Enforcing and enhancing consensus of spatially distributed filters utilizing mobile sensor networks

This paper considers the learning abilities of spatial filters utilizing mobile sensing devices. The spatial filters monitor and estimate the state of a spatially distributed process. Going beyond the use of mobile versus static sensing devices to enhance the learning properties of these filters, an agreement of the different spatial filters is imposed. This takes the form of an additive penalty term in each of the dynamic equations of the filters and whose goal is to minimize the agreement between them. The proposed consensus penalty term results in all the spatial filters agreeing with their static average and an enhanced convergence of each of the associated state estimation errors to zero. Numerical simulations for a 1D PDE are included to demonstrate the effectiveness of a such a consensus mobile sensor network in improving the system´s learning performance.