EdgeRIC: Empowering Real-time Intelligent Optimization and Control in NextG Networks
Planning to Explore via Self-Supervised World Models
EdgeRIC: Empowering Real-time Intelligent Optimization and Control in NextG Networks

Woo Hyun Ko
whko@tamu.edu
Ushasi Ghosh
ughosh@ucsd.edu
Ujwal Dinesha
ujwald36@tamu.edu
Raini Wu
ryw003@ucsd.edu
Srinivas Shakkottai
sshakkot@tamu.edu
Dinesh Bharadia
dineshb@ucsd.edu
NSDI 2024


Radio Access Networks (RANs) are increasingly softwarized and accessible via data-collection and control interfaces. RAN intelligent control (RIC) is an approach to manage these interfaces at different timescales. In this paper, we introduce EdgeRIC, a real-time RIC co-located with the Distributed Unit (DU). It is decoupled from the RAN stack, and operates at the RAN timescale. EdgeRIC serves as the seat of real-time AI-in-the-loop for decision and control. It can access RAN and application-level information to execute AI-optimized and other policies in real-time (sub-millisecond). We demonstrate that EdgeRIC operates as if embedded within the RAN stack. We showcase RT applications called μApps over EdgeRIC that significantly outperforms a cloud-based near real-time RIC (> 15 ms latency) in terms of attained system throughput. Further, our over-the-air experiments with AI-based policies showcase their resilience to channel dynamics. Remarkably, these AI policies outperform model-based strategies by 5% to 25% in both system throughput and end user application-level benchmarks across diverse mobile scenarios.


Website Template Originally made by Phillip Isola and Richard Zhang for colorful ECCV project; the code can be found here.