The WSN-DPCM ARTEMIS-JU Project

40 Pages·2016·11.39 MB·English

sensors
Article
Integrated Toolset for WSN Application Planning,
Development, Commissioning and Maintenance:
The WSN-DPCM ARTEMIS-JU Project
ChristosAntonopoulos1,†,KaterinaAsimogloy2,†,SarahChiti3,†,LucaD’Onofrio3,†,
SimoneGianfranceschi3,†,DanpingHe4,†,AntonioIodice5,†,StavrosKoubias1,†,
ChristosKoulamas6,†,LucianoLavagno7,†,MihaiT.Lazarescu7,*,†,GabrielMujica4,†,
GeorgePapadopoulos6,†,JorgePortilla4,†,LuisRedondo8,†,DanieleRiccio5,†,TeresaRiesgo4,†,
DanielRodriguez8,†,GiuseppeRuello5,†,VasilisSamoladas2,†,TsenkaStoyanova1,†,
GerasimosTouliatos1,†,AngelaValvo3,† andGeorgiaVlahoy2,†
1 AppliedElectronicsLaboratory,DepartmentofElectrical&ComputerEngineering,
UniversityofPatrasCampusofRion,Patras26500,Greece;[email protected](C.A.);
[email protected](S.K.);[email protected](T.S.);[email protected](G.T.)
2 TelecommunicationSystemsInstitute,TUCCampus,Kounoupidiana,Chania73100Greece;
[email protected](K.A.);[email protected](V.S.);[email protected](G.V.)
3 INTECSS.p.A.,ViaGiacomoPeroni130,Rome00131,Italy;[email protected](S.C.);
[email protected](L.D.);[email protected](S.G.);[email protected](A.V.)
4 CentrodeElectrónicaIndustrial,UniversidadPolitécnicadeMadrid,C/JoséGutiérrezAbascal,2,
Madrid28006,Spain;[email protected](D.H.);[email protected](G.M.);
[email protected](J.P.);[email protected](T.R.)
5 UniversitàdegliStudidiNapoliFedericoII,Dip.diIngegneriaElettricaedelleTecnologie
dell’Informazione,viaClaudio21,Napoli80125,Italy;[email protected](A.I.);
[email protected](D.R.);[email protected](G.R.)
6 IndustrialSystemsInstitute—ISI/R.C.“ATHENA”,PSPBldg.,StadiouStrt.,Platani,Patras26504,Greece;
[email protected](C.K.);[email protected](G.P.)
7 PolitecnicodiTorino,dip.ElettronicaeTelecomunicazioni,corsoDucadegliAbruzzi24,Torino10129,Italy;
[email protected]
8 MétodosyTecnologíadeSistemasyProcesosS.L.C/SantaLeonor65,Edif.C,Pl.4ª,Madrid28037,Spain;
[email protected](L.R.);[email protected](D.R.)
* Correspondence:[email protected];Tel.:[phone];Fax:[phone]
† Theseauthorscontributedequallytothiswork.
AcademicEditor:DavideBrunelli
Received:3March2016;Accepted:23May2016;Published:2June2016
Abstract: InthisarticlewepresentthemainresultsobtainedintheARTEMIS-JUWSN-DPCMproject
betweenOctober2011andSeptember2015. Thefirstobjectiveoftheprojectwasthedevelopment
of an integrated toolset for Wireless sensor networks (WSN) application planning, development,
commissioningandmaintenance,whichaimstosupportapplicationdomainexperts,withlimited
WSN expertise, to efficiently develop WSN applications from planning to lifetime maintenance.
The toolset is made of three main tools: one for planning, one for application development and
simulation(whichcanincludehardwarenodes),andonefornetworkcommissioningandlifetime
maintenance. The tools are integrated in a single platform which promotes software reuse by
automaticallyselectingsuitablelibrarycomponentsforapplicationsynthesisandtheabstractionof
theunderlyingarchitecturethroughtheuseofamiddlewarelayer. Thesecondobjectiveoftheproject
wastotesttheeffectivenessofthetoolsetforthedevelopmentoftwocasestudiesindifferentdomains,
onefordetectingtheoccupancystateofparkinglotsandoneformonitoringairconcentrationof
harmfulgassesnearanindustrialsite.
