Collaborative processing of signals recorded by distributed sensors brings about new opportunities as well as challenges to classical array signal processing. In this letter, we present a method for passive self-calibration of a distributed system in which each distributed node consists of an array of sensors. The proposed method estimates the positions and orientations of distributed sensor arrays, the synchronization timeline offsets as well as the positions of spatially distributed events emitted by an uncontrolled source or sources. The proposed two-step optimization consists in finding the maximum likelihood estimate of the relative geometry of a distributed system based on Directions of Arrival (DoAs) observed independently at each array. Next, the final positions of distributed arrays, the positions of acoustic events, and the synchronization offsets between the nodes are computed by solving a linear least square problem based on the observed Time Differences of Arrival (TDoAs) between the arrays and the relative positions estimated in the first optimization step. The evaluation results indicate that passive self-calibration of distributed sensor arrays can be conveniently achieved by observing acoustic events generated by an uncontrolled source based on the measured DoAs and TDoAs.
Products Included in this Bundle