A Survey of VBR Video Traffic Models

We have seen a phenomenal growth in video applications in the past few years. An accurate traffic model of VBR video is necessary for performance evaluation of a network design and also for creating synthetic loads that can be used for benchmarking a network. In view of this, various models for VBR video traffic have been proposed in the literature. In this paper, we classify and survey these models. In addition, we implemented four representative video traffic models and compared them using the H.264 AVC video traces available at the Arizona State University video traces library. These models are: the Markov Modulated Gamma (MMG) model, the Discrete Autoregressive (DAR) model, the second order Autoregressive AR(2) model, and a wavelet-based model. The results show that the MMG and the wavelet-based models are suitable for both video conference and IPTV, while the DAR model is good for video conference traffic only. According to our results, the AR(2) model is not suitable for generating any type of H.264 video. A brief overview of SVC, HD, and 3D video is also provided.