Distributed Particle Filtering Algorithm Based on Adaptive Hybrid Mesh Partitioning
-
摘要: 在分布式网络中, 传统粒子滤波方法难以在跟踪精度与计算开销之间取得良好平衡. 为此, 本文提出一种自适应混合网格分布式粒子滤波算法. 首先引入一种基于信息论准则的自适应网格策略, 动态优化状态空间的划分方式, 使计算资源集中于后验分布的高概率区域, 从而提升表示效率. 其次, 针对分布式信息融合, 设计一种新的融合机制, 使节点能够直接融合以网格权重表示的局部后验分布, 有效避免参数化方法中常见的模型失配与信息损失问题, 并显著降低了通信负担. 此外, 为保障融合过程的鲁棒性, 采用对数域Gossip协议驱动融合迭代, 并从理论上证明了其在概率域中的收敛性. 蒙特卡罗仿真结果表明, 相较于固定网格与高斯混合融合等基准方法, 所提算法在显著降低资源消耗的同时, 其跟踪精度最接近集中式性能上界, 展现出优越的估计精度与系统可扩展性.Abstract: In distributed networks, traditional particle filtering methods struggle to strike a good balance between tracking accuracy and computational cost. To address this issue, this paper proposes an adaptive hybrid grid distributed particle filtering algorithm. Firstly, it introduces an adaptive grid strategy based on information theory criteria to dynamically optimize the partitioning of the state space, concentrating computational resources on the high-probability regions of the posterior distribution, thereby enhancing representation efficiency. Secondly, for distributed information fusion, a new fusion mechanism is designed, enabling nodes to directly fuse local posterior distributions represented by grid weights, effectively avoiding common model mismatch and information loss problems in parametric methods and significantly reducing communication overhead. Additionally, to ensure the robustness of the fusion process, a log-domain Gossip protocol is adopted to drive the fusion iterations, and its convergence in the probability domain is theoretically proven. Monte Carlo simulation results show that, compared with fixed-grid and Gaussian mixture fusion benchmark methods, the proposed algorithm significantly reduces resource consumption while achieving tracking accuracy close to the upper bound of centralized performance, demonstrating superior estimation accuracy and system scalability.
-
Key words:
- Particle filter /
- adaptive algorithm /
- stochastic Gossip algorithm /
- grid-based /
- consensus fusion
-
图 4 无线传感器网络、通信链路以及目标轨迹
Fig. 4 Wireless sensor network, communication links, and target trajectory
图 5 不同时间步下网格平均KLD随Gossip迭代次数$ l$的收敛曲线
Fig. 5 Convergence profiles of the grid average KLD with respect to Gossip iterations $ l$ for various time steps
表 1 不同算法的相对性能量化对比
Table 1 Quantitative performance comparison of different algorithms
算法 ARMSE(m) 相对误差(%)(所提算法为基准) 自适应网格 0.66 0.00 20$ \times $20固定粗网格 1.95 195.45 100$ \times $100固定细网格 0.68 3.03 GMM-Fusion 0.97 46.97 集中式粒子滤波 0.52 -21.21 JE-MPF 1.33 101.50 
下载: 导出CSV
表 2 不同分布式滤波算法的融合计算复杂度对比
Table 2 Comparison of fusion computational complexity of different distributed filtering algorithms
算法 融合计算复杂度$ T_{\text{fusion}} $ $ 100 \times 100 $固定细网格 $ \mathrm{O}(L \cdot M_{\text{fix1}}) $ $ 20 \times 20 $固定粗网格 $ \mathrm{O}(L \cdot M_{ \text{fix2}}) $ GMM-Fusion $ \mathrm{O}(L \cdot K^2 \cdot d^2) $ JE-MPF (基准算法) $ \mathrm{O}\big(L \cdot (M_{\Omega} + N)\big) $ 本文算法 $ \mathrm{O}(L \cdot \bar{M}_{\text{active}}) $
下载: 导出CSV
-
[1] Mahler R. Statistical Multisource-Multitarget Information Fusion. Norwood, MA, USA: Artech House, 2007. [2] 孟琭, 杨旭. 目标跟踪算法综述. 自动化学报, 2019, 45(7): 1244−1260 doi: 10.16383/j.aas.c180277Meng Lu, Yang Xu. A survey of object tracking algorithms. Acta Automatica Sinica, 2019, 45(7): 1244−1260 doi: 10.16383/j.aas.c180277 [3] Yang B Y, Yang E F, Shi H B, Yu L J, Niu C. Adaptive square-root cubature Kalman filter based low cost UAV positioning in dark and GPS-denied environments. IEEE Transactions on Intelligent Vehicles, 2025, 10(5): 3587−3599 doi: 10.1109/TIV.2024.3457678 [4] Yoo T H, Bang H T, Youn W K. Adaptive DBSCAN-based probabilistic data association filter for maritime object tracking using RADAR. IEEE Sensors Letters, 2025, 9(6): 1−4 doi: 10.1109/lsens.2025.3565796 [5] 杨峰, 郑丽涛, 王家琦, 潘泉. 双层无迹卡尔曼滤波. 自动化学报, 2019, 45(7): 1386−1391 doi: 10.16383/j.aas.c180349Yang Feng, Zheng Li-Tao, Wang Jia-Qi, Pan Quan. Double-layer unscented Kalman filter. Acta Automatica Sinica, 2019, 45(7): 1386−1391 doi: 10.16383/j.aas.c180349 [6] Gordon N J, Salmond D J, Smith A F M. Novel approach to nonlinear/non-Gaussian Bayesian state estimation. IEE Proceedings F (Radar and Signal Processing), 1993, 140(2): 107−113 doi: 10.1049/ip-f-2.1993.0015 [7] 盛立, 刘一凡, 高明, 周东华. 