Software Defined Network (SDN) is a concept that separates all control logic from network devices to a logically centralized control system named controller. One of the network applications that can be applied to the controller is routing. Routing algorithms are used to generate paths, for example Dijkstra's algorithm which computes the least cost from one node to another on the network. The cost is an important factor in determining which path to choose. When the cost is a fixed value, for example distance, it may result in the shortest path but not the optimal path, because link conditions are not considered. With a programmable SDN, the cost of Dijkstra's algorithm can be developed by considering the link conditions and artificial intelligence can be applied to cost calculations. In this research, the cost are based on available bandwidth and packet loss by using fuzzy logic. The results of the test, the system can determine the cost of the link from fuzzy logic calculations based on the available bandwidth and packet loss and find the path for the packet. Based on the results of the delay test, the system outperforms Dijkstra's static algorithm. Then based on packet loss testing, the system has packet loss ranging from 0-1%, while Dijkstra's static algorithm has a significant increase in the number of clients 25 and above. However, on the results of the convergence time, the system has a longer time than Dijkstra's static algorithm.
Copyrights © 2022