By Chris Donkin
Ericsson, Nvidia plot virtual RAN revolution
Edge computing specialist Nvidia revealed it had worked with Ericsson to develop advanced virtualised RAN (V-RAN) infrastructure, an innovation the two said would help operators better deploy AI applications over 5G.
Speaking at a keynote held on the eve of the event’s formal opening, Nvidia CEO Jensen Huang (pictured, right) said the pair had partnered to develop software-defined 5G RAN, which he added initially neither company “knew if it could be done”.
Following the revelation, Ericsson EVP and head of Business Area Networks Fredrik Jejdling (pictured, left) was brought on stage to discuss the potential for software-defined RAN technology, along with its impact for both the vendor’s business and operators.
Jejdling said the companies explored how to define the “next generation of V-RAN using the GPUs produced by Nvidia and combined with our years of wireless industry knowledge to create something that’s powerful and flexible for our customers”.
“Mobile operators across the world are looking for different architectures enabling them to take advantage of 5G. Our collaboration is figuring out an efficient way of providing that,” he added.
Huang noted the deployment of 5G V-RAN by operators was one of the last key components of the “smart everything revolution,” alongside adoption of cloud-native applications, AI and GPU technologies.
Development work on the software-based RAN platform is still ongoing.
AI advances
In addition to the partnership with Ericsson, Huang highlighted advances in AI already made using Nvidia’s computing platform and backed 5G to provide further advances once advanced capabilities are brought to the edge.
To this end, he also unveiled a partnership with software company Red Hat to provide its GPU and AI computing platforms to operators, and unveiled the EGX, which it claimed was the “world’s first super-computing edge server optimised for AI”.
Among the sectors Huang pointed to as having potential for AI at the edge are connected cars; remote farming; and call centres, where data needs to be kept locally for privacy or legal reasons.