Joint Communication and Sensing for 6G is missing in Sionna

[akesha01]

Hi everyone!

I am researching on Joint Communication and Sensing capabilities which is one of key elements in 6G. I checked and I couldn't anything related in Sionna library. I was wondering isn't it necessary for this library to have such a element in it? I appreciate if you help how can I start to add Sensing capability to Sionna.

[jhoydis]

Hi @akesha01,

Would you have any concrete block in minds that are missing for Sensing in Sionna?

[akesha01]

Yes, here is a list of blocks I am developing for ISAC (Integrated Sensing and Communication) based on Sionna classes and modules:

Radar Channel: This is a two-way path that accounts for the round-trip delay of the signal, Swerling (RCS) fluctuations of the target, and clutter (with ray tracing). Doppler and multipath effects are already present in Sionna channel blocks.

Radar Receiver: This involves radar signal processing for target range, velocity, and angle-of-arrival estimation. It includes matched filtering (for range), Doppler processing (FFT), and beamforming (for angle estimation).

Waveforms: Additional waveforms, such as OTFS, need to be investigated for ISAC.

Designs for ISAC Scenarios: The following aspects need further investigation as they may change when integrated with radar: constellation design, waveform design, beamforming design, and space-time signal design.

[jhoydis]

Hi @akesha01,

Do you have already implemented some of these blocks that could be potential candidates for a PR? If yes, I would be happy to have look.

[akesha01]

Hi @jhoydis
Sure. I have worked on the radar channel and radar receivers. I will tidy up the code and provide it here by next week.

[sssxxxfan]

I would also like to know how to implement RCS target single point or multipoint modeling or calculation methods in ISAC channels in sionna。are u already have any idears or codes?

[akesha01]

In Radar theory, target RCS fluctuations are modeled using Swerling models (Swerling 1,2,3,4).

The target RCS fluctuates in time mainly because the RCS of complex shapes depends on the angle of radiation. Even small changes in angle can significantly alter the RCS. Other factors, such as target motion, frequency, and scatterer interference, also contribute.

The choice of Swerling models depends on how many scatterers the target has and their size.

For Swerling 1, RCS follows an exponential distribution, meaning only one random variable is needed to model fluctuations. This variable determines the received power level, which varies from pulse to pulse.

I am working on Radar simulation and I will put the code here soon.

[sssxxxfan]

I intend to use sionna for single-point or multi-point RCS modeling of the target object, and my initial idea is to emit a cluster of rays for a single point and get the ratio of the received power to the transmitted power. For multiple points, such as cars, I want to let a certain number of rays be emitted to the front of the car, and some of them are transmitted to parking spaces and other positions, and I can also get the ratio of received power to transmitted power, can this ratio be RCS?

[akesha01]

Great idea to use ray tracing for modeling channels in Sionna! This is what I intended to do as well.

For a simple simulation, you can assume that the target (e.g., a car) has a basic shape, such as a rectangular plate, and add it to the scene. Then, you can modify its radio_material attributes to model the reflection from the target. In this case, the radar cross-section (RCS) can be estimated as the ratio of the received power to the transmitted power, as you mentioned.

However, for a more realistic simulation, a target typically consists of multiple scattering points, meaning its RCS fluctuates over time and aspect angle. As a result, the received power cannot be modeled as a fixed deterministic value. Instead, we should introduce a statistical model to account for RCS fluctuations. In the case of a Swerling 1 model, for example, the RCS follows an exponential distribution, meaning the received power (given a constant transmitted power) is also exponentially distributed with a certain mean value.

[sssxxxfan]

I would also like to know how to implement RCS target single point or multipoint modeling or calculation methods in ISAC channels in sionna.are u already have any idears or codes?