Change search
ReferencesLink to record
Permanent link

Direct link
On the tightness of linear policies for stabilization of linear systems over Gaussian networks
KTH, School of Electrical Engineering (EES), Communication Theory. KTH, School of Electrical Engineering (EES), Centres, ACCESS Linnaeus Centre.ORCID iD: 0000-0002-0036-9049
KTH, School of Electrical Engineering (EES), Communication Theory. KTH, School of Electrical Engineering (EES), Centres, ACCESS Linnaeus Centre.ORCID iD: 0000-0002-7926-5081
2016 (English)In: Systems & control letters (Print), ISSN 0167-6911, E-ISSN 1872-7956, Vol. 88, 32-38 p.Article in journal (Refereed) PublishedText
Abstract [en]

In this paper, we consider stabilization of multi-dimensional linear systems driven by Gaussian noise controlled over parallel Gaussian channels. For such systems, it has been recognized that for stabilization in the sense of asymptotic stationarity or stability in probability, Shannon capacity of a channel is an appropriate measure on characterizing whether a system can be made stable when controlled over the channel. However, this is in general not the case for quadratic stabilization. On a related problem of joint-source channel coding, in the information theory literature, the source-channel matching principle has been shown to lead to optimality of uncoded or analog transmission and when such matching conditions occur, it has been shown that capacity is also a relevant figure of merit for quadratic stabilization. A special case of this result is applicable to a scalar LQG system controlled over a scalar Gaussian channel. In this paper, we show that even in the absence of source-channel matching, to achieve quadratic stability, it may suffice that information capacity (in Shannon’s sense) is greater than the sum of the logarithm of unstable eigenvalue magnitudes. In particular, we show that periodic linear time varying coding policies are optimal in the sense of obtaining a finite second moment for the state of the system with minimum transmit power requirements for a large class of vector Gaussian channels. Our findings also extend the literature which has considered noise-free systems.

Place, publisher, year, edition, pages
Elsevier, 2016. Vol. 88, 32-38 p.
Keyword [en]
Channel capacity, Gaussian channels, Networked control systems, Stabilization
National Category
Signal Processing
Identifiers
URN: urn:nbn:se:kth:diva-180930DOI: 10.1016/j.sysconle.2015.09.013ISI: 000370101400004ScopusID: 2-s2.0-84951037908OAI: oai:DiVA.org:kth-180930DiVA: diva2:897803
Note

QC 20160126

Available from: 2016-01-26 Created: 2016-01-25 Last updated: 2016-03-19Bibliographically approved

Open Access in DiVA

No full text

Other links

Publisher's full textScopus

Search in DiVA

By author/editor
Oechtering, TobiasSkoglund, Mikael
By organisation
Communication TheoryACCESS Linnaeus Centre
In the same journal
Systems & control letters (Print)
Signal Processing

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Altmetric score

Total: 16 hits
ReferencesLink to record
Permanent link

Direct link