Probability Estimation Based Adaptive Media Playout Algorithm

In recent years, with the rapid advances in internet and speech signal processing technology, the number of users increases rapidly which leads to the IP network streaming services become one of the biggest challenges on the internet. Streaming media services transmit media data to provide basic text, image information, as well as high-quality video. Its characteristics include timelivess, transmission effectiveness, convenience and inheritance and so on.Streaming media servers store the encoded and compressed videos and audios, so that client can watch the video program via receiving the streamed video, without having to download the entire multimedia files. Thus, the streaming technology can reduce the start delay. Streaming media technology mainly includes server technology, encoding technology, transmission technology, and channel adaptive technology.Adaptive media playback technology is one of the channel adaptive technologies for video streaming services. It allows clients to control the consumption rate of the buffered video frames.It triggers the actions of playout control, relying on the network channel conditions and the current client buffer fullness.An adaptive media playout algorithm based on probability estimation is proposed to reduce the video playout jitter and playout interuption caused by buffer underflow. It adjusts the duration of each frame according to the channel state and the estimated probability of underflow and overflow. Also it controls the playout rate range and the variation of the playout rate of the consecutive video frames to achieve the purpose of eliminating buffer outage and smoothing the video playout speed. Simulation results demonstrate that compared to other AMP schemes, the proposed algorithm improves visual quality by reducing the probability of buffer underflow, reducing the playout latencies as well as smoothing the visual perception.Since IP network is best-effort, available bandwidth may fluctuate frequently. We conceive an adaptive media playout regime based on underflow probability estimation, which requires no significant statistical knowledge of the previous teletraffic load. To achieve this, we present an underflow probability estimation model which is derived based on large deviation theory. We will then obtain the exact playback speed control mechanism, relying on which we directly adjust the frame duration, instead of using indirect methods based on a buffer fullness threshold or buffer fullness variation threshold. Our simulation results demonstrate an improved performance in comparison to other recent AMP algorithms.