Peak time bandwidth control algorithms with fitted traffic model on broadband network YouTube video traffic / Aini Azmi

This paper presents an analysis of video traffic and fitted to best distribution traffic model to control bandwidth usage in a broadband network. The study scope comprised of collections of inbound YouTube video traffic for 7 days with the time interval of each day is 3 hours. The broadband network is supported at 1 OGbps line speed to Wide Area Network (WAN). The objective of this research is to characterize YouTube video traffic on broadband network, to fit the original traffic to best traffic model and Bandwidth control algorithm is developed based on the real live traffic and using the fitted traffic model. Results presents traffic characterizations are identified based on two (2) parameters Cumulative Distribution Functions (CDF) traffic model. Maximum Likelihood Estimator (MLE) technique is used to fit the best distribution model. Four best traffic model is identified which are Extreme Value, Weibull, Normal and Rician traffic model. Among the four, Weibull shown as the best fitted model that presents value of MLE=-1178.4 with the Scale a=9.49411e+08 and Shape P=2.81324. Bandwidth Control Algorithms is developed based on Peak Time of day and night. Performance shows the bandwidth controlled as bandwidth save, processing time, bucket capacity and cost. Research benefits in the development of design network especially for bandwidth used on Video used in a network.