Hybrid intrusion detection using machine learning for wireless sensor networks

This Wireless sensor network (WSN) is a network of sensors, which is capable of communicating with each other and sensing some changes in parameters such as temperature, humidity etc. Such networks are beneficial in many fields, such as military industries, health monitoring, environmental tracking, monitoring of traffic. However, WSN’s are easy to be attacked because of its properties such as untrusted broadcast transmission media, physical accessibility of sensors. So, protecting networks against attacks is one of most important issues in network and information security domain. As Sensor nodes have limited resources, authentication and encryption cannot be implemented directly to it. Hence, we propose a Hybrid Intrusion Detection System, which consists of Host Based Intrusion Detection system (HBIDS) and Network Intrusion Detection System (NIDS). In NIDS anomaly in network traffic, is detected. In HBIDS, patterns of misuse are detected from information collected at particular host or sensor. The main idea is to collect each sensor node’s data and anomaly is detected in network and this detected intrusion is compared with signatures of attack in misuse detection system.