Differentiable Ray Tracing for Modeling of Complex Radio Environments

At EMS we wish to first measure, then estimate, and finally model radio channels as accurately as possible. For this we develop dedicated, high precision measurement devices, so called channel sounders, that are able to capture spectral, spatial and temporal information about radio wave propagation. However, the effort in terms of hardware and the subsequent data processing is rising disproportionally with measurement bandwidth, scenario complexity and required accuracy.

To avoid this, we wish to pair the conducted radio measurements with a suitable parametric model that allow to accurately simulate the dominating propagation phenomena. This would allow us to generate radio channel realizations cheaply, quickly, especially in complex scenarios without costly measurements. In our case, so-called differentiable ray tracing tools can turn a digital description of an environment or an object into a simulation. This can be done at rapid speed due to GPU-based hardware acceleration.

With their differentiability these ray tracing tools can be neatly integrated into deep learning models. This in turn immediately equips the resulting machine learning architectures with an understanding of the physical processes that underpin radio wave propagation. Potentially, this alleviates us a little from the burden of doing overly excessive measurements. Instead we let the simulation do the heavy lifting, while we provide accurate calibration data with our channel sounder. Ultimately this offers us exciting (at least for us) research directions that have applications in future communication, radar and hybrid systems.

Problems we want to solve with you §

• How can we use our cutting edge measurement systems to calibrate parametrizations of digital models of complex environments or objects?
• Can we derive deep learning assisted measurement strategies that turn our measurement platforms into intelligent and resource efficient agents that avoid recording of redundant information?
• What additional features and physical phenomena need to be offered by existing differentiable ray tracing solutions? How do we implement those efficiently?

You should be already skilled in or eager to learn the following things §

• You have a good understanding for the internal workings of an optical or RF ray tracer.
• You know how to write code that others love to read and use. You curate beauty and style in your daily routine when refactoring legacy code bases and writing code against high performance computing APIs.
• You get excited when code is running blazingly fast on massively parallel hardware that your team members use for their simulations.
• When you develop substantially new science you are eager to present it to the rest of the research community.

What we do not care about §

• Whether you have been programming in Python, or JS, or MATLAB, or PTX-assembly, or punch cards for that matter.
• Whether you studied physics, or engineering, or computer science or even mathematics. If you are passionate, so are we!

Contact §

Send your application documents (cover letter, cv, degrees) directly to ems@tu-ilmenau.de