Comparing performance of misbehavior detection based on Neural Networks and AIS

We compare two approaches for misbehavior detection in sensor wireless networks based on artificial immune systems (AIS) and neural networks (NN). We conclude that AIS and NN based misbehavior detection offers a decent detection performance at a very low computational cost. However both approaches are different regarding the length of the preprocessing phase, memory requirements, speed of computation and the rate of false positives. Both approaches are suitable for misbehavior detection in sensor networks, the decision which approach to choose for a specific sensor network depends on the details of the scenario.