Periodicity of Kalman-based scheduled filters

In this paper, the sensor scheduling problem for estimation is tackled, in which a set of sensors must share a common network to communicate their measurements. It is shown that, under some mild conditions, a Kalman-based scheduled filter results in a periodic selection of the sensors. In those cases, the computational cost and energy consumption can be reduced, as the sensors can await asleep between two consecutive measurements. A suboptimal one-step ahead strategy is thoroughly studied as a possible example of observation scheme.