A Distributed Bayesian Data Fusion Algorithm With Uniform Consistency

• 成果类型: Article
• 署名作者: Li, Yingke; Zhou, Enlu; Zhang, Fumin
• 署名单位: University System of Georgia; Georgia Institute of Technology; University System of Georgia; Georgia Institute of Technology
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2024.3375254
• 发表日期: 2024
• 页码: 6176-6182
• 关键词: Robot sensing systems; data integration; Bayes methods; Network topology; TOPOLOGY; probability distribution; data models; Bayesian consistency; Bayesian learning; distributed data fusion

摘要

Distributed data fusion methods, which possess guaranteed performance for ad hoc and arbitrarily connected networks, empower more scalable, flexible, and robust information fusion for multirobot sensor networks. This article proposes a novel distributed Bayesian data fusion algorithm, which ensures uniform consistency, i.e., all the locally estimated distributions converge to the true distribution, for arbitrary periodically connected communication graphs. Conservative fusion via the weighted exponential product (WEP) rule is utilized to combat inconsistencies that arise from double-counting common information between fusion agents, and the WEP fusion weight is chosen based on the dynamic communication network topology. The uniform consistency of the proposed algorithm is rigorously proved, and the cooperative consistency conditions that guarantee uniform consistency have been explicitly identified. The performance and convergence properties of the proposed algorithm are validated through simulations.