The RF Digital Twin

Physical · Differentiable · Rendering
Next-Gen Wireless Sensing and Communication Research Platform

Get StartedInstallLaunch Web

Get Started

Core

Core RF digital twin engine

pip install witwin
MoreGitHub

Studio

Visual simulation editor

Coming Soon

Maxwell

Differentiable full-wave EM solver

pip install witwin[maxwell]
MoreGitHub

Radar

Differentiable mmWave radar simulation

pip install witwin[radar]
MoreGitHub

Channel

Differentiable wireless channel modeling

Coming Soon

Satellite

Differentiable satellite link simulation

Coming Soon

Genesis

Synthetic RF data generation

Coming Soon

Build the Future of Wireless Simulation

A complete platform for RF simulation, optimization, and research

Multi-method electromagnetic solver

Next Generation EM Simulation

Choose the right simulation method for your needs. From fast ray tracing for large-scale scenarios to full-wave solvers for high-accuracy analysis.

Learn more
TXRX
Paths: 847Bounces: 3Time: 2.3ms
Gradient computation for optimization

End-to-End Differentiable

Compute gradients through the entire simulation pipeline. Enable neural network integration and inverse design optimization.

Learn more
LossObservationSimulated SignalSignal GradientRFDTSimulatorParameterizationSceneMeshingGradientParametersValueGradForwardGradient
Deep learning powered digital twin

Integrating with Neural Networks

Seamlessly integrate physics-based simulation with deep neural networks. Train end-to-end models that combine learned representations with accurate physical modeling.

Learn more
LossPhysical Twin IntrinsicRFSignalInputDNNShapeMaterialFormWaveSensorMotionRFDTSimulatorRFSignalOutputFeed ForwardPhysics Model
Node-based configuration interface

Visual Editor

Design and configure simulation scenarios with an intuitive visual editor. Real-time preview and parameter adjustment.

Learn more
HIERARCHY
▼ Scene
├ Transmitter
├ Building_01
├ Building_02
└ Receiver
3D Viewport
NODE EDITOR
∂ HyperparamTX PosFreqSceneComputeChannelPaths
PROPERTIES
Position0.0, 1.5, 0.0
Frequency77 GHz
Flexible scripting interface

Python SDK

Full programmatic control with a clean Python API. Integrate with your existing ML workflows and data pipelines.

Learn more
simulation.py
import rfdt
# Create scene
scene = rfdt.Scene()
scene.add_transmitter(pos=[0,0,0])
# Run simulation
result = scene.simulate()
grad = result.backward()
AI Assistant

Add a human and dog running in a living room. Use 60GHz MIMO radar sensing.

Creating scene with animated targets...

scene.load("living_room")
scene.add_actor("human")
scene.add_radar(freq=60e9)
Ready
Ask AI Agent to modify simulation...

Developers

Xingyu Chen

Xingyu Chen

Fourth Year PhD Student

UC San Diego

Founder and Developer

Prof. Xinyu Zhang

Prof. Xinyu Zhang

Professor, Director of Center of Wireless Communication

UC San Diego

Advisor

Citation

If you use RF Digital Twin in your research, please cite our paper

BibTeX
@inproceedings{chen2026rfdt,
  title     = {Physically Accurate Differentiable Inverse Rendering
               for Radio Frequency Digital Twin},
  author    = {Chen, Xingyu and Zhang, Xinyu and Zheng, Kai and
               Fang, Xinmin and Li, Tzu-Mao and Lu, Chris Xiaoxuan
               and Li, Zhengxiong},
  booktitle = {Proceedings of the 32nd Annual International Conference
               on Mobile Computing and Networking (MobiCom)},
  year      = {2026},
  doi       = {10.1145/3795866.3796686},
  publisher = {ACM},
  address   = {Austin, TX, USA},
}

Licensing

Research

Free

For academic and non-commercial research purposes

  • Full access to simulation capabilities
  • Academic publications allowed
  • Community support
Get Started

Commercial

Contact Us

For commercial and enterprise applications

  • Production deployment rights
  • Priority support
  • Custom integration assistance
Contact Us