JISE


  [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]


Journal of Information Science and Engineering, Vol. 37 No. 6, pp. 1313-1326


Automation based Active Queue Management using Dynamic Genetic Algorithm in Real-Time Application


MAJID HAMID ALI+ AND SERKAN ÖZTÜRK
Computer Engineering Department
Erciyes University
Kayseri, 38039 Turkey
E-mail: hammajid80@gmail.com; majidalbayti80@yahoo.com; serkan@erciyes.edu.tr


Nowadays, there has been significant interest on designing coordinated automation systems for wide range of industrial applications. Automation brings the dynamic control in real time communication systems also. For real time communications, recently the TCP/IP gained the significant attentions in usage of best-effort networks. The focus is on assured the Quality of Service (QoS) while handling real-time communications through IP networks. QoS of such communications based various components of networking like the Active Queue Management (AQM). The AQM methods mainly designed to handle the network traffic efficiently so that no QoS degradations due to congestion in networks. In this research described novel AQM depend on Random Early Detection (RED). RED AQM method mainly designed to solve the network congestion problems in Internet routers. However RED does not support the automation according to traffic dynamics in network which may degrade the performance. This paper proposed automation-based RED using the Dynamic Genetic Algorithm (DGA) called DGARED to handle the congestion in TCP/IP networks. The DGARED based on dynamic tuning parameters of optimization technique GA which adjust the weight parameter dynamically enhance overall queue scale sensitivity at routers in order to update actual queue size dynamically in RED. Using the dynamic GA, we provided the technique to find the effective values for weight parameter, maximum threshold, and minimum threshold. The simulation results of DGARED are evaluated with existing RED and GARED algorithms in terms of throughput by considering the different network conditions. The results show that DGARED overcomes the problems of both methods.


Keywords: automation, dynamic tuning, genetic algorithm, active queue management, queue size, RED, throughput

  Retrieve PDF document (JISE_202106_06.pdf)