Sionna
An open-source, GPU-accelerated link-level simulation library released by NVIDIA. Built on TensorFlow, it supports simulation of 5G NR and 6G candidate physical-layer links, including OFDM waveforms, channel models, DMRS pattern configuration, and differentiable end-to-end neural receiver training.
Timeline (last 90 days)
The paper evaluates using OFDM simulations generated by Sionna, with the CNN-based DeepRx receiver as the comparison target, covering three DMRS layouts—dense, two-symbol sparse, and one-symbol sparse—and also examines channel mismatch scenarios.
Sionna is an open-source GPU-accelerated link-level simulation library released by NVIDIA, based on TensorFlow, supporting fast simulation of 5G NR and 6G candidate physical-layer links, including OFDM waveform generation, channel models, DMRS pattern configuration, and neural receiver training.
All Facts (2)
The paper evaluates using OFDM simulations generated by Sionna, with the CNN-based DeepRx receiver as the comparison target, covering three DMRS layouts—dense, two-symbol sparse, and one-symbol sparse—and also examines channel mismatch scenarios.
50%UnverifiedSionna is an open-source GPU-accelerated link-level simulation library released by NVIDIA, based on TensorFlow, supporting fast simulation of 5G NR and 6G candidate physical-layer links, including OFDM waveform generation, channel models, DMRS pattern configuration, and neural receiver training.
50%