Adaptive policing and shaping algorithms on inbound traffic using generalized Pareto distribution / Nor Azura Ayop

This paper present an analysis of inbound internet traffic and development of Adaptive Policing and Shaping Algorithms on inbound internet traffic and fitted to traffic model. Network "bursting" is a normal event in internet traffic engineering, but the bustiness of throughput data will violent the committed rate that provided by ISP. The objective of this research is to characterize inbound internet traffic collected on real live IP-based campus network, to develop Adaptive Policing and Shaping Algorithms with percentage level on Inbound Traffic based on traffic characterization and to compare the policing and shaping performance on bandwidth used, processing time and packet loss. Then, traffic is fitted to best traffic model and percentage level Policing and Shaping algorithm is developed to control the bandwidth used. The research scope is based on collected of internet traffic on IP-based network real live traffic at 16 Mbps speed line. By using MATLAB software, the Open Distribution Fitting application is fitted to the collected data to identifying the best distribution and the results presents GPD shows the highest value for best fitted traffic model. Log likelihood estimation technique is used to fit the best 2-parameter CDF compared to Weibull, Normal and Rician distribution model. The percentage level 5% under original bandwidth used is developed on policing and shaping algorithms to control bandwidth used. Result present performances upgraded around 3% of time processing and approximately 73% of bandwidth saved. This result help to expand the view of new idea in modelling the tele-traffic algorithm based on bandwidth management and time processing improvement. The most important matter is the understanding about the internet traffic's flow and characteristic.