What is it about?
The article presents a solution for estimating the state-parameters of a dynamical process, such as estimating the traffic density profile on highways. The developed solution assumes a distributed implementation, which means that the entire set of state-parameters is estimated by numerous subsystems each having their own local estimation results. Subsystems can exchange their local results with other subsystems via a communication network.
Featured Image
Why is it important?
The proposed solution allows that the implemented estimation algorithm differs per subsystem, thereby increasing the practical feasibility as subsystems can differ in their communication and computational resources. For example, one subsystem with high computational power can run a sophisticated estimator suitable for nonlinear processes, while one of its neighboring subsystems with low computational power runs a simple estimator based on a linear approximation. Nonetheless, both subsystems with assist each other with improving their local estimation results.
Perspectives
This is a first study of a heterogeneous solution for distributed state estimation, where one subsystem can run a Kalman filter for processing its local measurement while others run an unscented Kalman filter. A key aspect of this solution is its objective. As opposed to most solutions aiming for optimal estimation results per individual subsystems, our goal was to guarantee that unique information obtained by a single node locally is automatically distributed across the entire network via the exchange of local estimation results between neighbors. Thereby, if you could prove stability of the local estimation results in one subsystem, then one has the guarantee that all other subsystems will have stable estimation results as well. The motivation for a heterogeneous solution has been the drawback of existing homogenous solution to have the same resources at every subsystem in the network. Nonetheless, it should be noted that this first study assumes that local estimation results are described by a Gaussian, though the study also shows the potential of heterogeneous estimation as a new topic in distributed estimation.
Dr Joris Sijs
TNO
Read the Original
This page is a summary of: Heterogeneous state estimation in dynamic networked systems, IET Control Theory and Applications, October 2015, the Institution of Engineering and Technology (the IET),
DOI: 10.1049/iet-cta.2014.0811.
You can read the full text:
Contributors
The following have contributed to this page
