A Data Association Approach for Angle-only Sensors based on Faster Clustering and Multidimensional Assignment

Combining clustering and assignment techniques to solve data association problem can not only improve the computational efficiency, but also obtain the global optimal association result. For angle measurements, the existing method uses pairwise clustering approach, and the computational complexity is quadratic with both the number of sensors and measurements. This paper proposes a faster clustering approach, which benefits from two operations. First, to avoid missing potential associations, the patching operation is added, thus the sequential clustering approach can be used. Second, to prevent the exponential growth in the number of clusters, the pruning operation is added, thus the clustering efficiency is improved. Thanks to these two kinds of operations, the computational complexity of the proposed approach reduces to linear with the number of sensors. Simulation results show that, while ensuring the association accuracy, the proposed approach improve the association efficiency significantly.