Sensors2016,16,804;doi:10.3390/s16060804 www.mdpi.com/journal/sensors
Sensors 2016, 16, 804 2 of 40
Sensors2016,16,804 2of40
Keywords: wireless sensor networks; simulation; Internet of Things; design flow; assisted
deployment
Keywords:wirelesssensornetworks;simulation;InternetofThings;designflow;assisteddeployment
1. Introduction
1. Introduction
Wireless sensor networks (WSN) are employed in a wide range of applications and application
Wirelesssensornetworks(WSN)areemployedinawiderangeofapplicationsandapplication
domains, and are the subject of considerable research efforts and technological advances to expand
domains,andarethesubjectofconsiderableresearcheffortsandtechnologicaladvancestoexpandtheir
their applicability. WSNs were among the enabling technologies of the Internet of Things (IoT)
applicability. WSNswereamongtheenablingtechnologiesoftheInternetofThings(IoT)paradigm
paradigm since it was coined more than 15 years ago [1], and consolidated and expanded their
since it was coined more than 15 years ago [1], and consolidated and expanded their application
application domains as the IoT paradigm evolved.
domainsastheIoTparadigmevolved.
However, the spread of WSN-based solutions is still limited by several factors, such as technical
However,thespreadofWSN-basedsolutionsisstilllimitedbyseveralfactors,suchastechnical
complexity, perceived reduced reliability and overall cost. The development of efficient WSN
complexity, perceived reduced reliability and overall cost. The development of efficient WSN
applications requires different and complementary views, but competency separation between the
applicationsrequiresdifferentandcomplementaryviews,butcompetencyseparationbetweenthe
typical stakeholders and the engineering disciplines concurring in the WSN domain is not yet mature,
typicalstakeholdersandtheengineeringdisciplinesconcurringintheWSNdomainisnotyetmature,
which leads to significant inefficiencies. For instance, WSN design and planning can be effort- and
whichleadstosignificantinefficiencies. Forinstance,WSNdesignandplanningcanbeeffort-and
risk-intensive, because they require detailed consideration of connectivity, coverage, cost, network
risk-intensive,becausetheyrequiredetailedconsiderationofconnectivity,coverage,cost,network
longevity, and service quality. In addition, predicting WSN performance before its real deployment
longevity,andservicequality. Inaddition,predictingWSNperformancebeforeitsrealdeploymentis
is very challenging, which leads to costly trial-and-error procedures.
verychallenging,whichleadstocostlytrial-and-errorprocedures.
ThTeh aeiamim ofo tfhthe eAARRTTEEMMIISS--JJUU ffuunnddeedd pprroojejecctt“ “WWSNSND Depelpolyomyemnet,nPtl, aPnlnainnngi,nagn,d aCndom Cmomissmioinsisniogn&ing
& MMaainintetennaanncec”e”( W(WSNS-ND-PDCPMC)Mp)r epsernesteedntiendt hiins atrhtiisc learwtiacslet owbausil dtoa nbuinitledg raante dinptelagtrfaotremd tpolsautfpopromrt to
sutphpeosryt sttheem siynstteegmra itnotresgtroadtoerssig tno, ddeesvieglno,p d,etevset,lodpe,p tleosyt,a dnedpmloayi natnadin mWaSinNtaaipnp WlicSaNtio anpsp.lTichaetipornosje. cTthe
proacjekcnt oawcklendogweldedfrgoemd ftrhoemb etghien bneinggintnhiantgt htheatte tchhen ticeaclhwniocrakl wofosryks otef msysintetemg rianttoergsraretoliress rheeliaevsi lhyeaovnily
one xeixsitsitnigngh ahradrwdawrearaen dansodf tswoaftrwecaorme pcoomnepnotsnaenndtss ahnoudl dshfoocuulds ofnocaupsp loicna taiopnp-lsipcaetciiofinc-psapretcs,ifaics sphaorwtsn, as
shoinwFnig iunr eFi1g.ure 1.

FigFuigrue r1e. 1W.WSNSN-D-DPCPCMM totooolsleset thheelplpss ssyysstteemm iinntteeggrraattoorrss ttoo ffooccuusso onnc cuustsotommererr erqeuqiureirmemenetnstbsy bhyi dhiindging
momsto ssttrsutrcutuctruarla alnadnd imimpplelemmeenntatattiioonn ddeettaaiillss ooff WWSSNN ppllaattffoorrmmc coommppoonnenetnst.s.
Sensors2016,16,804 3of40
ThustheprimarydesignobjectivesoftheWSN-DPCMplatformwere:
‚ Amodulardesignwithcomponentsthatperformwell-definedtasksandcommunicatethrough
well-definedinterfacesanddatamodels,inordertosimplifythemaintenanceandevolutionof
theplatforminlinewiththefastevolutionpaceofWSNresearchandtechnology;
‚ Tobeaccessibletoapplicationdomainexpertswhileallowingexperienceddeveloperstooptimize
thedesignsasmuchaspossible;
‚ ToreuseopenorclosedsourceIPblocksinordertoincreasethereliabilityofWSNprojects,to
reducethedevelopmenteffort,andtofostercommercialexploitationoftheplatform;
‚ Toallowthedeveloperstofocusonapplication-specificrequirementsandimplementationofthe
WSNapplicationbyreducingtheeffortneededtoselectandadapttheexistingcomponentstothe
application,aswellastoreducetheefforttoplan,commission,troubleshootandmaintainthe
WSNduringitslifetime.
