Applications of machine learning techniques in next-generation optical WDM networks

There has been an increase in demand of optical networks over the recent years due to which they have become sources of heterogeneous data. Several other challenges such as failure management, power optimization and low-margin optical netting have also been highlighted. However, with the use of machine learning (ML) techniques such support vector machine (SVM), k-NN (k-nearest neighbors), back-propagation, etc. most of these challenges have been overcome. The paper reviews the various optical network based applications in which ML-techniques have been used. Furthermore, the paper highlights the approach of using Wavelength Division Multiplexing (WDM) in optical networks as they provide huge bandwidth to the optical fibers. As a result, traffic can be transmitted simultaneously on several non-overlapping channels. This paper presents a comparative analysis of the existing approaches with the aim of attaining a future direction toward better outcomes.