QoE-Driven Adaptive K-Push for HTTP/2 Live Streaming
Zhimin Xu,Xinggong Zhang,Zongming Guo
DOI: https://doi.org/10.1109/tcsvt.2018.2849015
IF: 5.859
2019-01-01
IEEE Transactions on Circuits and Systems for Video Technology
Abstract:Dynamic adaptive streaming (DAS) over HTTP has been widely deployed over the Internet. However, due to the pull-based nature of HTTP/1.1, there exists intolerable streaming latency and high request overhead in the current DAS systems. With dynamic k-push, HTTP/2 live streaming promises to achieve low live latency with less overhead and small segment duration. In this paper, we propose a quality of experience (QoE) driven adaptive k-push mechanism (QK-Push) for HTTP/2 live streaming. The client just sends one request to set push length (K) and bitrate (v) parameters and the server would push back K segments in a batch. To determine k-push parameters, a probabilistic buffer model is first designed to avoid buffer underflow/overflow. Also, three QoE objective functions are designed to ensure the high streaming quality (bitrate), playback continuity, and smoothness. QK-Push casts this multi-objective optimization problem as a Pareto optimal problem. To solve it, a Nash bargaining solution is designed to balance the needs for video quality, bitrate smoothness, and request overhead. Finally, the segments in each push cycle are selected by solving the Nash problem with a discrete space Lagrangian method. We implement an HTTP/2 live streaming prototype system, with the QK-Push algorithm over modified dash.js and media presentation description. To evaluate the performances, the extensive live streaming experiments are carried out over a controllable network test bed and real Internet trace. The results demonstrate that the proposed QK-Push algorithm is able to improve the average bitrate up to 13%, reduce the bitrate oscillations up to 81%, decrease the startup delay up to 58%, and increase the estimate the mean opinion score up to 12% compared to the current HTTP/1.1 system.