PB-IDS: High Accuracy and Reduced False Reports through Performance Based IDS in MANET

Abstract

Mobile ad-hoc network (MANET) is infrastructure less wireless network with dynamic topology supports for handling the nodes mobility and authenticity. As the network the nodes are continuously entering and leaving the network, detecting the nodes behaviour is a very complicated task. The reliability over such network gets reduced if some unauthorized malicious actions are performed in the network which causes the drops in normal working. The MANET is always susceptible to the malicious attackers gives a rise to intrusion detection systems (IDS). The aim of IDS is to effectively analyse the unwanted access or drops in the networks which are intentionally planted by some of the misbehaving nodes known as attacker. During the last few years various techniques are developed for achieving the goals of early and complete detection of intruder’s node. But there are some of the practical realities by which a complete solution is still need to be developed. Some of factors which play a vital role in handling the intruder’s node and detecting them effectively are: mobility, routing behaviour, performance monitoring, reduced collision, drops, overheads etc. There are so many intrusion preventions and detection techniques had worked on above factors and gives effective results. This work mainly carries the previous direction of IDS improvements based on their performance factors monitoring and focuses towards achieving the complete solution. It mainly reduces the false detection rates and timely founds the ambiguous collisions. The approach is named as performance based intrusion detection system (PB-IDS). The approach uses AODV protocol for embedding the suggested PB-IDS code for improved, robust and scalable detection of intrusions. At the analytical level of evaluation, the approach is proving its strong presence to their parallel tools like Watchdog, EAACK, and Pathrater etc. Also it is observed that the malicious behaviour detection ratio is high and the false detection is very low with optimal overhead.