On Scalable Matching Algorithm for Multi-Targets Over Bearing-Only Sensor Networks

This paper addresses the problem of single-scan bearing-only measurement matching for multi-targets in three-dimensional space over sensor networks. Since bearing-only measurements lack one dimension of information about targets, the corresponding matching problem cannot be solved using existing data association algorithms designed for three-dimensional measurements. Also, single-scan matching requires algorithms to be efficient without relying on prior information. These challenges make the single-scan bearing-only measurement matching problem in sensor networks an open research question. To tackle this issue, this paper proposes a scalable matching algorithm for bearing-only sensor networks using a two-step approach. First, a cost function is designed to quantitatively evaluate the disparity between pairs of bearing-only measurements, incorporating analytic and geometric attributes. Then, a cost-function-based algorithm for matching measurements between two bearing-only sensors is developed. Finally, by combining the proposed two-sensor algorithm with the Minimum Spanning Tree technique, a scalable matching algorithm for bearing-only sensor networks is constructed. The effectiveness of the algorithm is demonstrated through simulation results under several typical scenarios.