Uncertainty Aware Sensor Fusion and Dynamic Calibration in Sensor Networks
Sensor fusion refers to the combining of data from sensors or those derived from sensor data such that the resulting information is in some sense better than would be possible when these sources were used individually.
In addition to measuring quantities that aren’t accessible by conventional means, combining information from multiple sensors can both improve the quality and reliability of data as well as increase the coverage area of a measured quantity. Although research on the incorporation of metrological methods to sensor fusion, particularly regarding sensor network applications is currently limited, there has been progress with respect to uncertainty-aware sensor fusion for detecting faulty or drifting sensors.
Traceability of utmost importance
Given their widespread usage, consensus filtering and Kalman filters were identified as key focus areas for further investigation in the project. The developed methods must be applied in a trustworthy manner and ensuring the uncertainty awareness and hence the traceability of the methods will be of utmost importance. As such methods will almost certainly involve the use of time-varying quantities, the uncertainty awareness must also account for the dynamic nature of the system.
Understanding transfer behaviour
The incorporation of dynamic uncertainty propagation in sensor networks requires a better understanding of the transfer behaviour of sensors among users including a basic understanding of the mathematical and statistical signal processing principles involved. The potential applications of sensor fusion and dynamic calibration to sensor network use cases are varied and will play a prominent role in the upcoming project activities of FunSNM.