Abstract
We study networked control of linear discrete-time systems using self-triggered strategies to reduce the amount of communication. At each transmission, the controller determines the next transmission time in advance based on the current state. We propose three self-triggered strategies which guarantee control performance based on a quadratic cost function. They have different characteristics with respect to the computation load for finding the transmission times. Through a numerical example, we demonstrate the tradeoff between computation loads and transmission frequencies.
| Original language | English |
|---|---|
| Pages (from-to) | 280-285 |
| Number of pages | 6 |
| Journal | IFAC-PapersOnLine |
| Volume | 49 |
| Issue number | 22 |
| DOIs | |
| Publication status | Published - 2016 |
| Externally published | Yes |
Keywords
- Hybrid systems
- Networked control systems
- Self-triggered control
ASJC Scopus subject areas
- Control and Systems Engineering
Access to Document
Other files and links
Fingerprint
Dive into the research topics of 'Self-triggered control for communication reduction in networked systems'. Together they form a unique fingerprint.Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS
In: IFAC-PapersOnLine, Vol. 49, No. 22, 2016, p. 280-285.
Research output: Contribution to journal › Article › peer-review
}
TY - JOUR
T1 - Self-triggered control for communication reduction in networked systems
AU - Akashi, Shigeru
AU - Ishii, Hideaki
AU - Cetinkaya, Ahmet
N1 - Funding Information: ∗∗∗ Department of Computer Science, Tokyo Institute of Technology, DepartmentofComYopukohamter Sacie226-nce8502,, TokyJoaIpnasntitute of Technology, Yokohama 226-8502, Japan Yokohama 226-8502, Japan Abstract: We study networked control of linear discrete-time systems using self-triggered Abstract: We study networked control of linear discrete-time systems using self-triggered Abstract: We study networked control of linear discrete-time systems using self-triggered dstertaetremgiensestotherednuexcte trhaensammisosuionnt toimf ecoimnmaudnviacnacteionb.asAedt oenachthetrcaunrsrmenistsisotna,tet.hWe ecopnrtorpololseer determines the next transmission time in advance based on the current state. We propose determines the next transmission time in advance based on the current state. We propose three self-triggered strategies which guarantee control performance based on a quadratic cost three self-triggered strategies which guarantee control performance based on a quadratic cost function. They have di∆erent characteristics with respect to the computation load for finding tfuhnecttrioann.smTihsesiyonhatvime edsi.∆eTrhenrotucghhara cntuermisetricicsawl ietxhamrepsplee,cwt etodtehmeocnosmtrpatuetatthioentrlaodadeof∆orbfeitnwdeineng the transmission times. Through a numerical example, we demonstrate the tradeo∆ between the transmission times. Through a numerical example, we demonstrate the tradeo∆ between computation loads and transmission frequencies. computation loads and transmission frequencies. © 2016, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: Networked control systems, Self-triggered control, Hybrid systems Keywords: Networked control systems, Self-triggered control, Hybrid systems Keywords: Networked control systems, Self-triggered control, Hybrid systems 1. INTRODUCTION et al. (2014); Zhang et al. (2015)). We develop three self-1. INTRODUCTION et al. (2014); Zhang et al. (2015)). We develop three self-1. INTRODUCTION tertigalg.er(ed2014)sch;emZhesantghaettraleq.u(2015)ire di∆).erWenetdleevveellops ofthreareel-tsiemlfe- In recent years, the use of communication networks in triggered schemes that require di∆erent levels of real-time IInn rreecceenntt yyeeaarrss,, tthhee uussee ooff cocommmmuunniicacattiioonn nneettwwoorrkkss iinn ttrriiggggerereded sscchhememeses tthhaatt rreqequuiirree ddii∆∆erereenntt lleevveellss ooff rreaeall--ttiimmee control systems has drastically increased for connecting computation for finding the next transmission