Power aware video traffic classification in the compression domain

Abstract

We propose a power aware video traffic classification scheme in the compression domain. A Bayesian classifier and a nearest neighbor classifier (NNC) for MPEG variable bit rate (VBR) video traffic are proposed based on the I/P/B frame sizes only; they can reduce power consumption vastly. Our simulation results show that: 1) MPEG video traffic can be classified based on the I/P/B frame sizes only using the Bayesian classifier or the nearest neighbor classifier, and both classifiers can achieve quite low false alarm rate; 2) the nearest neighbor classifier performs better than the Bayesian classifier which sounds ridiculous because the Bayesian classifier is recognized as the optimal classifier. The reason is because the recognized lognormal distribution is not a good approximation for I/P/B frame sizes. We have based the Bayesian classifier on the lognormal distribution model, but the nearest neighbor classifier is model free, so it can perform better than the Bayesian classifier.