Optimal operation of sensor systems using transputers

In the framework of the sensor platform optimization problem, the case of a dynamic system subject to random disturbances and observed by a single sensor, that can or cannot be connected with renewable finite energy and whose output is also subject to disturbances is analyzed in this paper. The resulting model describing the system and sensor dynamic are stochastic differential equations. A sensor operation regime that minimizes a cost function, which is a function of the estimation error is looked for. In order to minimize the estimation error, the state vector estimation given by the Kalman filter, that minimizes the error variance is used; therefore the evolution of the covariance associated with that filter is an ordinary differential equation. In this way the optimization can be solved with dynamic programming techniques, solving an appropriate system of quasi-variational inequations of Bellman type. This system is solved numerically using finite element approximations and computational parallelization techniques implemented in a transputer network. Finally, the simulation of the system operation under a suboptimal policy found with this methodology is shown