Skip to main navigation Skip to search Skip to main content

Fully Distributed Filtering With a Stochastic Event-Triggered Mechanism

  • Jiachen Qian
  • , Peihu Duan
  • , Zhisheng Duan*
  • *Corresponding author for this work
  • Peking University
  • The University of Hong Kong

Research output: Contribution to journal › Article › peer-review

Abstract

This article focuses on distributed filtering for a discrete time-varying system observed by multiple smart sensors, where every sensor only measures partial state information of the target system and then sends it to a corresponding remote estimator. Subsequently, the estimator performs the local Kalman filter and shares its estimates with the estimators in its neighborhood in a distributed way. This article aims to reduce the communication rate between sensors and estimators, and guarantee the estimation performance, simultaneously. To achieve this goal, a novel distributed information fusion algorithm is designed by embedding a stochastic event-triggered communication mechanism. Based on a new developed mathematics technique, the consistency and stability of the proposed distributed state estimation algorithms are both ensured. Furthermore, compared with the literature, the stability can be guaranteed with a milder collectively uniformly observable condition. Moreover, the tradeoff between the communication rate and estimation performance is analyzed in a closed-form expression. Finally, the effectiveness of the theoretical results is demonstrated by several comparative numerical examples.

Original languageEnglish
Pages (from-to)753-762
Number of pages10
JournalIEEE Transactions on Control of Network Systems
Volume9
Issue number2
DOIs
Publication statusPublished - 1 Jun 2022
Externally publishedYes

Keywords

  • Distributed filtering
  • event triggered communication
  • information fusion
  • smart sensor

Fingerprint

Dive into the research topics of 'Fully Distributed Filtering With a Stochastic Event-Triggered Mechanism'. Together they form a unique fingerprint.

Cite this