Ad hoc sensor networks, constraint programming and distributed agreement

Numerous applications rest on sensor networks. Managing such networks in an efficient manner involves solving complex distributed computing problems, that problems are magnified in ad hoc sensor networks, where a limited knowledge of the topology is available. The two major issues that must be addressed simultaneously are, on the one hand, optimal sensor allocation and, on the other hand, tolerance to sensor and communication failures, which leads to solving dynamic assignment/reconfiguration problems. We describe a class of solutions that combines the problem solving capabilities of constraint programming techniques with the properties of distributed real-time fault-tolerant agreement. With such solutions, those safety, timeliness and dependability properties that define the management problems under consideration. The efficiency of these solutions are illustrated with real world examples.