Real-time quality evaluation of adaptation strategies in VoD streaming

Abstract

HTTP Adaptive Streaming (HAS) has become a popular trend for multimedia delivery nowadays. Because of throughput variations, video adaptation methods are needed to avoid buffer underflows. In this context, it is also important to evaluate the overall video quality of a session. In this paper, we investigate a quality model that can evaluate a session quality as well as different adaptation strategies in real time. We use the histogram of segment quality values and the histogram of quality gradients in a session to model the overall video quality. Then, our quality model is employed to evaluate, for the first time, the cumulative quality of typical adaptation methods in real time. It is found that, to provide a high quality level, the client should avoid changing versions frequently and drastically.