000000"/>
		<transform name="to_world">
			<rotate x="1" angle="-153.5592908827048"/>
			<rotate y="1" angle="-46.691938147394204"/>
			<rotate z="1" angle="-179.999991348578"/>
			<translate value="7.358891 4.958309 6.925791"/>
		</transform>
		<sampler type="independent">
			<integer name="sample_count" value="$spp"/>
		</sampler>
		<film type="hdrfilm">
			<integer name="width" value="$resx"/>
			<integer name="height" value="$resy"/>
		</film>
	</sensor>

<!-- Materials -->

	<bsdf type="twosided" id="mat-itu_concrete">
		<bsdf type="principled">
			<rgb value="0.800000 0.800000 0.800000" name="base_color"/>
			<float name="spec_tint" value="0.000000"/>
			<float name="spec_trans" value="0.000000"/>
			<float name="metallic" value="0.000000"/>
			<float name="anisotropic" value="0.000000"/>
			<float name="roughness" value="0.500000"/>
			<float name="sheen" value="0.000000"/>
			<float name="sheen_tint" value="0.500000"/>
			<float name="clearcoat" value="0.000000"/>
			<float name="clearcoat_gloss" value="0.173205"/>
			<float name="specular" value="0.500000"/>
		</bsdf>
	</bsdf>

<!-- Emitters -->

	<emitter type="point">
		<point name="position" x="4.076245307922363" y="5.903861999511719" z="-1.0054539442062378"/>
		<rgb value="79.577469 79.577469 79.577469" name="intensity"/>
	</emitter>

<!-- Shapes -->

	<shape type="ply">
		<string name="filename" value="meshes/Plane.ply"/>
		<boolean name="face_normals" value="true"/>
		<ref id="mat-itu_concrete" name="bsdf"/>
	</shape>
</scene> ```

[jhoydis]

Hi,

I cannot execute this code as there are undefined variables. Could you also please use a scene that is provided with Sionna?
In your updated example, could you please explain clearly which outputs are unexpected?

[DerBambus]

Sorry my bad, the following code should run normally:

from sionna.rt import load_scene, Transmitter, Receiver, PlanarArray, Camera
from sionna.channel.utils import subcarrier_frequencies
from sionna.channel.utils import cir_to_ofdm_channel
import sionna
import tensorflow as tf

simulation_scene = load_scene(sionna.rt.scene.simple_reflector)
simulation_scene.tx_array = PlanarArray(num_rows=1,
                                    num_cols=1,
                                    vertical_spacing=0.5,
                                    horizontal_spacing=0.5,
                                    pattern="tr38901",
                                    polarization="V")

# Configure antenna array for all receivers
simulation_scene.rx_array = PlanarArray(num_rows=1,
                                    num_cols=1,
                                    vertical_spacing=0.5,
                                    horizontal_spacing=0.5,
                                    pattern="tr38901",
                                    polarization="V")

# Create transmitter
alicetx = Transmitter(name="aliceTX",
                 position=[0, 0, 1.5],
                 orientation=[0, 0, 0])

simulation_scene.add(alicetx)

 #TODO: rebuild simulation with two transmitters?   
bobtx = Transmitter(name="bobTX",
                 position=[10, 0, 1.5],
                 orientation=[0, 0, 0])

simulation_scene.add(bobtx)

# Create a receiver
alicerx = Receiver(name="aliceRX",
              position=[0, 0, 1.5],
              orientation=[0, 0, 0])

simulation_scene.add(alicerx)

bobrx = Receiver(name="bobRX",
              position=[10, 0, 1.5],
              orientation=[0, 0, 0])

simulation_scene.add(bobrx)

## RX points towards TX
bobrx.look_at(alicetx)
alicerx.look_at(bobtx)

simulation_scene.frequency = 2e9
simulation_scene.synthetic_array = True
current_paths = simulation_scene.compute_paths()
a, tau = current_paths.cir()
frequencies = subcarrier_frequencies(1000, 15e3)
current_cir = cir_to_ofdm_channel(frequencies, a, tau)                                                                     
current_cir = tf.reshape(current_cir, (4, 1000))
print(current_cir)

When running this code, the tensor with the current cir values looks as follows:

So am I correct in assuming the four values here are in this order?
alice -> alice
bob -> alice
alice -> bob
bob -> bob

If this is the case, I would expect the CIRs of bob -> alice and alice -> bob to be the same, however these values never seem to match (except for channel noise, which currently isn't part of the simulation I believe).
I've tried using different frequencies, different scenes, etc., but the problem seems to persist.

Thank you for your troubles^^

[jhoydis]

Hi,

The issue in your example is that you are using directive antenna patterns. AliceTX is oriented towards BobRX because the default orientation points toward the positive x-axis. However, BobTX points away from AliceRX which is located in the opposite direction..

If you change the antenna patterns to "iso", you get the results you expect. Alternatively, you can make the transmitters to look at the receivers. You would need to add these two lines to your code to achieve this:

## TX points towards RX
bobtx.look_at(alicerx)
alicetx.look_at(bobrx)