A Bayesian Theory of Cooperative Calibration and Synchronization in Sensor Networks
Shigeru Ando, Nobutaka Ono · 2006
This paper proposes a method for calibrating networked sensors from local duplicated measurements in couples of sensors among them. In the Bayesian framework, we show the system and procedure for the optimum collaborative calibration/synchronization is composed by: 1) sensor-wise maintenance of the offset values and a corresponding column of inverse of the estimation error covariance matrix (confidence matrix), 2) incremental updates of the confidence matrix elements for the coupled sensors with the confidence of the duplicated measurement, and 3-1) centralized computation for inverting the confidence matrix into the estimation error covariance matrix and sensor-wise computation for obtaining updated estimates of the offset values, or 3-2) global iterative computation of the updated estimates among all the sensors.