time at the control systems has drastically increased for connecting computation for finding the next transmission time at the control systems has drastically increased for connecting goivnetnroclloenrt.roWlhpielrefoarlml aonfcteh,etmheyareexhgiubaitratnratedeedo∆tsobaecthwieeevne plants with controllers which may be remotely located. controller. While all of them are guaranteed to achieve Dpluaentsowthiethshcaornetdronlaletrusrewohfinchetwmoaryksbaeswreemlloatsetlyhelolicmaitteed. givencontrolperformance,theyexhibittradeo∆sbetween DDueuettootthehesshaharreeddnanatturureeooffnenettwwoorrkkssaasswweellllaasstthehelliimmiitteedd ggiivveenn coconnttrrooll ppererffoorrmmaannce,ce, tthheyey exexhhiibbiitt ttrraaddeoeo∆∆ss bbeettwweeeenn computation in embedded devices, it is important to de-the necessary computation and the length of waiting times computationinembeddeddevices,itisimportanttode- bthheeefonnereecceethsssseaarrnyyexccootmmtrpputauntsaamttiiioosnnsioaanndnsd. ttHhheenlleecnngeg, ttdhheoopffenwwdaaiiinttiignnggonttiimmtheeess sciogmnpsutcahtinoentwinorekmedbceodndterdoldseyvsitceems,siwt iitshicmeprtoaritnancotntsoiddeer--before the next transmissions. Hence, depending on the signsuchnetworkedcontrolsystemswithcertainconsider- bbyeesffooterrmee ttrhheeequnneireexxmtt ettnrraatsnns,stmmhiiessssmiiooonnssst.. aHHpeepnncrcoeep,,riddeaetppeeeonndipdtiinngogn ooshnnottuhhelde asitgionnssuctoh nkeeetwp otrhkeedcocmonmtruonlicsyastitoenmas nwditchocmeprtuatianticoon sliodaedrs-system requirements, the most appropriate option should atatiiononssttookkeeeepptthheeccomommmuunniiccatatiiononananddccomomppuuttatatiiononlloadoadss ssyysstteemm rreeqquiuirreemmeennttss,, tthehe mmoosstt aapprpprooprpriiaattee ooptptiioonn sshohoululdd low. In this respect, conventional digital control techniques be chosen. The first strategy is based on computing the low. In this respect, conventional digital control techniques be chosen. The first strategy is based on computing the leomwp.loInyinthgispreesriopectdic, scoamnvpelinntigonmaalydingoittabl coe idnteraol.l techniques futotmurpreutsattaiotenausllsyingignttehenesivplel.anTthemosedeceolnadnddstriastehegeynccreeqmuiorrees employing periodic sampling may not be ideal. computationally intensive. The second strategy requires Reduction in communication can be achieved by activat-computationally intensive. The second strategy requires Reductionincommunicationcanbeachievedbyactivat- much less real-time computation by using bounds on the iRRngededuutrctctansiioonnmiiinnssicocoonsmmmmonluunnyiicacawttheiioonnncacaitnnisbbneeeaacccehhssiieeavvryee.ddTbbhiyysaactctisiivvtaahett-- mmtauutccehhtlleerassssjerrceetaaollr--itteiismm,eebutccoomminputputgaaenettiioornnalbbiyysususmiiongngrebbdeooundsundsmandioongnn tthehein ing transmissions only when it is necessary. This is the state trajectories, but in general is more demanding in uinngdetrrlayninsmg iisdseioansinonthlye wsthraentegitiesisonfeecveesnsatr-tyr.igTgheirsedisctohne-state trajectories, but in general is more demanding in underlying idea in the strategies of event-triggered con- terms of communication. In the third one, the amount of tuurnnoddleerarllnyydiinnsggeliifdd-teearaigiignnerttehhdeecssottnrraatrttoeeglg,iieeswshooicffheevvheeannvtte--ttrrlaiiggtggeleeryreedgdaicconoennd-- toen-rmlines ocfocmomputmautnioicnatiisofn.urtIhen trhereducthireddobney,pathertitaiomniongunttheof trol and self-triggered control, which have lately gained on-line computation is further reduced by partitioning the mttrroloulchaannaddttsseeenlltff--iottrrnii;gggegserreee,ddecc.goo.nn,ttrrol(oMl,,awwzohhiiacchhndhhaaTvvaeebullaatatdeeallyy(ggai2a0i0nn8ee)dd; oon-n-lliinene ccoommputputaattiioonn iiss ffururttheherr rreeducduceedd bbyy paparrttiittiiooniningng tthehe much attention; see, e.g., (Mazo and Tabuada (2008); state space into a finite number of regions, where each Hmeuecmhelatsteetntalion. ;(2013)see, ;e.Cg.e,ti(nMkaazyaoeant ald.T(a2016)buad)aan(2008)d the; restagitone shpaascethinetcoorrea