Analysis of Calibration Performance of Networked Sensors Using Measurements Graphs

In this article, we analyze the relationship between the accuracy of calibration of networked sensors and the structure of the network. Networked sensors can be calibrated by maximizing a likelihood of given measured data. Though, the calibrated values can contain indeterminacies and have some degrees of freedom. For removing this indeterminacies, we need some standards. Incorporating one sensor that is calibrated in advance into the networked sensors, then, we can uniquely calibrate them. If the incorporated sensor and another sensor measure a same object, then we can take useful information for calibrating the latter sensor by comparing the two measurements of the object. The information is, though, incomplete because of measurement noise. In this article, the relationships between the accuracy of calibration and the structure of the network of sensors are analyzed. For this analysis, we describe the structure with a measurement graph