7/11/2025 Jenny Applequist
Professor Milton Feng at The Grainger College of Engineering at the University of Illinois Urbana-Champaign has been working with Foxconn Interconnect Technologies on a series of Networked Intelligent Components and Environments research projects.
Written by Jenny Applequist
Milton Feng has been working with Foxconn Interconnect Technologies (FIT) on a series of Center for Networked Intelligent Components and Environments (C-NICE) research projects on high-speed signal integrity for five years—almost since C-NICE first kicked off as a collaboration between the University of Illinois Urbana-Champaign and FIT. The projects first bore fruit through a simulation effort that solved challenging real-world design problems. The team more recently headed into an ambitious new phase in which real devices are being tested, and a new FIT-supported testing program is being established on campus this summer.
FIT is one of the world’s leading suppliers of AI interconnects, presenting it with a fundamental problem faced by the whole industry: as data rates get faster and faster, it becomes more and more difficult to maintain the integrity of signals as they pass through interconnects. In 2020, FIT provided Feng with some of their design specifications, and told him that they had been struggling with signal integrity issues. Could he help?
Feng—who is the Nick Holonyak, Jr. Endowed Chair Emeritus Professor in Electrical and Computer Engineering (ECE), the Coordinated Science Laboratory (CSL), and the Holonyak Micro & Nanotechnology Laboratory (HMNTL)—explained that the current and rising generations of AI interconnects are “no longer just simple circuits or wires.” It was no easy matter to determine why signals were being degraded on their way to receivers, but his team worked tirelessly to develop sophisticated simulations that finally teased out the answer. “We had to go back to fundamental physics of electromagnetic waves... and understand how that worked, what was the root cause of the problem.”
Further, on top of the simulation effort that began in 2020, in 2024 FIT started bringing him their real 224Gbps connector products to test physically in his HMNTL laboratory—and the team successfully isolated a problem through testing.
“And so therefore they went back to fix the problem right away... and their products suddenly had the best performance... better than competitors’,” Feng said. “So that’s a joyful point, right?”
Feng repeatedly used the word “joyful” in describing the success of the work. He said it’s one thing to publish a paper that describes interesting ideas—but that for an engineer, there’s far more satisfaction when someone tells you that because you told them how to fix their problem, their product is now the best. “So that’s what the joyful research I have is,” he said. “We were able to help FIT to identify the problem and solve the problem for use in high-speed data communication.”
In technical terms, the achievements included identification of underlying causes of ground line cavities and signal stubs that were causing signal loss and crosstalk on connectors, development of a multimodal theory identifying energy transfer mechanisms from signals to ground cavities, and discovery of correlations between FIT connector frequency responses and “eye diagrams” (graphical representations of the quality of digital signals), among others.
The successful work yielded not just FIT’s gratitude but multiple conference awards for student Yulin He, who graduated this summer and is moving to a position at Apple. Feng singled out students Kewei Song, Zetai Liu and Haonan Wu for their high-speed testing setup contributions as well; Wu graduated this summer too, and is going to Broadcom. Jianming Jin, who is the Y. T. Lo Chair in ECE, and his group also contributed to the fast simulation work.
But the collaboration isn’t over. On top of over $2.75 million FIT previously invested in the 224Gbps work, it has just decided to provide over $1 million for a two-year high-speed testing program in Feng’s lab that will also study next-generation designs, which, at 448 gigabits per second, will have double the data rate of 224Gbps designs. “The whole industry is moving that direction,” Feng noted.
He praised C-NICE for providing a compelling “real problem” to solve, and reflected on the value of working with industry, which he said he enjoys. “I find success is coming to a mutual understanding of a problem,” he said. “A lot of people have a hard time to work with industry, because it’s not ‘freedom’... But my story is, I think we should patiently work to understand their problems, in order for us to really solve a real problem.”