A University of Nebraska-Lincoln team led by computing professor M. Can Vuran is part of a national push to combine artificial intelligence with next-generation wireless networks — one of just 278 projects selected for the U.S. Department of Energy's new Genesis Mission.
On July 22, the U.S.Department of Energy announced the awards for the Genesis mission. The DOE has called the response the largest to any funding opportunity in the department's history. The agency's own list of the 278 selected projects confirms UNL's inclusion, naming Vuran's project "SwarmSlicer."
The DOE's Genesis Mission combines artificial intelligence, supercomputing, quantum systems and advanced scientific instruments to build what DOE calls the world's most powerful scientific platform.
Vuran's team is building what he calls an AI "autopilot" for designing 6G networks – the next generation of cellular network technology.
A new generation of cellular network technology
"We are all connected these days with our cellphones and devices. A majority of the traffic is supported through Wi-Fi cells or cellular networks,” Vuran said. “So currently, we are using 5G standards for cellular networks. This is what our cell phones are connected to," The next generation, he said, will bring more than just faster speeds.
"The next generation will be the sixth generation, 6G networks, which will provide significantly new capabilities — including higher throughputs, more data rates, more diversified services, larger coverage, and several other functionalities," Vuran said. "Core to 6G technology will be a native integration of artificial intelligence in the operation of these systems."
That added complexity, Vuran said, is outpacing what engineers can design using human oversight alone, driving the need for AI integration in the new technology.
"These systems are becoming more and more complex. So it's becoming more challenging to design and innovate on these systems using only conventional, human-driven design solutions," he said. "So our project is geared towards building what you can call an autopilot — an artificial intelligence-based autopilot system — that helps humans co-design next-generation wireless technology."
Faster design, faster science
Vuran said accelerating innovation is important. He said his team's goal is to turn a current design process that takes months into just a few days.
"Our ultimate goal is to turn this design process into one that takes only a few days, so that we can accelerate the innovation pipeline," Vuran said. "We can try out new methods much faster. We can get results much faster, and then we can iterate on them."
That speed matters for more than convenience, he said. DOE's broader scientific missions increasingly rely on robots and sensors operating in remote or unpredictable environments — including, he said, potential future missions to Mars.
"Our goal is to leverage the robots and sensors within these missions to provide scientific information to the scientists in a more reliable manner, in unpredictable terrains," Vuran said.
The cost problem
Vuran said cost is another problem the team is trying to solve. Outfitting every sensor or robot with its own AI-capable GPU is expensive and power-hungry.
"The cost of GPUs is increasing significantly with the increasing demands for AI workflows," Vuran said. "There is also the energy cost perspective — as you add more of these components into scientific equipment, sensors, robots, you're increasing the cost of that equipment as well."
Rather than scale back scientists' access to AI because of that cost, Vuran said, the team is designing the network itself to share the load.
"We don't want to strip scientists from these advancing AI capabilities just because of cost," Vuran said. "That's where 6G comes in, where we can offload some of these AI workflows — whether on the robot, on the edge, on a cloud setting, or even on leadership facilities and supercomputing facilities.”
A team effort, with an unexpected member
"We have a strong team of collaborators, both from academia, from national labs and industry," Vuran said.
The project brings together Brookhaven National Laboratory, contributing supercomputing and high-performance computing expertise; ALPEMI Consulting LLC, specializing in data movement and cloud/HPC operations; and Hewlett Packard Enterprise, bringing AI hardware and software expertise — along with Vuran's own Cyber-Physical Networking Lab at UNL, which specializes in wireless systems and Internet of Things technology.
Vuran also credited Santosh Pitla, a professor and associate head for research and innovation in UNL's Biological Systems Engineering department, and a co-director of NU's AI Institute.
"With Dr. Pitla's involvement, we will leverage the AI Institute's capabilities. We also hope to contribute to the AI Institute, as well as the College of Engineering's PRAIRIE Initiative," Vuran said. "You can see that UNL will have significant leadership in AI research and innovation, as well as education. Our goal is to provide leadership in AI technologies when it comes to connectivity."
Vuran said landing a spot in a program this competitive is a reflection of the team he's built.
"We are honored to receive this award. It is a testament to the teams that we have built, the team's expertise, and we are really excited to be selected as one of the groups being considered for this grant, and we hope to contribute significantly to the Genesis Mission," Vuran said.
The DOE has framed the Genesis Mission around maintaining U.S. leadership in AI amid growing competition with China. Vuran cast his own project in similar terms, calling it a "once-in-a-lifetime" scientific program.
"We believe our project will be a critical step in securing U.S. leadership in the global race, and we also hope to advance the DOE's scientific missions by providing connectivity in challenging environments," Vuran said.