Sensor Networks Protocol Design Using Random Markov Field Theory

In this paper, we present a new general framework allowing sensor networks design using random Markov fields (MRF) theory. We explain how the principles underlying MRF theory naturally fit design requirements in sensor networks in particular the need to rely on decentralised and distributed methods to solve global optimisation problems. We illustrate the potential of this new general concept with the practical problems of power control and resource allocation. In addition, we present a new method to improve the convergence rate of the simulated annealing when the sensor network design problem is a constrained optimisation problem.