Inordertomeettheseaims,theWSN-DPCMplatformincludesseveralusefulcapabilities,such
asradiofrequency(RF)simulation,networksimulationwithhardware-in-the-loop(HiL)capabilities,
topologyoptimization,automaticsoftwaresynthesis,monitoring,debugging,asdescribedintherest
ofthearticle.
TheeffectivenessoftheWSN-DPCMtoolsetistestedintwocasestudiesindifferentdomains.
ThefirstisaWSN-basedsystemtomonitortheoccupancystateofparkinglots,ofwhichaprototype
hasbeentestedintheparkingoftheUniversidadPolitécnicadeMadrid(UPM),Spain[2,3].Thesecond
casestudyisaWSN-basedsolutionformonitoringairconcentrationofharmfulgasesnearanindustrial
site,ofwhichaprototypehasbeentestedattheINTECSpremisesnearPisa,Italy[4,5]. Notethatboth
casestudiesrefertooutdoordeploymentscenarios. Infact,theproposedtoolkitisprimarilyintended
foroutdoorWSNapplicationsespeciallybecausetheraytracing-basedRFsimulationiseffectiveto
simulateoutdoorelectromagneticfieldpropagation. Forindoorenvironmentsasimpleheuristicpath
lossone-slopemodelmaybeused,butwithsignificantlylowerpredictionaccuracy[6].
Therestofthepaperisorganizedasfollows:Section2brieflyreviewsthestateoftheart. Section3
presentstheWSN-DPCMtoolsetarchitectureandthetypicalapplicationdesignflow. Section4details
the main tools of the platform. In Section 5 the two use cases of the project are presented, and
concludingconsiderationsaremadeinSection6.
2. RelatedWork
ThemainadvancementoftheWSN-DPCMplatformwithrespecttothestateoftheartisthatit
providesacompleteintegratedtoolsetforWSNdesignandmanagement,whichisahighlydesired
feature[7]. ExistingWSNframeworkssuchasProFuNTG[8]orGOAT[9]provideafullsetoftools
tosupportnetworksimulation, application-logicdesignandnetworkmonitoring. However, their
featuresandscopearefocusedonspecificsegmentsofnetworkdevelopmentandmaintenance,and
donotaggregateorenablesupportforthewholeWSNtoolchain,fromdevelopmenttodeployment
andmaintenance.
AlmosteveryWSNframeworkandoperatingsystemprovidesintegratedsimulationsupport.
RecentsurveysofsimulationtoolsforWSN[10–13]listdozensoftools,includinginstruction-level
simulators for popular microprocessor families, RF connectivity simulators, mobility simulators,
protocol-levelsimulators,etc. Operatingsystem-basedsimulators,likeTOSSIM[14]andCOOJA[15],
provide good support for code written for the respective platforms, and are often coupled with
hardware-levelemulatorsforpopulararchitectures,forinstruction-levelsimulation. Suchtoolsare
best suited to develop and test low-level components, such as routing protocols. Several of the
availabletools,suchasCastalia[16],MiXiM[17]andNesCT[18]areextensionsbuiltontopofthe
general-purposediscrete-eventsimulatorOMNeT++[19]. Thesetoolsprovidehigher-levelabstraction
oftheWSNandingeneralcanscaletolargernetworksandlongersimulatedperiods.
Sensors2016,16,804 4of40
AnimportantoriginalfeatureoftheWSN-DPCMtoolsetisasite-specificRFsimulationusinga
newraytracing-basedRFsimulatorthatiscustomizedforWSNapplicationsandconsidersindetail
thesurroundingenvironmentandmainobstacles. Manywirelessnetworksimulationtools,suchas
OMNeT++[19]andTOSSIM[14],currentlyrelyonverysimple,heuristicpropagationmodelsthatdo
notaccountforthedetailsofthesurroundingenvironment. Ontheotherhand,theelectromagnetic
propagationpredictiontoolsthataccountforcomplexoutdoorenvironment[20–23]aretailoredfor
radioandtelevisionbroadcasting,cellulartelephonysystemsorWi-Fi,andhavenotbeenemployed
forWSNplanning.
Several high-level programming abstractions and design automation have been proposed to
improveapplicationdevelopmentproductivity. Shimizuetal.[24]proposeapplicationdecomposition
atnetwork-,group-andnode-levelandaDomain-SpecificModelingLanguageoptimizedforeach
level. Application development using this model is restricted mainly to optimization of several
parameters. REMORA [25] is an advanced component-based design framework that uses C-like
behavioral models wrapped in an XML-based abstraction of, e.g., services, interfaces and events.
