Wireless sensor networks have recently received an increasing interest in many communities, way beyond the networking and signal processing areas. Wireless sensor networks provide a great infrastructure for many contemporary distributed applications. Many research groups are working for a few years on the design and implementations of distributed algorithms on such networks. More specifically, they are targeting data aggregation and how to make the aggregate traverse efficiently such a network However, most of the proposed approaches rely on a simplifying assumption: they assume that the network is composed of reliable and well-behaved nodes, all collaborating with each other towards a common goal. We believe this is a paradox as sensor networks are precisely very often deployed in hostile environments. The probability that some sensors are actually faulty is pretty high as well as the possibility of the presence of sensors exhibiting malicious behaviours. If the formers ones are sometimes considered, malicious behaviours are mostly ignored.
We believe that for the sake of the credibility of wireless sensor networks and their applicability to a wide range of applications, they should be able to support reliably a number of key functionalities, even in the presence of malicious sensors. Obviously such algorithms need to be themselves adapted to sensor networks and more specifically should take into account sensors’ reduced resources. The objective of this postdoctoral research is to consider the following functionalities in such a setting : (1) collection and aggregation of data observed in the sensor network and their routing to specific nodes ; (2) generation and management of alerts when some of the observed measures reach a given threshold. Obviously, identifying the potential attacks in such a network is an important step of this research. Expected outcomes should be both theoretical (algorithm design and proofs) and practical (simulation and implementation) in order to fully validate the proposed solutions .