Optimal Active Sensing for Process Tracking
Arpan Chattopadhyay, Urbashi Mitra · 2018
Motivated by the Internet-of-things and sensor networks for cyberphysical systems, the problem of low complexity dynamic sensor activation for the tracking of a time-varying process is examined. The tradeoff is between energy efficiency and fidelity. The problem of minimizing the time-averaged mean-squared error over infinite horizon is examined under a constraint on the mean number of active sensors. The proposed method artfully combines two key ingredients: Gibbs sampling for sensor subset selection, and stochastic approximation for learning, in order to create a high performance, energy efficient tracking mechanism with active sensor selection. Tracking of an i.i.d. process with unknown parametric distribution is considered; the main challenge here is that the unknown parameter vector must be learned. The key theoretical result proves that the proposed algorithm converges to locally optimal solutions. Numerical results suggest that global optimality is in fact achieved in some cases.