Therun-timeoverheadislow,buttheapproachcannothandlearbitrarybehavioralcodeorhardware
specifications. Songetal.[26]demonstratedamodel-basedhardware-softwareco-designframework
basedonwidelyusedtoolslikeMathworksSimulink[27]andStateflow[28].Theyprovideahigh-level
graphicalapplicationentrybasedonabstractconcurrentmodels,node-levelsimulation,automatic
generation of network simulation models (including hardware-in-the-loop) and implementation
models for popular embedded OSs. However, the developer is required to manually define and
connectallmodulesthatcomposetheapplication.
InHiLtesting,theprogramundertestrunsonthephysicalsensornodeswithsomeassistant
middleware. Forexample,TOSSIMsimulatesthesensornetworkbyreplacinglow-levelcomponents
andintroducingadiscreteeventqueue.DeveloperscantesttheprogrambeforetheTinyOSapplication
isdeployed.However,TOSSIMcannotrevealifthelengthofamessageissettobelessthanitsintended
sizeoritcannotrevealproblemsstemmingfromthedurationoftasks,sincethetasksinTOSSIMare
executedinstantaneously. Moreover,ithasnotageneralmechanisminordertoaccuratelyestimate
thepowerconsumptionofthesensornodes. In[29]asolutionaimingtoalleviatetherequirementthat
notallTinyOSprogramsrunonphysicalnodesisproposed,becausethatincreasescostsexcessively
whilebeinginconvenient,sotheyallowonlyonephysicalnodetoexistwhichcanbeconfiguredtobe
aneighborofanyvirtualnode. Thus,thepotentialfaultswhichcannotbecapturedbypuresimulation
testing tools can be captured through the interaction between the physical node and other virtual
nodes. AnotheradvantageoftheirworkisthatthepowerconsumptionofanodeinalargeWSNcan
beestimatedthroughtheuseoflow-costdigitalmultimeters. Someconsiderationsnottakenunder
accountonthatworkincludethesignalgainsthatcurrentlyaredesignatedbytheuserandthefact
thatpowerconsumptionestimationmaybeimproved.
In [30] a hybrid simulation framework for WSN application development is proposed, that
interconnectsavirtualnetworkwithaphysicalnetworkandthenallowsonetosimulatethenetworkas
awhole. Theauthorscreatedanovelmodel-basedhybridsimulationframeworkforWSNapplications,
introducingthenotionofhybridsimulation. TheyexploitHiLextensionsinordertolinksomeof
the simulated WSN nodes with hardware-dependent features. Thus, by using Hy-Sim users can
buildahybridnetworkconsistingofvirtualandrealnodesandthensimulateitasawhole. Actually,
Hy-Simsupportstheexistenceofpurelyvirtualnodes,ofpurelyphysicalnodes,butalsoofpartially
simulatednodes. Thelattercanhaveaccesstorealsensorsinordertocompletepartoftheiroperations.
Moreover,Hy-Simallowsnodestohaveaccesstoarealchannelandevenhaveaccesstodatacoming
fromrealsensors.
Although there have been contributions in the literature regarding the optimization of the
wireless sensor networks from the point of view of planning and simulation technologies, there
isstillashortageofmethodologiesandframeworkstocarryoutthein-fielddeployment,configuration
and performance assessment of the overall system application. In [31], the authors proposed an
Sensors2016,16,804 5of40
embedded initialization and maintenance mechanism in which a flooding-based synchronization
strategyisadoptedtoconfigurediscoveryandexpirationtimewindowsintowhichnodescanperform
neighborsearchandlinkassessment. Operationalmodesaretriggeredbydedicatedpointsofthe
network,suchassinkandrootnodes,fromwhichthemodechangingprocessisrealized. Inorderto
enhancethefloodingprocessandwake-upcalls,techniquessuchasTrickle[32]arealsoconsidered.
Another maintenance system to be embedded into the sensor nodes has been proposed in [33],
whichintendstoprovidemaintenanceservicesduringtheoperationalstageoftheWSNbydefining
systemcoherencystatepolicies. Suchmaintenancepoliciesmainlyrelyontriggeringconditionswhen
encounteringsensingfailuresandcommunicationcoveragefailures,particularlyconsideringprotocol
coherent states, which are organized as Global view of coherent states (system-wise, such as time
synchronizationrequirements),andLocalviewofcoherentstate(node-wise,suchasneighborhood
awareness). Themaintenanceenginecollectsthecoherentrequirementsandstates,andthenfloods
the list of protocols for the maintenance service to the network. Other works, such as the one
proposed in [34], make use of Deployment Support Networks (DSN), which are wireless network
temporarilyinstalledalongsidetheactualWSNduringthedeploymentprocess.Themaintargetofthis
infrastructureistoprovidetwodifferentradiocommunicationcapabilities,oneforpassivelistening
ofnetworktrafficandonefordebuggingchannelconnectionsthroughDSNnodestodistributethe
gatheredinformationtoanearbysinkpoint,fromwhichthedeploymentanalysiswillbeperformed.
