Modeling video sources for real-time scheduling

What is the impact of autocorrelation of variable bit rate (VBR) video sources on real-time scheduling algorithms? Our results show that the impact of long term, or interframe, autocorrelation is negligible, while the impact of short term, or intraframe, autocorrelation can be significant. Such results are essentially independent of the video coding scheme employed. To derive these results, we introduce a model that is based on statistical analysis performed on actual video data. Our model accurately captures the distribution and the autocorrelation function of the source bit stream on both the frame and the slice level. We show that the original video data sequence can be modeled as a collection of stationary subsequences called scenes. Within a scene, a model is derived for both the sequence of frames and of slices. In previous work at the slice level, the pseudo-periodicity of the autocorrelation function made it difficult, to develop a simple yet accurate model. One of the new elements introduced in this work is that we present a generalization of previous methods, that can easily capture this pseudo-periodicity and is suited for modeling a greater variety of autocorrelation functions. The generality of our model lies in that, by simply tuning a few parameters, it is able to reproduce the statistical behavior of sources with different types and levels of correlation