Latency Aware Reliable Intrusion Detection System for Ensuring Network Security

Abstract

In network communication based applications, Intrusion detection system plays a vital role. In these applications, so as to collapse the system, the malevolent nodes may create enormous amount of traffic. In order to identify and prevent the intrusion activities, there are numerous researches that are presented by diverse researchers. LEoNIDS: architecture solves the energy-latency tradeoff by giving minimum power utilization as well as minimum detection latency all at once. However LEoNIDS architecture focus on the latency but not the precise detection of intrusion. This work resolves these issues by presenting the technique called Attack Feature based Fast and Accurate Intrusion Detection System (AF-FAIDS). This work presents a solution that enables intrusion detection in an effective way with enhanced delay and latency parameter. This also track and eliminate the intrusion attack existing in the system in an effective way with guaranteed energy saving. By using Machine learning, methods like Support Vector Machine, attack detection ratio is enhanced that would learn the attacks features in an effective way. Those features, which are taken in this presented technique, are Average Length of IP Flow, One-Way Connection Density, Incoming and Outgoing Ratio of IP Packets, Entropy of Protocols as well as Length in IP Flow. This technique provides the precise and quicker identification of intrusion attacks that will decrease the latency. The technical work is carried out in the NS2 environment under numerous performance measures and it is confirmed that the presented technique gives improved outcome when compared with the previous research methodologies.