Therefore, such a debugging implementation relies on a passive monitoring system based on the
deploymentofadditionalnetworkplatforms.
3. WSN-DPCMArchitectureandDesignFlow
The WSN-DPCM toolset flow was designed to reduce the application development and
deploymenteffortandthetime-to-marketfornewWSNapplications. Itenablessoftware-assisted
engineeringandprovidesmostofitsintegratedcapabilitiesasservicesovertheInternet.
ItsoperationflowiscorrelatedwiththeDPCMdatamodel,whichholdsallinformationitems
usedbythedifferenttoolsofthePlatformasinput/output,fromthecreationofanewProjecttoonline
monitoringofthedesignedWSNnetworkinareal-lifeenvironment. Modelentitiescanbecreated,
modifiedordeletedthroughoutatoolsetsession. ThehierarchyofthemodelisshowninFigure2.
Itcomprisesthefollowingmainentitytypes:
‚ PROJECT—projectdescriptionandattributes. ItincludesreferencestotheUSERandthelistof
PLANentitiesdefinedfortheProject.
‚ PLAN—deployment plan definition and attributes. It includes references to SIM entities as a
resultofthesimulationsexecutedovertheplanspecification,theCONNECTIVITYmatrixentity
as attachment and all references to the definition entities of the node types considered by a
particulardeployment.
‚ USER—contactinformationandcredentialsoftheproject’suser.
‚ CONNECTIVITY MATRIX—structured file in JSON format that contains the communication
propertiesandtransmissionvaluesestimatedbythesimulationoftheradiolinksbetweeneach
pairofnodes.
‚ NIDDEF—typedefinitionentityfornetworkinterfacedevices(NID,e.g.,gateways). Itcontains
all details that characterize a NID type structurally (type identifier, nature, functional blocks,
parametersetc.) andfunctionally(properties,configurationchoices,etc.). Thisspecificationis
correlatedwiththeembeddedcodegeneratedforthisparticulartypeofnode.
‚ ROOTDEF—typedefinitionentityforsinknodedevices, similartotheNIDDEFentity. Itcan
containnodeprogrammingsoftwarewhenthecodesynthesisphaseiscompleted.
‚ NODEDEF—type definition entity for regular leaf node devices, similar to the NIDDEF and
ROOTDEFentities. Itcancontainnodeprogrammingsoftwarewhenthecodesynthesisphase
iscompleted.
‚ SIM—RF simulation entity that contain its configuration attributes and results for the
targetednodes.
Sensors2016,16,804 6of40
Sensors 2016, 16, 804 6 of 40
Sensors 2016, 16, 804 6 of 40

FFiigguurree 22.. DDiiaaggrraamm ooff tthhee ddaattaa mmooddeell hhiieerraarrcchhyy aanndd tthhee mmuullttiipplliicciittiieess aammoonngg iittss eennttiittiieess..
Figure 2. Diagram of the data model hierarchy and the multiplicities among its entities.
The platform is based on three main tools, the Planning Tool, the Development Tool and the
TThhee ppllaattffoorrmm iiss bbaasseedd oonn tthhrreeee mmaaiinn ttoooollss,, tthhee PPllaannnniinngg TTooooll,, tthhee DDeevveellooppmmeenntt TTooooll aanndd tthhee
Commissioning & Maintenance Tool, and supporting components, Library of components and the
CCoommmmiissssiioonniinngg && MMaaiinntteennaannccee TTooooll,, aanndd ssuuppppoorrttiinngg ccoommppoonneennttss,, LLiibbrraarryy ooff ccoommppoonneennttss aanndd tthhee
Middleware, each consisting of integrated services that are accessible through a local/web GUI
MMiiddddlleewwaarree,, eeaacchh ccoonnssiissttiinngg ooff iinntteeggrraatteedd sseerrvviicceess tthhaatt aarree aacccceessssiibbllee tthhrroouugghh aa loloccaall//wweebb GGUUII
(WebTop Environment) and/or RESTful services (Figure 3).
((WWeebbTToopp EEnnvviirroonnmmeenntt))a anndd//oorr RREESSTTffuull sseerrvviicceess ((FFiigguurree 33))..

Figure 3. Overview of the WSN-DPCM Toolset communication and typical development flow.
Figure 3. Overview of the WSN-DPCM Toolset communication and typical development flow.
