Expressing validity estimates in smart sensor applications

Distributed control applications need reliable data from remote sensors. This paper presents a method for describing the validity of continuously valued sensor data that represents the confidence an application can put in remote information. Validity is dependent on the completeness of the sensor failure model and the quality of the respective checking mechanisms. Finally, the results of the various individual tests have to be combined into a single validity value. The paper describes how to derive an expressive notion of validity and gives an example how to exploit it in a Kalman filter for sensor fusion.