Introducing Sionna: A TensorFlow-Based Library for Prototyping and Evaluating Communication Systems
Sionna is a TensorFlow based, open source library for link level simulations with GPU concurrency. It streamlines the iterative design for complex communication systems architectures and leans towards a native neural networks integration. Sionna has a collection of numerous thoroughly validated techniques which can be used for performance assessment and verification for integral performance. This helps the scientists to concentrate more on their work leading to a more productive and efficient action with the time that would have been used implementing features that they have no training on. This white paper presents Sionna, including its main principles and features as well as areas for extensions: forSionna, such, for ray tracing and integrating custom CUDA cores. We consider that Sionna is a useful instrument for the conduction of investigation over communication systems of the future generations like 6G and we invite initiatives from our community.
