Adaptive rate control for Internet video streaming

Abstract

The main contribution of this paper is an end-to-end rate-based congestion control algorithm for unicast quality adaptive video streaming that we call Adaptive Rate Control (ARC). The algorithm is based on end-to-end estimation of both available bandwidth and queue backlog. ARC has been compared with the TCP-Friendly Rate Control (TFRC) algorithm via the ns-2 simulator. In particular, both single bottleneck and multi-hop scenarios with and without lossy links have been considered. Simulation results have shown that ARC: (1) is more friendly than TFRC towards Reno; (2) remarkably improves the good-put with respect to TFRC and Reno TCP in the presence of lossy links; (3) provides no rate oscillations in the presence of stationary network load; (4) exhibits a less oscillating rate dynamics with respect to Reno TCP.