K-nearest neighbor algorithm analysis for path determination in network simulation using software defined network

Abstract

Software defined network (SDN) is a new approach concept for planning, developing, and operating computer networks. Routing is the process of selecting paths on a network system to send or forward packets to the destination network. This study aims to obtain the results of the calculation analysis using the k-nearest neighbor (k-NN) method as the implementation of the recommendations. The results of the analysis use a mesh topology design approach with predicted values of throughput, jitter, delay, and packet loss. This value is used as a recommendation to the network manager to determine the best path. The best path selection from the analysis results is i) path-1 (very good) which includes switch 1, switch 2, and switch 4; ii) path-2 (good) to switch 1, switch 3 and switch 4 switch 4; iii) channel-3 (moderate) through switch 1, switch 2, switch 3 and switch 4; and iv) channel-4 (bad) through switch 1, switch 3, switch 2, and switch 4. While the calculation using the confusion matrix is accuracy=72.31%, precision=96.08% and recall=87.84%.