Hi,
I’m currently generating wireless datasets using NVIDIA Sionna (e.g., channel coefficients, CSI matrices, and received signal samples) for AI-RAN simulation research.
I’d like to confirm whether these Sionna-generated datasets can be directly imported into Aerial Omniverse Digital Twin (AODT) simulations, or if pyAerial is the only supported interface for dataset ingestion and simulation orchestration.
My goal is to establish an offline integration workflow between Sionna and AODT — without deploying the CUDA-Accelerated RAN container.
Any insights or documentation references would be greatly appreciated.
Thanks!
@hychen2
Sionna dataset cannot be imported into AODT. What data do you wish to import? AODT should be able to generate a similar dataset on its own.
This is directly relevant to something
I am working on.
I am building Sionna Visualizer a web
dashboard for Sionna simulation results —
and planning AODT integration in Phase 2.
The main challenge I am anticipating is
the data format bridge between Sionna
tensor output and AODT input format.
Did you find a clean way to export Sionna
channel matrices into AODT-compatible
format? Or did you end up using pyAerial
as the bridge layer regardless?
Any insight on the integration path would
help me a lot as I plan Phase 2.
@vinayakgote44 Can you please elaborate on what you mean by “data format bridge between Sionna tensor output and AODT input format”.
AODT does not consume any external data. Do you mean Aerial rather than AODT?
@kpasad1 Thank you for the clarification that helps a lot.
You are right, I was conflating AODT and Aerial. Let me clarify
what I actually mean.
My goal is to take Sionna simulation outputs specifically BER
curves, channel coefficients, and path loss data and visualize
them alongside a digital twin scene. The idea is not to feed
Sionna data into AODT as input, but to overlay Sionna simulation
results on top of an AODT-rendered environment for visual analysis.
Is there a supported way to do this kind of overlay Sionna
results displayed in context of an AODT scene? Or is Aerial
the right layer to be looking at for this use case?
@vinayakgote44 AODT is composed of two parts, the back end and the front end. The front end is a vizualizer that is build on top of omniverse. The vizualizer opens a map in the USD format. The USD map can be created from a CityGML/OpenStreetMaps using our GIS pipeline.
Do you wish to plot the metrics at the specific lat/long on the USD maps in Omniverse?
This is not supported out of the box in AODT 1.4.1. In AODT 1.5, which will be released towards end of April, AODT will go headless- i.e no UI. However, there will be guidance on using any UI of your choice, including Omniverse or Cesium JS, to overlay prims on top of a map.
That makes a lot of sense especially the shift towards a headless AODT with external UI control.
Yes, exactly my intention is to map simulation metrics (like BER, path loss, etc.) to spatial coordinates and visualize them over a scene.
If AODT 1.5 enables external UI integration, then using something like Omniverse or even a custom web-based UI for overlay sounds like a really interesting direction.
I’m currently exploring how to structure simulation outputs so they can be mapped cleanly to spatial context would be great to experiment once that release is available.