Finite-horizon Gaussianity-preserving event-based sensor scheduling in Kalman filter applications

详细信息    查看全文

• 作者：Junfeng Wu^a ; ^b ; ^{junfengw@kth.se" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Xiaoqiang Ren^a ; ^{xren@connect.ust.hk" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Duo Han^a ; ¹ ; ^{dhanaa@connect.ust.hk" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Dawei Shi^c ; ^{dshi@ualberta.ca" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Ling Shi^a ; ^{eesling@ust.hk" class="auth_mail" title="E-mail the corresponding author}Author Vitae
• 关键词：Networked control systems ; Estimation ; Kalman filtering ; Sensor scheduling ; Dynamic programming
• 刊名：Automatica
• 出版年：2016
• 出版时间：October 2016
• 年：2016
• 卷：72
• 期：Complete
• 页码：100-107
• 全文大小：827 K

文摘

This paper considers a remote state estimation problem, where a sensor measures the state of a linear discrete-time system. The sensor has computational capability to implement a local Kalman filter. The sensor-to-estimator communications are scheduled intentionally over a finite time horizon to obtain a desirable tradeoff between the state estimation quality and the limited communication resources. Compared with the literature, we adopt a Gaussianity-preserving event-based sensor schedule bypassing the nonlinearity problem met in threshold event-based polices. We derive the closed-form of minimum mean-square error (MMSE) estimator and show that, if communication is triggered, the estimator cannot do better than the local Kalman filter, otherwise, the associated error covariance, is simply a sum of the estimation error of the local Kalman filter and the performance loss due to the absence of communication. We further design the scheduler’s parameters by solving a dynamic programming (DP) problem. The computational overhead of the DP problem is less sensitive to the system dimension compared with that of existing algorithms in the literature.