旋转导向钻井工具系统实时测量的智能粒子滤波方法. 自动化学报, 2025, 51: 1−11 doi: 10.16383/j.aas.c250136Sheng Li, Liu Yi-Fan, Gao Ming, Zhou Dong-Hua. An intelligent particle filter method for real-time measurement of rotary steerable system. Acta Automatica Sinica, 2025, 51: 1−11 doi: 10.16383/j.aas.c250136 [8] 卢锦, 马令坤, 吕春玲, 章为川, 孙长明. 基于代价参考粒子滤波器组的多目标检测前跟踪算法. 自动化学报, 2024, 50(4): 851−861 doi: 10.16383/j.aas.c220635Lu Jin, Ma Ling-Kun, Lyu Chun-Ling, Zhang Wei-Chuan, Sun Chang-Ming. A multi-target track-before-detect algorithm based on cost-reference particle filter bank. Acta Automatica Sinica, 2024, 50(4): 851−861 doi: 10.16383/j.aas.c220635 [9] Xia W, Sun M Q, Wang Q. Direct target tracking by distributed Gaussian particle filtering for heterogeneous networks. IEEE Transactions on Signal Processing, 2020, 68: 1361−1373 doi: 10.1109/TSP.2020.2971449 [10] Jiang C Y, Wu J, Gou R, Fu J F. Research on adaptive particle swarm optimization particle filter target tracking algorithm in wireless sensor networks. Advanced Control for Applications: Engineering and Industrial Systems, 2024, 6(3): e205 doi: 10.1002/adc2.205 [11] Liu Y, Coombes M, Liu C J. Mesh-based consensus distributed particle filtering for sensor networks. IEEE Transactions on Signal and Information Processing over Networks, 2023, 9: 346−356 doi: 10.1109/TSIPN.2023.3278469 [12] Jeon H C, Park W J, Park C G. Grid design for efficient and accurate point mass filter-based terrain referenced navigation. IEEE Sensors Journal, 2018, 18(4): 1731−1738 doi: 10.1109/JSEN.2017.2779463 [13] Zhou C L, Chen S Y, Wang C Y. Degradation estimation for distributed nonlinear systems: a PDF-consensus particle filtering method. IEEE Transactions on Network Science and Engineering, 2025, 12(2): 1408−1419 doi: 10.1109/TNSE.2025.3530158 [14] Li J C, Nehorai A. Distributed particle filtering via optimal fusion of Gaussian mixtures. IEEE Transactions on Signal and Information Processing over Networks, 2018, 4(2): 280−292 doi: 10.1109/TSIPN.2017.2694318 [15] Liu Y, Wang Z D, Liu C J, Coombes M, Chen W H. A novel algorithm for quantized particle filtering with multiple degrading sensors: degradation estimation and target tracking. IEEE Transactions on Industrial Informatics, 2023, 19(4): 5830−5838 doi: 10.1109/TII.2022.3176910 [16] Hu J, Fan S T, Chen C, Liu H J, Yi X J. Encoding-decoding-based distributed fusion filtering for multi-rate nonlinear systems with sensor resolutions. IEEE Transactions on Signal and Information Processing over Networks, 2023, 9: 811−822 doi: 10.1109/TSIPN.2023.3334496 [17] Gu D, Sun J, Hu Z, Li H. Consensus based distributed particle filter in sensor networks. In: Proceedings of the International Conference on Information and Automation. Zhangjiajie, China: IEEE, 2008. 302–307. [18] Li T, Corchado J M, Sun S. Partial consensus and conservative fusion of Gaussian mixtures for distributed PHD fusion. IEEE Transactions on Aerospace and Electronic Systems, 2018, 55(5): 2150−2163 doi: 10.1109/taes.2018.2882960 [19] Battistelli G, Chisci L, Fantacci C, Farina A, Graziano A. Consensus CPHD filter for distributed multitarget tracking. IEEE Journal of Selected Topics in Signal Processing, 2013, 7(3): 508−520 doi: 10.1109/JSTSP.2013.2250911 [20] Guldogan M B. Consensus Bernoulli filter for distributed detection and tracking using multi-static Doppler shifts. IEEE Signal Processing Letters, 2014, 21(6): 672−676 doi: 10.1109/LSP.2014.2313177 [21] Battistelli G, Chisci L. Kullback-Leibler average, consensus on probability densities, and distributed state estimation with guaranteed stability. Automatica, 2014, 50(3): 707−718 doi: 10.1016/j.automatica.2013.11.042 [22] Boyd S, Ghosh A, Prabhakar B, Shah D. Randomized gossip algorithms. IEEE Transactions on Information Theory, 2006, 52(6): 2508−2530 doi: 10.1109/TIT.2006.874516 [23] Kayaalp M, İnan Y, Telatar E, Sayed A H. On the arithmetic and geometric fusion of beliefs for distributed inference. IEEE Transactions on Automatic Control, 2024, 69(4): 2265−2280 doi: 10.1109/TAC.2023.3330405 [24] Bar-Shalom Y. Tracking and Data Fusion: A Handbook of Algorithms. Storrs, CT, USA: YBS Publishing, 2011. -
下载:
计量
- 文章访问数: 3
- HTML全文浏览量: 4
- 被引次数: 0
