Probe or wait: Handling tail losses using multipath TCP

Packet losses are known to affect the performance of latency sensitive Internet applications such as media streaming and gaming. Transport protocols recover from packet loss in order to provide reliable end-to-end communication and improve user experience. The efficiency of loss recovery mechanisms influences the completion time of flows, and thus also the application performance as perceived by the end user. In this paper we focus on state-of-the-art loss recovery mechanisms for TCP and Multipath TCP. We use controlled tail loss scenarios to evaluate the performance of loss recovery mechanisms and, based on the observations, we propose an enhanced tail loss recovery mechanism for Multipath TCP, to improve the loss recovery time. Our experiment results, using the Linux Multipath TCP implementation, show consistent end-to-end latency performance improvement in considered scenarios.