Minimal-time network reconstruction for DTLTI systems

Reference : Minimal-time network reconstruction for DTLTI systems


	Document type :	Scientific congresses, symposiums and conference proceedings : Paper published in a book
	Discipline(s) :	Engineering, computing & technology : Multidisciplinary, general & others
	To cite this reference:	http://hdl.handle.net/10993/20353

Title :	Minimal-time network reconstruction for DTLTI systems
Language :	English
Author, co-author :	Yuan, Y. [> >]
	Goncalves, Jorge [University of Luxembourg > Luxembourg Centre for Systems Biomedicine (LCSB) > >]
Publication date :	2010
Main document title :	The proceedings of the 49th IEEE Conference on Decision and Control (CDC)
Publisher :	IEEE
Pages :	2676 - 2681
Peer reviewed :	Yes
ISBN :	978-1-4244-7745-6
Event name :	49th IEEE Conference on Decision and Control (CDC)
Event date :	December 15-17, 2010
Event place (city) :	Atlanta, GA
Event country :	USA
Abstract :	[en] This paper considers the problem of obtaining in minimal time the “dynamical network structure” (DNS) from partial state observations of a discrete-time linear time-invariant system. From the DNS, we can not only obtain the network structure of the system at the measurement level, but also estimate the minimal number of hidden states which are not observed directly. First, we discuss when reconstruction of the DNS is and is not possible. Then, we give an algorithm to find the minimal number of successive outputs to find the DNS. Finally, we discuss extensions of the results to non-linear and noisy systems. These results can be directly applied to the decentralised network control problem of multi-agent systems to find network connections of the observed agents.
Permalink :	http://hdl.handle.net/10993/20353
DOI :	10.1109/CDC.2010.5718100

File(s) associated to this reference

Fulltext file(s):

	File	Commentary	Version	Size	Access
Limited access	Minimal-time network reconstruction for DTLTI systems.pdf		Publisher postprint	338.78 kB	Request a copy

All documents in ORBi^lu are protected by a user license.