Tile-Level Multipath Content-aware Live Video Streaming

The growing popularity of live video streaming applications even in remote areas underscores the need for a high-quality and seamless experience. However, live video streaming is a bandwidth-hungry application that challenges the current cellular networks to maintain consistent bandwidth and latency. To address these shortcomings, we propose using the collaboration of multiple devices with their individual cellular networks to support live video streaming. Prior approaches on multipath streaming, however, lead to high end-to-end latencies due to head-of-line blocking when used with live video streaming. To mitigate this problem, we design a content-aware system Compact that splits the video into foreground and background using video tiles (independently encoded spatial blocks) and streams them over different paths. Compact depends on its scheduler, which exhaustively searches for the best quality based on network estimates. We implement and validate Compact on two transport protocols: SCTP and QUIC. We extensively evaluate our system using network traces while walking and traveling in a bus or car for three live streaming applications: online classes, live sports, and live news. For the online class, we find that compared to the single path, Compact manages to reduce the median stall and E2E lag by \(70.6\%\) and \(28.57\%\) , and the tail stall and lag by \(83.9\%\) and \(\approx 80\%\) on a bus trace. We also observe that Compact performs better than the baselines, for live news and live sports as well. We further performed a live experiment to test Compact on an actual cellular network, and show that it works well in practice.