A Prediction Scheme for VBR Video Traffic Using a Fast Orthogonal Search Algorithm

In this paper, we propose a fast autoregressive prediction scheme that takes into consideration the characteristics of variable-bit-rate (VBR) encoded video. The proposed scheme is based on the fast orthogonal search (FOS) algorithm and is characterized by its low computational complexity in addition to its accuracy in predicting the sizes of video frames. In addition, the proposed scheme handles the short-term and long-term dependence inherent in the sizes of the video frames without the need to assume any specific distribution. Simulation and numerical investigations are carried out to study the efficacy of the proposed predictors on video sequences with different spatial and temporal contents.