High-Definition MPEG-4 AVC Traffic Analysis and Bandwidth Prediction

Abstract

In this paper, we propose and evaluate the accuracy of two new MPEG-4 AVC video traffic models. The first predicts the size of the B-frames of MPEG-4 AVC compressed video. The second predicts the size of the group of pictures (GOP) of the MPEG-4 AVC compressed video. We also propose an enhancement to these models using a new scene change detector metric. These traffic models can be employed by a network scheduler to provision MPEG-4 AVC video bandwidth efficiently.