End-to-end Enterprise 5G and Edge AI utilizing GPU technologies

The trends of network virtualization are coming to 5G radio access network (RAN), and GPU is expected to be one of the accelerators for radio signal processing, as well as AI applications on multi-access edge computing (MEC). We've evaluated the performance of “NVIDIA Aerial,'' which is a GP- based radio signal processing SDK for vRAN itself. We've also evaluated the resource separation functionalities for coexistence with some AI applications in the edge computing environment. Finally, NVIDIA Aerial is integrated into the commercial mobile network product developed by Mavenir. It means a GPU-based vRAN solution is ready to evaluate end-to-end 5G communication. We’ll share the overview of our recent activities toward enterprise 5G network utilizing GPU technologies. We evaluate the performance of end-to-end 5G communication using a GPU-based vRAN solution. We also demonstrate and evaluate the AI on the MEC, such as IVA and Virtual Positioning System.