Perbandingan Algoritma Breadth First Search dan Bellman-Ford Dalam Pemilihan Path Secara Dinamis Pada Software Defined Network Berbasis Ryu Controller

Abstract

The increasing complexity of modern networks requires dynamic path selection mechanisms that can adapt to changing network conditions. Software Defined Network (SDN) is a networking architecture that separates the control plane from the data plane, enabling centralized network management through a controller. This study aims to implement and compare the performance of the Breadth First Search (BFS) and Bellman-Ford algorithms in dynamic path selection within an SDN environment based on the Ryu Controller. BFS performs path selection based on the minimum hop count, whereas Bellman-Ford performs path selection based on the minimum path cost. The experiments were conducted using the Mininet network emulator with a partial mesh topology representing the East Java Backbone network. Performance evaluation was carried out under three network scenarios, namely normal links, heterogeneous links, and link failures. The evaluated parameters consisted of Quality of Service (QoS), including throughput, delay, packet loss, and convergence time. The results show that both algorithms were successfully implemented on the Ryu Controller and were able to recompute alternative paths when network conditions changed. In the normal link scenario, BFS and Bellman-Ford exhibited relatively similar performance because all links had identical characteristics. In the heterogeneous link scenario, Bellman-Ford achieved higher throughput by considering link costs during the path selection process, whereas BFS selected paths solely based on the minimum hop count. In the link failure scenario, both algorithms successfully maintained network connectivity by recomputing alternative paths after topology changes. However, BFS achieved lower convergence time due to its simpler path selection process. These findings indicate that BFS is more suitable for networks with homogeneous link characteristics because of its lower computational complexity and faster convergence, whereas Bellman-Ford is more appropriate for networks with heterogeneous link conditions because it considers link costs when selecting communication paths