Adventures in Learning-Based Rate Control, Brighten Godfrey, UIUC

Abstract:
Rate control algorithms are central to the Internet, for transport layer congestion control, and application layer video delivery.  Rate control is also one of the most subtly difficult and persistent challenges in networking, having to infer good decisions at millisecond timescales in diverse, opaque environments using only a trickle of information.  In this talk, we'll discuss how this problem can be approached with an online learning-based architecture, where an explicit performance objective function guides a rate controller learning from real-time observations. After introducing the basic Performance-oriented Congestion Control framework, we'll discuss how one can plug in new objectives (for example, to build a "scavenger" transport) and new control algorithms (for example, deep reinforcement learning).  Finally, we'll discuss challenges in the approach and future directions including using ML agents as an "adversaries" to produce challenging environments for rate controllers.

Bio:
Brighten is an Associate Professor in the computer science at the University of Illinois at Urbana-Champaign. He co-founded and served as CTO of network verification pioneer Veriflow, through its 2019 acquisition by VMware where he now serves as a Technical Director. He received his Ph.D. at UC Berkeley in 2009. His research interests lie in the design of networked systems and algorithms. He is a winner of the ACM SIGCOMM Rising Star Award, the Sloan Research Fellowship, and the National Science Foundation CAREER Award.