finispontedinnugmtbranersomfisrseigonionstim, weshe. re each Heemels et al. (2013); Cetinkaya et al. (2016)) and the regionhasthecorrespondingtransmissiontimes. Heemels et al. (2013); Cetinkaya et al. (2016)) and the Inallthreestrategies,wefollowthecontrolmethodregionhasthecorrespondingtransmissiontimes. trhefeerpelnacnets itshecroeninti.nIunouesvleyntm-tornigitgoerreedd, cbounttronl,lythwe hsteantethoef In all three strategies, we follow the control method tthehe plplaanntt iiss ccoonnttiinnuuooususllyy mmooninittoorreed,d, butbut oonlnlyy wwhehenn tthehe IInn aallll ththrreeee sstrtraateteggiieess,, wwee ffoollllooww ththee ccoonnttrrooll mmeeththoodd state value has sufficiently changed and satisfies certain developed by (Ishii and Francis (2002)) in the context of state value has sufficiently changed and satisfies certain qdueevvaeenlltoopipzeeddcbboyyn((tIIrssohhl.iiiiTaanhneddreFF,rraaannLcciiyssap((22002)u0n0o2v)-))baiinnsettdhheeapccpoornnottaeexxchtt oofisf cstoantdeitvioanluse, choams ms ufnfiiciaetnitolny icshtarnigggedereadndfosrathisefiecsoncterrotlalienr quantized control. There, a Lyapunov-based approach is conditions, communication is triggeredfor the controller qquuaevaennlttoiipzzeedd fccooornnfttinrroodlli..nTTg hhetherreees,,oaa-caLLlyyleaadppunoduwnoevvll--bbatiamsseeddinaacpppropnrtooiaanccuhhouiisss tcoonbdeitiuopndsa, tceodm(mTuabnuicaadtaion(2i0s07tr)i)g.gOernedthfoerotherchonatnrdo,llienr developed for finding the so-called dwell time in continuous to be updated (Tabuada (2007)). On the other hand, in developed for finding the so-called dwell time in continuous self-triggered control, when the controller transmits the tiomnte,r,owl.hTichhesiesrienferfaencttceclslofuselrltyherrellcaotnedsdidteor qevueant-itzraitgigoernredodf newcontrolinputtotheplant,itisaccompaniedwiththesoelfb-treiupggedaretdedco(nTtarbuaol,dawhe(n20th07e)).coOnntrotlhelerotrtheanrsmhaitsnd,thine tcohoennttrrcoolln..tTTrohhleessineeprrueeftfeerraeennndcceseshoffuuwrrttthhoeerrrccoeodnnussciieddeertrhqqeuudaaannttaiizzaarttaiiootenn ioonff innefworcmoantiroonl irnepguatrdtointghethpelatnitm, eit wishaecncotmhepasneinesdorwisthhotuhlde the control input and how to reduce the data rate in information regarding the time when the sensor should ttheheommccoounnnttirrcooalltiiionputnputn.Hoaawndndevehohor,wwintttoohisrreepaducducpeeer,tthehewedadaemttaaplorryaattoeenliinny send the state the next time (Wang and Lemmon (2009); communication. However, in this paper, we employ only sendthestatethenexttime(WangandLemmon(2009);informationregardingthetimewhenthesensorshould tchooemmmmiduueannsiiccfaaottriiootnn.h.eHHsooawwmeepvvleeinrr,,giinnpattrhhitissofppatahppeeerr,,eswwueeltseemmthpplelooreyy. ooTnnlhlyye Gseonmd mthaensteattealt.h(e20n1e4x)t).tiSminec(eWthaengtranndsmLiesmsiomnotnim(2e0s0a9r)e; the ideas for the sampling part of the results there. The Gommansetal.(2014)).Sincethetransmissiontimesare itnhheteeriieddesetaainssgffooarrsptthheeectssaaismmtpplhlaiinngtgthppaearrstt aootffettshhepeacrreessuulislttpssrttohhejeecrreete..dTTohhene dGeotmermainnesdetinala.d(v2a0n1c4e),).seSlifn-tcreigtgheretdracnosnmtrisosliomnatyimreeqsuairree interesting aspect is that the state space is projected on determined in advance, self-triggered control may require interesting aspect is that the state space is projected on determined in advance, self-triggered control may require reductwoe-diremale-nstimioenaclomspaputcea,tiwohincinh tehenablo∆e-slineusctaosee.specially more communication in general compared to the event-a two-dimensional space, which enables us to especially mtrioggererecodmcmasuen.icaThteioadnvinangtenageeriaslhcoowmepvaerredthattomthoneietorivenntg-reducereal-timecomputationintheo∆-linecase. otrifggetherestdatceasies.unneThecadessvaarnytageat tihes hsoewnsevoerrsitdehat. monitoring This paper is organized as follows. In Section 2, we formu-of the state is unnecessary at the sensor side. Tahteistphaepneertiws orkgeadniczoendtraoslfpolrlobwles.mInstSuedciteidon. I2n, wSeecftoiornmu3-, In