期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：20
出版社：Journal of Theoretical and Applied
摘要：This paper proposes an efficient modeling of the Dynamic (DGRED) algorithm using a three state Markov Modulated Bernoulli arrival process (MMBP-3) for primary period congestion detection at the router buffer. The purpose of using Markov is two folds, the first is to implement DGRED with multiple traffic classes, where in each class may have different priorities and the second is to enhance the mechanism of DGRED in stabilizing the average queue length (aql) between the allocated threshold values of minthreshold and maxthreshold, using variable calculating parameters, which is stored in the utilized states of Markov. The (MMBP-3) is employed to replace the conservative and widely used Bernoulli process (BP) in assessing DGRED method. Accordingly, a three dimensional discrete time Markov chain is provided to implement DGRED algorithm for three traffic classes where each dimension corresponds to a traffic class with its own parameters. The (MMBP-3) is deal with correlation and burstiness in the network traffic. The developed algorithm allows for faster response to the changes in the network, congestion-to-non-congestion and non-congestion-to-congestion, which lead to decrease packet losses and improvement of network performance. The proposed method is evaluated in comparison to the DGRED and other AQM methods, the results reveal that the proposed algorithm provides better performance compared to DGRED, RED, GRED, and Adaptive GRED in relations of delay packet loss, delay and mean queue length. The most appropriate arrival process for DGRED method is IBP for D, mql, PL, and DP while congestion occurs. Though, this process cannot discourse correlation. Thus MMBP-3 is the greatest choice as it detects both correlation and bursty things.
