A Counterexample in Distributed Optimal Sensing and Control
Serdar Yüksel, Sekhar Tatikonda · IEEE Transactions on Automatic Control · 2009
This note exhibits that in a distributed multi-sensor, single-controller scenario, for the minimization of a quadratic cost function, linear sensing policies over Gaussian channels are, in general, not optimal. This is in contrast with the corresponding single-sensor problem, which does admit an optimal linear solution.