this paper, we study self-triggered control strategies for late the networked control problem studied. In Section 3, In this paper, we study self-triggered control strategies for late the networked control problem studied. In Section 3, Ilinnetahristpimape-eir,nvwaeriastnutdsyyssteelfm-tsriinggetreheddcisoncrterolte-sttimrateedgioemsaforin wnetepreretsheentdeasLiryeadpunoluevneolvo-bafacsoendtrsouffilffpiceirefnotrmcoandindciet.ioSnecttoiognuasa4r-, (liEnqeatarmtimeet-inalv.ar(i2a0n1t0)s;ysBtermunsnienr tehteadl.isc(2re0t1e5-t)i;mGeodmomaanins antee the desired level of control performance. Sections 4, (Eqtami et al. (2010); Brunner et al. (2015); Gommans aann, ttaeeenedtthh6ee ddpeesrsoiivrreediddelleetvvheeell oodffetccoaoinnlsttrroollf ppteerhreffoorrtmmhraaennecce.ese. lSSf-eecttcrtigiioognnessre44d,, (Eqtami et al. (2010); Brunner et al. (2015); Gommans 5, and 6 provide the details of the three self-triggered 5, and 6 provide the details of the three self-triggered 1 This work was supported in part by the JST-CREST Program control strategies. We illustrate the results through a nu-11 This work was supported in part by the JST-CREST Program moenrtircoal setxraamtepgilesi.nWSecitlliuosntr7a.teInthSecrteisounlt8s,twhreoguigvhe asonmue-a1nTdhbisywJoSrPkSwausndsuerppGortaendt-in-Apaidrtfboyr tShceienJtSiTfic-CReEsSeaTrcPhroGgranmt merical example in Section 7. In Section 8, we give some and by JSPS under Grant-in-Aid for Scientific Research Grant merical example in Section 7. In Section 8, we give some anod. 1b5yH0J4S0P20S. under Grant-in-Aid for Scientific Research Grant concluding remarks. NNo.oE. 15H1-m5Hai04020.0l4s:0a2k0a.shi@sc.dis.titech.ac.jp(S.Akashi),ishii@c.titech.ac.jp concluding remarks. No.E15H-mails04020.:akashi@sc.dis.titech.ac.jp(S.Akashi),ishii@c.titech.ac.jp (H.EI-smhiaii)l,s:aahkmaesht@i@dscl..mdies.it.tititeecchh.a.acc.j.pp(S(A. A. Ckaesthini)k,aiysah)ii@c.titech.ac.jp (H. Ishii), ahmet@dsl.mei.titech.ac.jp (A. Cetinkaya) (H. Ishii), ahmet@dsl.mei.titech.ac.jp (A. Cetinkaya) C24o0p5y-r8i9g6h3t ©© 22001166, IIFFAACC (International Federation of Automatic Contr2o8l)0 Hosting by Elsevier Ltd. All rights reserved. CCPooeeppryy rrreiiggvhhiettw ©© u 22n00d11e66r rIIFFesAApCConsibility of International Federation of Automa2802ti8c0 Control. Copyright © 2016 IFAC 280 10.1016/j.ifacol.2016.10.410 Publisher Copyright: © 2016
PY - 2016
Y1 - 2016
N2 - We study networked control of linear discrete-time systems using self-triggered strategies to reduce the amount of communication. At each transmission, the controller determines the next transmission time in advance based on the current state. We propose three self-triggered strategies which guarantee control performance based on a quadratic cost function. They have different characteristics with respect to the computation load for finding the transmission times. Through a numerical example, we demonstrate the tradeoff between computation loads and transmission frequencies.
AB - We study networked control of linear discrete-time systems using self-triggered strategies to reduce the amount of communication. At each transmission, the controller determines the next transmission time in advance based on the current state. We propose three self-triggered strategies which guarantee control performance based on a quadratic cost function. They have different characteristics with respect to the computation load for finding the transmission times. Through a numerical example, we demonstrate the tradeoff between computation loads and transmission frequencies.
KW - Hybrid systems
KW - Networked control systems
KW - Self-triggered control
UR - http://www.scopus.com/inward/record.url?scp=84994159044&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84994159044&partnerID=8YFLogxK
U2 - 10.1016/j.ifacol.2016.10.410
DO - 10.1016/j.ifacol.2016.10.410
M3 - Article
AN - SCOPUS:84994159044
SN - 2405-8963
VL - 49
SP - 280
EP - 285
JO - IFAC-PapersOnLine
JF - IFAC-PapersOnLine
IS - 22
ER -