Performance Analysis of Optical Mesh Network using Machine Learning Approach

Recent developments in devices and automations have created huge demand for network resources. Traditional networks are becoming less effective in handling a large number of traffic generated by new technologies and requirements, so the data center needs a well-tested automatic monitoring system. Such automation can be provided by the machine learning-based architectures. In this paper, we propose the implementation of machine learning algorithms in the optical
$16 \times 16$
mesh network (intra datacenter optical network) and the performance is monitored continuously to detect the network's efficiency. The failure in the network can be immediately addressed by continuous monitoring of data transmission parameters, which helps to maintain the network's Quality of Service. An Optical Mesh network is designed using a discrete event simulator, and its performance is analyzed using various machine learning algorithms. The implementation of ANN algorithm for network performance analysis has achieved an accuracy of 96% in predicting the testing values, when compared with other ML algorithms such as SVM, K-Nearest Neighbor, Decision Tree, and Naive Bayes.