EdgeRIC: Empowering AI-based Real Time Intelligent Control, Optimization and Security in NextG Networks
# 213
Abstract
In the rapidly evolving landscape of robotics, AR/VR/XR, automotive perception, and machine learning applications, traditional approaches to network optimization focused solely on QoS optimization cannot deliver diverse application requirements. Furthermore, optimizing QoE is crucial to delivering an enhanced user experience. Such requirements are often impossible to consider during standardization and are incredibly challenging to optimize. In this talk, I will first present the limitations of existing RAN intelligent controllers (RICs) in adapting to highly mobile wireless channels, which restricts their ability to meet the on-demand needs of applications. I will next introduce EdgeRIC—a real-time RAN intelligent controller that leverages the power of AI, specifically Reinforcement Learning, to elevate the performance of ORAN (Open RAN) stacks and address the diverse requirements of various applications in real-time. Decoupled from the RAN stack, EdgeRIC functions as an intelligent controller that employs AI-powered optimization techniques to provide control decisions to the RAN across multiple layers. Next, to train these AI models, we have developed a digital twin that ensures the spatial and temporal consistency of the wireless channel. Our deployment showcases the integration of EdgeRIC with an open-source ORAN stack, highlighting the remarkable over-the-air performance improvements achieved. With an application-aware intelligent scheduling policy, we present compelling results demonstrating a substantial 90 percent reduction in video streaming stalls. In addition to this, we would present MIMO apps that secure connectivity and cancel unwanted interference in challenging urban environments. Finally, I briefly cover several activities in my group -- from sensing to communication.
Dinesh Bharadia has been an associate professor in the ECE department at the University of California San Diego since July 2022, where he directs the WCSNG group. He received early promotion to a tenured professorship and held Assistant Professorship for four brief years from 2018–2022. He received his Ph.D. from Stanford University in 2016 and was a Postdoctoral Associate at MIT. Specifically, he built a prototype of a radio that invalidated a long-held assumption in wireless that radios cannot transmit and receive simultaneously on the same frequency, which inspired research on this topic from different communities (communication theory to RFIC). From 2013 to 2015, he worked to commercialize his research on full-duplex radios, building a product that underwent successful field trials at Tier 1 network providers worldwide like Deutsche Telekom and SK Telecom. He serves as a technical advisor for multiple startups. Dinesh was named to Forbes 30 under 30 for the science category worldwide list in recognition of his work. Dinesh was named a Marconi Young Scholar for outstanding wireless research and was awarded the Michael Dukakis Leadership Award. MIT Technology Review also named him among the top 35 Innovators under 35 worldwide in 2016. At UC San Diego, his group WCSNG designs and prototypes systems for Wireless Communication, Computing, Sensing, Networking, and sensor design with applications to privacy, security, robotics, health, and everyday life. Much of the group's research has inspired new research areas for border communities: communication theory, circuits, RFIC, and robotics. Much of his research has been translated into startups and commercial products (Haila, Kumu Networks, Totemic Labs).