Impact of Misbehaviour and its Detection in Ad-hoc Wireless Sensor Networks using AIS

Characteristic for ad-hoc wireless sensor networks is the lack of a centralized control. Instead each node serves as a routing device, which is able to forward packets to its neighbors1 and receive packets from them. Node movement is allowed, but rare for sensor nodes. Each node is able to be switched on and off at any time depending on its power saving configuration. Such networks are extremely vulnerable to user misbehavior. Since nodes within an ad-hoc sensor network are expected to have limited computational power and be battery powered, a system that is going to protect them has to be lightweight. It has to be adaptive too, as ad-hoc sensor networks are expected to operate autonomously with only spare maintenance. Classical intrusion detection approaches which are based on intrusion signatures therefore do not apply for such a task. Artificial immune systems (AIS) on the other hand seem capable of handling such demands. AIS are inspired by the Human immune system (HIS) using selected features of this defense system. The basic feature of an HIS is the ability to discriminate between self and non-self. In case of sensor networks non-self is any kind of user behavior that impacts the network in a negative way. The purpose of our simulation based experiments is to show that AIS are a suitable approach for detecting misbehavior in ad hoc sensor networks.