Evaluation of an Intrusion Detection System for Routing Attacks in Wireless Self-organised Networks

Abstract

Wireless Sensor Networks (WSNs) are<br />becoming increasingly popular, and very useful in military<br />applications and environmental monitoring. However,<br />security is a major challenge for WSNs because they are<br />usually setup in unprotected environments. Our goal in this<br />study is to simulate an Intrusion Detection System (IDS)<br />that monitors the WSN and report intrusions accurately and<br />effectively. We have thus simulated an IDS that uses<br />anomaly-based technique to monitor traffic pattern on the<br />network following a fixed-width clustering algorithm. Our<br />simulation is based on the sensor network simulation<br />package by the Naval Research laboratory (NRL). To<br />evaluate the IDS, we simulated a sensor network,<br />investigated it with the presence of phenomenon, and<br />extended it to generate denial of service attacks. We have<br />used the phenomenon contribution to generate a realistic<br />traffic pattern for accurate evaluation of protocols, and<br />compared it to the traditional method of using only cbr<br />traffic, which is usually been employed by most<br />researchers. We further adapted the IDS into this simulated<br />network, and our results show that the selected IDS has<br />detection rate of over 90% with a very low false positive<br />rate of less than 1%. We obtained this by configuring every<br />node to independently monitor detect and report intrusions.