Delay prediction in best effort type network and its application to media synchronization

点击翻译按钮获取中文摘要

When using one or more media streams in best effort type network, media synchronization is necessary to mitigate the effect of delay and jitter on the applications. From this point of view, this study aimed at constructing a new media synchronization method which is based on the short-term delay transition and its prediction. The major results are the followings.First, we measured the short time end-to-end delay on Internet and found that the exponential distribution with an offset fits the real network case very well. Then, by analyzing the shift of the parameters of this distribution minutely, we showed that it is not necessary to determine the transition time of the parameters is not necessary. Therefore, we can use this model easily when we simulate a certain network on computers.Second, we introduced phase plot method for end-to-end delay measurement. Although this method is rather accurate, it could not be directly applied to media synchronization since it requires some off-line calculation. So, we proposed an innovative method to use phase plot in real-time and also improved the performance. It was also applied to delay prediction and media synchronization.On the application side, we introduced a media synchronization for haptic collaboration systems. We first made a very intensive experiment which clarifies the effect of delay and synchronization on the subjective quality. The result was quite counter intuitive, which means that the precise synchronization is not always the best. In other words, when there are two or more haptic streams, it is sometimes desirable to activate the haptic interface as soon as possible without waiting the signals of other haptic streams. Considering it, we constructed a suitable media synchronization mechanism for haptic streams.