A Correlative Method to Rank Sensors with Information Reliability: Interval-Valued Numbers Case

Multisensor systems are increasingly used to obtain information when solving problems of investigating incidents and incidents, risk assessment, etc. However, as the complexity of the tasks being solved grows, the amount of data grows too, what leads to the well-known Big Data problem. One of the ways to overcome existing difficulties is to expand the functionality of sensors, but this requires knowledge of the information reliability of sensors. The paper proposes a method to rank sensors via information reliability criterion, which is based on the idea that the information reliability of any sensor can be calculated based on estimates of the proximity of its data to the data of other sensors. The implementation of this idea allows to solve the problem of assessing the information reliability of the sensors of a multisensor system in the absence of information about the class affiliations of test objects. To assess the information reliability of the sensors, it is proposed to use the correlation approach, at that the case is considered when the data has the form of interval-valued numbers. The developed method makes it possible to determine the information reliability of the sensors in an ordinal scale. To facilitate the practical application of the developed method, a numerical example is given.