Sionna

Sionna™ is a hardware-accelerated differentiable open-source library for researchon communication systems. It is composed of the following modules:

• Sionna RT: A lightning-fast stand-alone ray tracer for radio propagation modeling

• Sionna PHY: A link-level simulator for wireless and optical communication systems

• Sionna SYS: System-level simulation functionalities based on physical-layer abstraction

The core principles of Sionna are modularity, extensibility, and differentiability.

Every building block is an independent module that can be easily tested,understood, and modified according to your needs. The documentation is completeand includes references. Similar to constructing a deep neural network bystacking different layers, complex communication system architectures can berapidly prototyped by connecting the desired blocks.

Sionna PHY andSionna SYS are written inTensorflow, whileSionna RT is built on top ofMitsuba 3 andDr.Jit. These frameworks provide automatic differentiation and can backpropagate gradients through an entire system. This is the key enabler for gradient-based optimization and machine learning, especially the integration of neural networks.

NVIDIA GPU acceleration provides orders-of-magnitude faster simulation, enablingthe interactive exploration of such systems, for example, inJupyter notebooks that can be run on cloud services such asGoogle Colab. If no GPU is available, Sionna will run on the CPU.

TheSionna Research Kit (SRK) allows to deploy trained AI/ML components in a real software-defined 5G NR radio access network (RAN). It is based on theOpenAirInterface project and is powered by theNVIDIA Jetson AGX Thor platform.

Sionna is developed, continuously extended, and used by NVIDIA to drive 5G and 6G research.