Detection and Classification of Anomalies in IP Communications Networks

Abstract

Communication systems and networks are constantly threatened, so the detection of adverse security events is an important step in maintaining vital security services, such as their confidentiality, integrity and availability. Intrusion detection systems are the most widely used line of defense in information and communication technology for monitoring and detecting security events. This article aims to  identify and classify the anomalies encountered in network traffic by using different machine learning methods and comparing their performances. Results show that random forests are able to classify attack types with high accuracy of about 99%, while support vector machine have a marginal poorer performance at this task.