A distributed subband valley fusion (DSVF) method for low frequency broadband target localization

A distributed subband valley fusion method is proposed in this paper for target localization for a multi-array fusion system. Instead of employing two-step processing like the traditional bearings-only target localization (BOTL) methods, the target location estimate is directly obtained from the intersection of the estimated bearing line of each array, which is automatically produced after applying a subband valley energy detection process to the reciprocal of the steered response power over predefined grid points. Bearing information at all frequencies and ranges are used to generate the bearing line and the intersection structure, leading to higher localization accuracy in comparison with the BOTL methods. In multiple-target situations, the data association problem is solved by distinguishing true targets from the ghost ones from the power and the number of bearing lines at the corresponding intersection grid point. The method is subsequently tested by the SWellEx-96 shallow water experiment data set. Simulation and experimental results demonstrate the localization accuracy of the proposed method.