Sionna: Introduction to Embedded Open-Source Semantic Communication Platforms

Recently, in the wireless communication protocol, the machine learning (ML) technique has emerged as the most promising approach for designing the entirely model-free end-to-end protocol or the partly model-free protocol by replacing only a few parts. Yet, its main challenge is the consistency of the simulator for each functional block in wireless communication. Even though many research outputs shed light on the ML for wireless communication topic, most of the results do not provide a reasonable baseline or are incompatible with conventional systems, such as 4G-LTE, and 5G-NR. This paper introduces the overview of the python-based open-source library, Sionna [1], enabling us to easily employ the 4G-LTE and 5G-NR compatible functional blocks in CUDA-GPU ML framework. Besides, one exemplary result, "trainable QAM constellation with Polar coding,"is presented, which motivates us to study more practical research on ML for wireless communication.