Queue Management for the Heavy-Tailed Traffics

The purpose of this research is to design the new queueing algorithm effectively controlling the heavy-tailed traffics based on the comparison between the performance of the passive queue algorithm and that of the active queue algorithm. We adopted the tail-drop algorithm for PQM and the Random Early Detection (RED) for AQM, and conducted the experiments using ns-2 simulator. As the results, we extracted the following features. Firstly, the different $\alpha$ of the Pareto distribution made the completely different throughput performance. Secondly, the heavy-tailed traffic could improve the throughput performance for both queueing management systems even if the average of file size of distribution increases. Thirdly, RED could effectively reduce the overall load leading the throughput increase and provided the fairness in terms of the throughput. Finally, if the traffic pattern could be changed to the heavy-tailed distribution, the performance could improve.