This is the sixth of six (6/6) posts serializing the FAQ report from the most recent NVIDIA 6G Developer Day (sessions playback here). All questions and answers are directly from attendees on the day and the answers provided to them.
The Sections are:
1/6: Introducing NVIDIA 6G Research & Development Portfolio
2/6: Prototyping 6G with the NVIDIA Sionna Research Kit
3/6: Designing a Modular 6G System Using NVIDIA Aerial Framework
4/6: Aerial Framework – Performant GPU-Accelerated Code From Python
5/6: Optimizing 6G Performance With CUDA and NVIDIA DOCA
6/6: Simulating 6G with NVIDIA Aerial Omniverse Digital Twin
The full FAQ report is here:
2025_6GDD_NVIDIA AI Aerial_FAQ.pdf (374.9 KB)
Simulating 6G with NVIDIA Aerial Omniverse Digital Twin
Q 6.1: In local setup using RTX 5070 TI, will it be possible to run AODT?
A: Possible after release 1.5, the scenario size might be small on such HW.
Q 6.2: Is AODT compatible with end-to-end Aerial Testbed stack?
A: No
Q 6.3: Can we deploy digital twin in open RAN technology?
A: Yes
Q 6.4: Is/will it be possible to simulate mesh networks with AODT? How many nodes can be handled by single GPU settings?
A: It will be possible - starting from release 1.5 - to have D2D for channel simulation.
Q 6.5: What is the strong point of AODT compared to existing Radio Planning Tools such as Atoll, Asset, etc. that can already do/did the RF simulation in a 3D model along with detail RF propagation characteristics, etc.
A: AODT EM solver is GPU-accelerated, more accurate, faster, and can be scaled to very large deployments across multiple GPUs.
Q 6.6: Is it possible to architecturally separate the “source of truth” for RAN (OAI/Sionna) and the “source of truth” for geometry (USD/Omniverse), so that the digital twin can be used both without Omniverse (CLI pipeline) and with Omniverse (visualization) without duplicating logic?
A: Yes, but full RAN / UE stack is currently not implemented in AODT.
Q 6.7: What Geo sources/formats can be used for map layer, OSM?
A: OSM, cityGML, and Blackshark from release 1.6.
Q 6.8: Would it be possible to simulate shaping of the moving objects with microwave and some massive MIMO?
A: This is not yet possible with AODT native EM solver.
Q 6.9: How can AODT fit within NVIDIA Aerial Framework context?
A: AODT can make sure of concepts developed with the Aerial Framework, at different levels of hardening.
Q 6.10: How would the AODT service be used for an indoor system, and would the workflow be different?
A: No. Same workflow. The solver supports indoor environment including stairs and elevators inside the building.
Q 6.11: Are Sionna RT and AODT interchangeable?
A: Integration is not direct, but the channel impulse responses derived from AODT can be used in simulation using Sionna, with the required integration taking place in Python.
Q 6.12: Does AODT have any installation security pre-requisites? I have noticed following the installation instructions that the .zip file produces 403 errors.
A: Membership of NVIDIA 6G Developer Program is required to download AODT 1.4 and below.
Q 6.13: Can AODT be used for measuring indoor coverage?
A: If the scene can be described in a format AODT support (i.e., cityGML), indoor is supported.
Q 6.14: Where does this flow fit in the entire network connectivity path - like is it between RU and DU or DU and CU?
A: The flow illustrated in the presentation is only about the channel, which is between RUs and UEs.
Q 6.15: In Tokyo map used, are the material of buildings considered?
A: Yes
Q 6.16: How can you define Data Traffic between UEs and RUs?
A: At this stage, AODT is only providing channel realizations and a full-buffer RAN mode.
Q 6.17: For maximum accuracy in sub 6GHz, is there a rule of thumb on how to configure diffraction in the AODT?
A: In terms of depth, only 1 diffraction per ray is supported.
Q 6.18: How similar would the AODT results of Tokyo scenario be with measurements? Is there any study on comparing the Tokyo scenario with measurements?
A: Measurement data for Tokyo is unavailable. However, with measurement data from Finland the accuracy is quite good, especially after calibration.
Q 6.19: How similar or different would AODT be with Sionna?
A: AODT is an enterprise tool, which aims at providing the highest possible accuracy for the EM solver to realize a full radio environment twinning, and its infrastructure is scalable (starting from release 1.6) to support diverse dimensions of the simulations.
Q 6.20: Any plan to allow time convolution channel emulation or manipulation of time IQ samples/waveforms in AODT?
A: It’s already present when running in RAN mode.
Q 6.21: Can we see PHY/MAC logs for each UE such as PRACH failures etc.?
A: Currently, the RAN mode is limited to PxSCH and SRS. Acquisition and cell mobility is achieved with custom logic.
Q 6.22: Can field data log be used with AODT to improve accuracy?
A: Yes, from release 1.5, SS-RSRP can be used to calibrate the EM solver.
Q 6.23: How can we use this for ISAC based Environmental reconstruction research?
A: This is not yet possible with AODT native EM solver.
Q 6.24: What are the hardware requirements and best practices for effectively deploying and running Digital Twin technologies in O-RAN use cases, including large-scale simulations on platforms like Sionna?
A: Sionna runs on any (recent) hardware. However, for large scale simulations sufficient GPU memory is required. For example, we simulated 5G coverage across the continental US in under five minutes by tracing more than 35 trillion rays on 96 NVIDIA L40S GPUs. The code is available as open source GitHub - NVlabs/sionna-large-radio-maps: Large-scale simulation of wireless network coverage maps using Sionna RT. · GitHub