Recently, HTTP-based adaptive streaming has become the de facto standard forvideo streaming over the Internet. It allows clients to dynamically adapt mediacharacteristics to network conditions in order to ensure a high quality ofexperience, that is, minimize playback interruptions, while maximizing videoquality at a reasonable level of quality changes. In the case of livestreaming, this task becomes particularly challenging due to the latencyconstraints. The challenge further increases if a client uses a wirelessnetwork, where the throughput is subject to considerable fluctuations.Consequently, live streams often exhibit latencies of up to 30 seconds. In thepresent work, we introduce an adaptation algorithm for HTTP-based livestreaming called LOLYPOP (Low-Latency Prediction-Based Adaptation) that isdesigned to operate with a transport latency of few seconds. To reach thisgoal, LOLYPOP leverages TCP throughput predictions on multiple time scales,from 1 to 10 seconds, along with an estimate of the prediction errordistribution. In addition to satisfying the latency constraint, the algorithmheuristically maximizes the quality of experience by maximizing the averagevideo quality as a function of the number of skipped segments and qualitytransitions. In order to select an efficient prediction method, we studied theperformance of several time series prediction methods in IEEE 802.11 wirelessaccess networks. We evaluated LOLYPOP under a large set of experimentalconditions limiting the transport latency to 3 seconds, against astate-of-the-art adaptation algorithm from the literature, called FESTIVE. Weobserved that the average video quality is by up to a factor of 3 higher thanwith FESTIVE. We also observed that LOLYPOP is able to reach a broader regionin the quality of experience space, and thus it is better adjustable to theuser profile or service provider requirements.
展开▼
