Sensor nodes situated spatially close to each other tend to have similar behaviour. The neighbour-based detection technique is based on this principle and should provide means for anomaly intrusion detection in wireless sensor networks without prior training. Recently, this technique has been successfully applied to detect faulty sensor readings in wireless sensor networks [1, 2].

The thesis should include a description of selective forwarding, jamming, packet alternation and sinkhole attacks together with their symptoms. The author should consider whether these symptoms can be used in the neighbour-based intrusion detection scheme for the detection of corresponding attacks. The author should design and implement such an intrusion detection system (IDS) for the operating system TinyOS. The IDS should come in two modifications – one with local knowledge of immediate neighbours only and one involving information exchanged among one-hop or two-hops neighbours. The last part of the thesis is an effectiveness evaluation of the proposed IDS using the simulator TOSSIM. The author should provide results of the simulations, namely the number of false negatives and the number of false positives.