Comparative Analysis of Community Detection Methods for Link Failure Recovery in Software Defined Networks

Abstract

The complexity of IP networks leads toward the minimum utilization of network resources. To address this problem the concept of SDN (Software Defined Network) has been introduced. SDN is a revolutionary networking paradigm that overcomes the limits of standard IP networks while also modernizing network infrastructures. SDN makes the IP networks into programable networks and upgrade the network infrastructure. Like traditional IP networks, SDN technology can experience network failures. Several research papers have investigated this issue utilizing several methods. One technique in SDN is to employ community detection methods for link failure recovery. Although a variety of comparing analyses have been given across community detection approaches, however, they have not considered the special comparative analysis for link failure recovery situations in SDN. This paper presents a comparative analysis of the most likely used community detection methods based on the Dijkstra algorithm for link failure recovery in SDN. Extensive simulations are performed to evaluate the performance of the community detection methods. The simulation results depict that the Infomap and Louvain community detection methods perform better and have more modularity by 0.12% and less average end-to-end latency by 27%, avg data packet loss by 0.8% than the Girvan and Newman community detection methods.