Real Time Network Intrusion Detection using Machine Learning Technique

Abstract

This paper reflects the work carried out in network security using distinctive machine-based learning techniques. In response the exponential increase in network space breaches and data leaks, the demand for a system that can detect anomalies and notify the system admin is imperative. Using packet sniffing modules, we capture the packets and then compare them to a pre-trained machine learning module trained on the NSL KDD dataset to detect ambiguous packets. By selecting the desired port, the IDS (Intrusion Detection System) sniff all incoming packets and categorizes them as anomalous if their behavior is not normal. On successful prediction, we present the user with a choice to act against the prescribed threat or ignore it as per the user's request. A detailed analysis report shall then be presented periodically to provide an overview of the overall health of the system on which our IDS system has been deployed.