LearningCC: An online learning approach for congestion control

It is noticed that traditional loss based algorithms are not moving data very well in today's Internet. To get bottleneck bandwidth fully utilized, these will keep increasing the congestion window and maintain a standing queue of packets routers, which cause large end to latency. Such long latency quite annoying ever growing delay sensitive applications. Recently, much effort has been devoted by researchers from both academia industry design better control protocols make In this article, we present an online learning approach named LearningCC for control. Instead adjusting with fixed rule as do, there several options endpoint choose. Each option mapped arm bandit machine. The exploration exploitation scheme used guide selection on update rule. Through trial error, improves its performance choosing action reward. Experiments conducted ns3 platform verify effectiveness LearningCC. Results indicate it achieves lower transmission than algorithms. Especially, found makes significant throughput improvement link suffering random loss.