Figure3.OverviewoftheWSN-DPCMToolsetcommunicationandtypicaldevelopmentflow.
The WebTop Environment is a web desktop application that requires no local installation and
The WebTop Environment is a web desktop application that requires no local installation and
acts as a unique entry point to the toolset through user authentication. It provides the main user
TheWebTopEnvironmentisawebdesktopapplicationthatrequiresnolocalinstallationandacts
acts as a unique entry point to the toolset through user authentication. It provides the main user
interface to the WSN-DPCM toolset, all visualization support needed by its tools and other
asauniqueentrypointtothetoolsetthroughuserauthentication. Itprovidesthemainuserinterface
interface to the WSN-DPCM toolset, all visualization support needed by its tools and other
information, as well as the controls needed to advance through the development flow. As shown in
totheWSN-DPCMtoolset,allvisualizationsupportneededbyitstoolsandotherinformation,aswell
information, as well as the controls needed to advance through the development flow. As shown in
Figure 3, it has control interfaces with each of the main platform tools in order to control their
asthecontrolsneededtoadvancethroughthedevelopmentflow. AsshowninFigure3,ithascontrol
Figure 3, it has control interfaces with each of the main platform tools in order to control their
execution (e.g., start, stop, status) and data interfaces with the Project Repository of the platform in
interfaceswitheachofthemainplatformtoolsinordertocontroltheirexecution(e.g., start, stop,
execution (e.g., start, stop, status) and data interfaces with the Project Repository of the platform in
order to check the advance status of the various development stages, as reported by the tools.
status)anddatainterfaceswiththeProjectRepositoryoftheplatforminordertochecktheadvance
order to check the advance status of the various development stages, as reported by the tools.
The Planning Tool (PT) provides the developer with a graphical input interface for the
statusofthevariousdevelopmentstages,asreportedbythetools.
The Planning Tool (PT) provides the developer with a graphical input interface for the
deployment field map, for the definition of WSN node types and their positions in the field, and for
ThePlanningTool(PT)providesthedeveloperwithagraphicalinputinterfaceforthedeployment
deployment field map, for the definition of WSN node types and their positions in the field, and for
other platform tools that need to analyze various aspects of node deployment (e.g., RF propagation,
fieldmap,forthedefinitionofWSNnodetypesandtheirpositionsinthefield,andforotherplatform
other platform tools that need to analyze various aspects of node deployment (e.g., RF propagation,
signal strength and coverage strength, optimal topology and connectivity). It has a control interface
toolsthatneedtoanalyzevariousaspectsofnodedeployment(e.g.,RFpropagation,signalstrength
signal strength and coverage strength, optimal topology and connectivity). It has a control interface
with the WebTop Environment and a data interface with the Project Repository of the platform to
andcoveragestrength,optimaltopologyandconnectivity). IthasacontrolinterfacewiththeWebTop
with the WebTop Environment and a data interface with the Project Repository of the platform to
save new projects or retrieve older ones.
EnvironmentandadatainterfacewiththeProjectRepositoryoftheplatformtosavenewprojectsor
save new projects or retrieve older ones.
retrieveolderones.
Sensors2016,16,804 7of40
AnimportantPTanalysisisradiocoverageofthenodesbasedontheirpositionsintheapplication
field,theirradiocharacteristicsandfieldfeatures. Thisanalysisiscontrolledfromtheuserinterface
ofthePlanningTool. Theresultsaregraphicallyoverlaidontheapplicationfieldmapanddetailed
informationisalsoavailableintextualformforeachnodeinthefield. Thetoolusestheresultsofradio
simulationtocalculatethenodeconnectivitymatrix,whichissavedintheProjectRepositoryforlater
usebytheNetworkSimulatorwithintheDevelopmentTool.
ThedeploymentplansdesignedbythePTtoolareliveelementsthatcanbeadaptedandrefined
usingthefeedbackfromsimulationanalysisorthedeploymentreportsproducedbyPlatformtools
andsystems. Ingeneral,theDPCMtoolsetconsidersaniterativeapproachtosolutionprototyping
using the proposed work flow (Figure 3) through sequences of planning, simulation, synthesis,
deployment, monitoring activities, by updating the data models to the new environment and
prototyperequirements.
TheDevelopmentTool(DT)allowsthedevelopertoproducetheapplication-specificsoftware
to program the WSN nodes and to simulate their behavior in the field using network simulation,
eitherpurelyinsoftwareorincludingrealhardwarenodes(hardware-in-the-loopsimulation: HiL)to
increaseresultaccuracy. HiLisatechniqueincreasinglyconsideredinthedevelopmentandtestingof
embeddedsystems,wherethemaingoalistotesttheembeddedsystemitselfinsteadofanabstract
model. The complexity of the real plant under control is embodied in the simulation, while the
designedsystemmustbetestedusingrealisticsimulationconditionsinavirtualenvironment. HiL
isakindofreal-timesimulationwheretheinputandoutputsignalsofthesimulatorshowthetime
dependentvaluesastheyareintherealprocess.
