Fiber-optic networks have formed the backbone of the internet for decades, but today’s explosive bandwidth demands are pushing core networks toward their operational limits, requiring network operators to extract greater performance, efficiency, and resilience from already-deployed infrastructure. The training of artificial intelligence (AI) models, as well as virtual reality and gaming technologies, has placed tremendous strain on these networks in recent years.
While machine learning and broader AI techniques have recently shown promise in improving network efficiency and adaptability, the field lacks a systematic framework for the deep integration of AI into optical network control, management, and failure recovery.
To address these problems head-on, Electrical and Computer Engineering (ECE) Professor and GW Vice Provost for Graduate and Postdoctoral Affairs Suresh Subramaniam was recently awarded a three-year, $600,000 grant from the National Science Foundation (NSF) to develop an AI-integrated architecture for next-generation optical networks, together with AI-driven algorithms for resource allocation, performance optimization, and failure management.
“The way these networks have been provisioned and operated has been largely without machine learning or AI modeling. It’s largely been based on rules. For example, if internet traffic is a certain amount, the network bandwidth will expand to a certain amount,” Subramaniam explained.
“We’re trying to take it to the next era, so the decision-making is automated, and you can predict where bandwidth will be needed. Instead of waiting for the traffic to show up and then resource appropriately, we will be able to predict and allocate bandwidth ahead of time to minimize delays,” he said.
The project will be supported by GW ECE graduate students, who will develop models, write simulations, and analyze and interpret data alongside Subramaniam.
“Optical networking has been my main area of research for the last 30 years, and I’ve been fortunate enough to be funded by NSF for this work without interruption since 1999. I look forward to continuing this effort and helping to build a foundation for more efficient integration of AI into optical networks,” he concluded.