IP Network Anomaly Detection using Machine Learning
Roshan Nair, Chaithanya Pramodh Kasula, Sravanthi Vankayala, Niloy Chakraborty · 2019
The proliferation of network technologies and its associated threats have made it indispensable to develop different techniques to effectively detect network attacks. The present paper focuses on discovering viable anomalies that have a significant potential to be associated with abnormal network behavior. Three approaches based on ML (Machine Learning) have been proposed to detect suspicious network behavior. These methods are an extension of the techniques discussed as part of the introduction below. The developed approaches in the current paper are evaluated by testing their efficiency against a real-time network attack using available open-source network tools. The results of the experiment demonstrate successful identification of anomalous instances from the telemetry data with a low false alarm rate. Further, we believe that our approaches can be directly deployed in a real-time environment (independently on the edge device or over the cloud) to strengthen the network security.