A Learning Automata Approach for Load Balancing in Software-Defined Networks

International Journal of Smart Electrical Engineering

مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 15 دی 1400  

نوع مقاله: Research Paper

شناسه دیجیتال (DOI): 10.30495/ijsee.2022.1944225.1158

نویسندگان

Mohammad Amin Irandoost ؛ Mona Salehi

Department of Computer Engineering, Hamadan Branch, Islamic Azad University, Hamadan, Iran

چکیده

Since Software-Defined Network is a logically centralized technology, the importance of scalability of the control plane has increased with increasing network scales. Therefore, the use of multiple controllers was proposed instead of a centralized controller. Although multiple controllers have solved scalability in software-based networks, they faced load imbalance on controllers due to the static assignment between the controllers and the switches. As a result, switch migration is proposed as an efficient approach to solving static allocation between the controller and switch. Switch migration allows a dynamic connection between controllers and switches, but which controller or switch is suitable for migration has become a vital problem issue in itself. A learning automaton with a variable structure is proposed to select the target controller in the proposed method. All selection and environment reaction cases are evaluated with learning automata, and choose the best controller for migration costs. The proposed method has been compared with state-of-the-art algorithms. The results showed that the proposed approach could reduce the delays of sending packets in the network by balancing the controllers with the optimal selection of target controllers for switch migration.

کلیدواژه‌ها

Software-Defined Networks؛ Load Balancing؛ Switch Migration؛ Learning Automata