ThetoolmaintainsacontrolinterfacewiththeWebTopEnvironmentoverwhichitcanbestarted
and can communicate the completion status for its internal development steps (e.g., application
synthesis,networksimulation). IthasalsoadatainterfacewiththeProjectRepositoryoftheplatform
tosavenewprojectsandresults(e.g.,nodeapplicationspecifications,simulationresults)andtoretrieve
olderones.
AsshowninFigure3,DTinterfaceswiththePlanningToolmainlytoextractthelistofrequested
nodetypesandtheirfeaturesfromthenetworkplanssavedbyPTintheProjectRepositoryorthenode
connectivitymatrix. Foreachnodetype,DTautomaticallycreatesadevelopmentprojectskeletonfor
whichitprovidesseveralnode-levelapplicationdevelopmentflows. Forapplicationdomainexperts,
thetooloffersaguided,graphicalflowbasedonstatecharts,whichrequiresjustbasicprogramming
skills. Nevertheless,experiencedprogrammerscanuseotherinputmethods,whichallowthemafiner
controlovernodeprogramdevelopment. Regardlessoftheinputmethodoftheapplication-specific
businesslogic,DTprovidesautomaticsynthesisofthenecessarysoftwareandhardwaretosupport
theapplicationrequirementsusingalibraryofhardwareandsoftwarecomponents,withorwithout
thesupportofanembeddedoperatingsystemormiddleware. Thesynthesisprocessaimstohide
mostimplementationdetailsthatarenotexplicitlyaddressedbytheapplicationspecification(e.g.,
somelow-levelsensorimplementations,OStasks,interfaces)andtofacilitatesoftwarereuse,allowing
thiswaythedeveloperstofocusonapplication-specificissues.
TheCommissioningandMaintenanceTool(CMT)assiststhedeveloperinthepreparationofthe
nodesforfielddeployment,duringtheactualfielddeploymentoperations,andformonitoringand
debuggingafterdeploymentforlong-termmaintenance.
For node preparation for in-field deployment, the CMT uses the programming images and
configurationssavedintheProjectRepositorybytheDevelopmentToolforallnodes,andthenode
typeandfieldpositionsavedintherepositorybythePlanningTool. Oncedeployedinthefieldand
activated,theactualoperationconditionsofthenodesarecollectedbytheCMTandcanbecompared
withtheresultsoftheradioandnetworksimulationssavedintheProjectRepositorytocorrectand
optimizevariousnetworkparameters. TheCMTisinsyncwiththedeploymentspecification,being
abletoreconfigureitselfautomaticallywheneveranewPlanisgeneratedormodified,orthefirmware
imagesarechanged,reconfiguredorregenerated.
Sensors2016,16,804 8of40
The Middleware (MW) provides an abstraction layer for node resources, software and
network services, and communication capabilities within the network. This layer allows to write
implementation-independentapplicationstoincreasesoftwarereuseanddevelopmentproductivity.
ItcaninterfacewiththeDevelopmentTooltosupportapplicationsynthesisanditsembeddedsupport
isusedatrun-timebytheCMTtomonitortheoperationofthenetwork.
AtypicalapplicationdevelopmentusingtheWSN-DPCMplatformstartsbyauthenticatingthe
developerwiththeplatform,throughtheWebtop-basedUI(seethelowerpartofFigure3). TheUI
helpsthedevelopertofollowtheplatformdevelopmentflowbyactivatingonlythecommandsthat
meetallconditionstobeexecuted. Forinstance,atthebeginofanewprojectonlythePlanningTool
canbeactivatedtoallowthedevelopertoinput,e.g.,thedeploymentfieldmap,thedefinitionofthe
nodetypesandtheirpositionsintheapplicationfield. Afterwards,thedevelopercanrunasimulation
oftheRFpropagationtoestimatethenumberandqualityofRFcommunicationlinks. Ifthesearenot
satisfactory,thedevelopercanrevisitthemapandnodepropertiesandmakethenecessarychangesto
fulfillallqualitycriteria.
