OpenNetLab: An Open Platform for RL-based Congestion Control for Real-Time Communication

Опубликовано 6 декабря 2022, 21:21

Research talk
Byung-Gon Chun, Seoul National University

With the importance of real-time communications (RTC), much attention is being paid to designing congestion control (CC) algorithms for RTC to achieve high Quality of Experience. In this talk, we introduce a project on OpenNetLab that provides system-level support for training and testing Reinforcement Learning(RL)-based CC algorithms for RTC. OpenNetLab supports diverse environments ranging from simulators to real-world testbeds, in which one can design and experiment with her RL-based CC algorithms by exploring tradeoffs in the design space. Early use cases demonstrate that OpenNetLab facilitated coming up with new RL-based CC algorithms that outperform widely used rule-based baselines.

Learn more about the Next Generation Networking and its platform Workshop: microsoft.com/en-us/research/e...

This workshop was part of the Microsoft Research Summit 2022: microsoft.com/en-us/research/e...