Multisensor multitarget distributed fusion for discrepant fields of view.

In a distributed sensor network, how to synthesize the target information and estimate the trajectories under the limitations for both the fields of view (FoVs) and the information channels is a challenging problem. Due to the difference between the FoVs of the sensors, the local multitarget densities can be different. The classical RFS-based fusion methods may discard the information that is only observed by partial sensors. Besides, the fusion method for the labeled RFS may encounter the label inconsistency sensitivity (LIS) problem. In this paper, we give the conditions for avoiding the LIS problem. Furthermore, the consensus information selector method is proposed for the distributed fusion problem with different local FoVs. At each time step, the local multitarget densities are converged within a finite number of iterations. The simulation results demonstrate that the theoretical conclusion is reasonable and the algorithm is effective. • The LIS problem can be avoided given the condition that all sensors share the same prior density. • The consensus information selector is proposed to solve the distributed fusion problem. • The consensus information selector is proved to converge within finite iterations. [ABSTRACT FROM AUTHOR]