Cong Shen, an associate professor of electrical and computer engineering at the University of Virginia’s School of Engineering and Applied Science, is combining his engineering experience in the field of telecommunications and his recent machine learning research to spin out a new company called WiSights Lab, which announced its formation this spring.
Shen’s new venture, is providing artificial intelligence-based tools for telecom engineers in a space called ORAN, which is short for open radio access network.
If you haven’t heard of ORAN before, you will. It’s the future of telecommunications, and it’s a huge opportunity for entrepreneurs. Globally, the ORAN market is estimated to be well over $1 billion currently. But by 2032, it’s expected to grow to $146 billion.
Understanding ORAN
Telecommunication providers have traditionally operated using closed systems, restricting access to only certain users working within their proprietary frameworks. But what happens when 5G providers open up their networks for greater collaboration?
The answer is an open radio access network. The network is essentially a set of industry standards that promote open, interoperable and software-defined approaches to designing and deploying mobile network infrastructure.
“Telecommunication traditionally is a very closed system,” Shen said. “ORAN is trying to open up this ecosystem so that everybody can contribute within the major components: cloud-connected hardware and software, the intelligent management systems that drive modern telecom networks and the physical radio apparatus itself.”
The idea is that by promoting open interfaces and standardized access, tech vendors can more quickly develop and scale up improvements, all for the benefit of the customer.
“It’s like a Lego kind of thing,” Shen said regarding interoperability. “You know, maybe Ericsson only provides one part of it, Motorola maybe can do something else, and some small companies may be doing a third part.”
Making Collaboration Easier
Prior to joining academia, Shen worked for Qualcomm Research, a large mobile technology pioneer, as well as SpiderCloud Wireless, a smaller company that develops scalable, small-cell network platforms.
He knew collaboration through ORAN was where the industry was heading. So he and partner Vijay K. Shah, an assistant professor of electrical and computer engineering at North Carolina State University, dreamed up a business that could guide vendors who need to use it.
They started by creating the open-source ORANSight, which could answer developer questions and generate ORAN code through large language models. These LLMs help accelerate ORAN development by optimizing network management, predicting failures, automating configurations and enhancing security — all through AI-driven insights.
We saw an opportunity of actually enabling some of the first large language models for ORAN, just because this is so new. Even seasoned telecom engineers are facing challenges to their understanding of how to develop under the ORAN frameworks.
Shen said that open 5G systems like ORAN speed up innovation, lower costs and boost competition by allowing more tech vendors to jump in improve the networks. That same openness also allows for greater collaboration among vendors. Either way, the result is more rapid deployment.
Shen and Shah are now working to make their tools more customized, based on individual clients’ needs.
“We saw an opportunity of actually enabling some of the first large language models for ORAN, just because this is so new,” Shen said. “Even seasoned telecom engineers are facing challenges to their understanding of how to develop under the ORAN frameworks.”
Their flagship large language model has now evolved to ORANSight2.0. The domain-specific generative AI is essentially like a smart assistant for these open 5G developers. What otherwise might take a day of internet searches and phone calls now takes seconds.
Enhancing Security Amid Openness
WiSights’ other major service is WiLM-Sec, which is short for Wireless Language Model Security. This collection of tools can help protect against attacks such as “prompt injections,” which try to trick an LLM model into following malicious instructions. The answers could inadvertently reveal sensitive information.
An example of an exposure might be real people’s data that was used to train the learning model.
Shen said the United States and its military have been active proponents of ORAN development for years, but they insist that it has to be secure. The locations and movement patterns of Armed Services telecom users could be exposed.
“But now there are maybe 20, 30 vendors coming into your infrastructure,” Shen said as a hypothetical. “So we are providing a large number of models to do this kind of a security check.”
Leading Through Innovation, Involvement
Shen, a 2022 National Science Foundation CAREER Award winner who has been using his research to designing for the step-up from 5G to sixth-generation networks, is a leader in this tech space.
He remains on top of industry developments as an editor with several of the Institute of Electrical and Electronics Engineers publications, as well as by serving as an IEEE conference organizer.
For the new company, Shen provides the expertise on the machine learning side, while he defers to Shah as the ORAN expert.
He said his recent research into LLMs, transformers and in-context learning, for example, has made a difference in their new pursuit.
Supporting Academic Entrepreneurs
Knowing UVA supports future-leaning entrepreneurial efforts has also made a big difference in maintaining his momentum, Shen said.
He has been working with the Enterprise Studio of the UVA Licensing and Ventures Group as well as David Chen, assistant dean for entrepreneurship and innovation partnerships, in his capacity as the managing director of the Wallace H. Coulter Center for Translational Research.
“I am honored to have been talking with Cong about this opportunity,” Chen said, emphasizing that both the Coulter Center and Licensing and Ventures strongly support faculty members who dive into the “thriving entrepreneurial ecosystem UVA has cultivated.”
It’s a culture that Engineering Dean Jennifer L. West has promoted since joining UVA in 2021. The idea is that big research breakthroughs will translate more quickly to society if academic researchers feel invested in the outcome.
For Shen’s part, he said, “I really appreciate the University for trying to commercialize more of what we do here. Over time, it benefits everyone.”