Is AIS Based Misbehavior Detection Suitable for Wireless Sensor Networks?

Sensor networks are a flavor of ad hoc wireless networks with limited computational capabilities. The task to protect such networks against misbehavior is therefore more complicated as any detection mechanism has to be simple and efficient. We employed mechanisms based on artificial immune systems (AIS) in order to detect misbehavior. We conclude that AIS based misbehavior detection offers a decent detection performance at a very low computational cost. We show that misbehavior detection when applied at both the MAC and network layers may still not be sufficient, instead it is necessary to extend it to layers with end-to-end connection information; this would also allow for classifying misbehavior by its potential to cause harm. These results have a direct impact on the design of AIS for sensor networks and on engineering of sensor networks.