Asian Journal of Pure and Applied Mathematics

Over-the- top (OTT) applications are known for their large consumption of network resources. Trailing to this, network operators need to efficiently classify network traffic according to the application that generated them. The introduction of new OTT applications on the internet are on the increase, and has made the traditional methods such as simple port-based, and analyses of packet payloads of networks inefficient. Therefore, network operators are beginning to explore machine learning methods to analyse the trend of communication of OTT applications in order to be able to classify them and to deploy application and user specific degradation policy to manage the network efficiently. This becomes imperative for the network operators to effectively manage network security monitoring, congestion avoidance, Internet Protocol (IP) management, Quality of Service (QoS) enforcement, bandwidth management, and estimation of bills for usage. This study makes use of two deep learning algorithms to develop OTT applications traffic analysis models for network. The deep learning algorithms used are Multilayer perceptron, and Long Short-Term Memory. Both algorithms gave feasible results in that LSTM have a precision of 90% recall of 91%, F1-score of 91.21% and accuracy of 93.5 while MLP have a precision of 99%, recall of 99%, F1-score of 99% and Accuracy of 98. It was therefore concluded from this study that the MLP model outperforms the LSTM model in carrying out Over-the-top applications traffic analysis on network.