SDP-based adaptation for quality control in adaptive streaming

HTTP Streaming has been a new trend for video delivering via IP. Currently, most of the adaptation algorithms developed for HTTP Streaming are qualitative, which means the performance metrics could only be shown after the streaming session. In this paper, we embark on this problem by discretizing the whole system. We then formulate an infinite horizon problem (IHP) and solve it by Stochastic Dynamic Programming (SDP). To deal with the time-varying characteristics of video bitrate, we estimate the instant bitrate from the previous bitrate. We also develop mathematical models that predict the average performance of the adaptation policy, which helps adjust the settings before a streaming session. In the evaluation, we compare our method with a previous work which applies bitrate estimation and another method which only applies SDP.