A sensor selection method considering communication delays

In connection with the method of selecting among sensors connected by a network those sensors that can efficiently acquire information on the object, this paper presents a discussion in a form that explicitly treats the existence of communications delay when the information is received via a network. First, we propose a calculation algorithm that is scalable with respect to the increase in the number of the sensors when the information from several sensors is combined by a Kalman filter and the communications delay is taken into account. When several sensors have different communications delays and these delays change so that the sensor information arrival is unrelated to the actual time, the information is correctly combined. Next, we propose a method for selecting an appropriate sensor when state estimation by this algorithm is performed. This method makes use of the cross-information previously used as a reference for sensor selection. However, the reduction of the information content due to communications delay, affecting the estimated value for the present time, can be expressed quantitatively. These algorithms are verified by numerical experiments and their effectiveness is demonstrated.