Measurement based dimension descent association algorithm for OTHR multi-detection multi-target tracking

• Published in: ISA transactions Vol. 140; pp. 18 - 31
• Main Authors: Zhang, Yujie, Zhang, Zheng, Yu, Jianglong, Li, Qingdong, Dong, Xiwang, Ren, Zhang
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.09.2023
• Subjects: Convergence analysis; Dimension descent association; Multi-detection multi-target tracking; Target detection; Dimension descent association; Multi-detection multi-target tracking; Target detection; Convergence analysis
• ISSN: 0019-0578; 1879-2022; 1879-2022
• DOI: 10.1016/j.isatra.2023.05.016

The problem of multi-detection multi-target tracking (MDMTT) using over-the-horizon radar in dense clutter environment is studied in this paper. The biggest challenge of MDMTT is the 3-dimensional multipath data association among measurements, detection models and targets. In particular, a lot of clutter measurements are generated in dense clutter environment, which increase the computational burden of 3-dimensional multipath data association greatly. A measurement based dimension descent association (DDA) algorithm is proposed to solve the 3-dimensional multipath data association, which decomposes the 3-dimensional multipath data association into two 2-dimensional data associations. The proposed algorithm can reduce the computational burden compared with the optimal 3-dimensional multipath data association and the computational complexity is analyzed. Besides, a time extension method is designed to detect the new-born targets that appear in the tracking scene, which is based on the sequential measurements. The convergence of the proposed measurement based DDA algorithm is analyzed. The estimation error can convergence to 0 as the number of Gaussian mixtures tends to infinity. The effectiveness and rapidity of the measurement based DDA algorithm are demonstrated by the comparative simulation with the previously proposed algorithms. •Multi-detection multi-target tracking using OTHR is studied by dimension descent.•The detection of new-born targets with unknown prior knowledge is considered.•The convergence of the novel multi-target tracking algorithm is analyzed.•The labeled GMPHD is proposed to reduce the computational burden.