GTC 2020: Streamlining Signal Processing and Deep Learning for Radio and 5G Networks
After clicking “Watch Now” you will be prompted to login or join.
Click “Watch Now” to login or join the NVIDIA Developer Program.
Streamlining Signal Processing and Deep Learning for Radio and 5G Networks
Adam Thompson , NVIDIA | John Ferguson, Deepwave Digital
With emerging bandwidth and data-rate hungry technologies like 5G, there's an essential need to make better use of the wireless spectrum—allowing signal transmission in bands designated for other purposes when available. In this joint talk between NVIDIA and Deepwave Digital, NVIDIA will provide an overview of the RAPIDS cuSignal library (GPU-accelerated SciPy Signal), focused on building high performance signal processing workflows from Python. Deepwave will expand on this infrastructure and highlight their work creating the first deep learning radio frequency sensor for the Citizens Broadband Radio Service (CBRS) 5G network. By leveraging the Artificial Intelligence Radio Transceiver (AIR-T) and its NVIDIA Jetson TX2i, the Deepwave team implemented a deep neural network on the AIR-T that is capable of channelization, detection, classification, and reporting the presence of signal emissions coming from non-cooperative priority users with extreme accuracy.