A Google Congestion Control Algorithm for Real-Time Communication

Abstract

This document describes two methods of congestion control when using real-time communications on the World Wide Web (RTCWEB): one sender-based and one receiver-based. It is published as an input document to the RMCAT working group on congestion control for media streams. The proposed approach combines receiver-side arrival-time filtering, over-use detection, and remote rate control with sender-side control based on round-trip time, packet loss, and receiver-estimated available bandwidth. The receiver-side algorithm uses RTP timestamp information, RTCP feedback, TMMBR or REMB messages, and adaptive filtering to estimate network parameters and detect bandwidth over-use or under-use. The sender-side algorithm adjusts the sending bitrate using packet loss thresholds, receiver-side bandwidth estimates, and TCP Friendly Rate Control constraints. The draft discusses interoperability scenarios, implementation experience in the open-source WebRTC project, further work, IANA considerations, and security considerations.