Once a field plan is approved and saved in the Project Repository (PR), the UI activates the
nextdevelopmentstepintheflow,theDevelopmentTool. Thistoolcollectsnodedataandnetwork
connectivityfromtheplan.Nodetypesareautomaticallyextractedandforeachofthemthetoolcreates
aprojecttemplatethatcanbeusedbythedevelopertosubsequentlyspecifyallnodecharacteristics
andtheapplicationitneedstorun. Foreachoftheseprojects,thedevelopercanruntheapplication
synthesis engine whose output are software for node programming, hardware specification and
networksimulationmodel. Thelattercanbeusedtogetherwiththeconnectivitymatrixgeneratedby
theRFsimulatorinthePlanningTooltoperformpuresoftwarenetworksimulationormixed,with
hardwarenodesintheloop. Ifthesynthesisorsimulationresultsarenotsuitable,thedevelopercan
iterativelychangethenodespecificationsorthesimulationparameterswithintheDevelopmentTool,
orcangobacktothePlanningTool,e.g.,tochangenetworkcompositionortopology.
WhennodedevelopmentiscompletedandsavedintheProjectRepository,theUIenablesthenext
developmentstep,theCommissioningandMaintenanceTool.Thistoolextractsthenodeprogramming
software,configurationsandhardwarespecificationsfromtherepositoryandusethemtoprogram
deployment-readynodes. Moreover,itusesthefieldmapandnodelocationsfromtheplangenerated
bythePlanningTooltoprogramtheCMTinterfaceforlong-termnetworkmonitoringandtoprogram
theHand-helddevicesthatwillbeusedduringnetworkdeploymentandforin-fielddebugging. Ifthe
datacollectedinthefieldarenotsatisfactory,thedevelopercangobacktothePlanningorDevelopment
toolstomakethenecessaryamendmentstothenetworkplan,ornodeorapplicationspecifications.
Iftheenvironment(i.e.,thescenarioinwhichsensorsaredeployed)changes,theRFsimulation
andtheconnectivitymatrixcanberunagainforallnodesintheupdatedscene. Ifonlysomenodes
changed,thesimulationcanberunjustforthosenodestosavecomputationtime(seeSection4.1.2).
Iftheconnectivitymatrixchanges,thenetworksimulationoftheDevelopmentToolmayneedtobe
runagaintocheckthesuitabilityofthenewnetwork. However,ifnewnodetypesareaddedoreither
thenodeorapplicationspecificationschange,thenitisnecessarytore-runthesynthesisengineofthe
DevelopmentTooltoupdatethesimulationmodels,thenodeconfigurationandprogrammingcode,
andthehardwareconfiguration.
TheCommissioningandMaintenanceToolshouldalsobeupdatedwiththenewplan,software
andhardwarecomponentsinordertocorrectlyinterfaceandinterpretthefielddataduringdeployment
andthelong-termnetworkexploitation.
4. MainToolsoftheWSN-DPCMPlatform
TheoperationofthethreemaintoolsoftheWSN-DPCMplatform(PlanningTool,Development
Tool,andCommissioning&MaintenanceTool,seeFigure3)iscontrolledfromaweb-basedplatform,
whiletheoperationoftheunderlyinghardwareandsoftwarecomponentsandservicesisabstracted
throughaMiddlewarecomponent,alsodevelopedwithintheproject.
Sensors2016,16,804 9of40
Sensors 2016, 16, 804 9 of 40
4.1. Tool for WSN Application Planning and RF Simulation
4.1. ToolforWSNApplicationPlanningandRFSimulation
The Planning Tool consists of:
ThePlanningToolconsistsof:
 An input interface that displays the map of the deployment field and allows WSN node selection
‚ AninputinterfacethatdisplaysthemapofthedeploymentfieldandallowsWSNnodeselection
and positioning,
 aAn sdetp oofs iftuionnctiniogn,al components for RF propagation simulation and node connectivity prediction.
‚ AsetoffunctionalcomponentsforRFpropagationsimulationandnodeconnectivityprediction.
Different types of nodes can be selected from those available in the repository and/or new ones
can bDe idffeefriennetdt ybpye tshoef dneovdeelospcearn. bOensceel escetleedctefrdo,m thteh nosoedaevs acialanb blee ipnlathceedr eopno tshiteo rmyaapn dw/itohr an ecwlicokn oens
tchaen bdeedsierfiedn edpobsyittihoen.d eTvheelo poethr.eOr nfcuenscetiloecntaeld ,ctohmepnoondeenstcsa ninbteegprlaatceedd oinn tthheem PaTp wariteh daecslcicrkiboend thine
tdhees ifroeldlopwoisnitgi.o n. TheotherfunctionalcomponentsintegratedinthePTaredescribedinthefollowing.
44..11..11.. 33DD BBuuiillddiinngg RReeccoonnssttrruuccttiioonn
TToo imimpprorovveet htehsei msiumlautiloantioancc uarcaccuyr,a3cDy, re3cDo nrsetrcuocntsiotrnuoctfiopnh yosifc aplheynsviicraoln menevnitroisnvmeeryntd eiss irvaebrlye.
Wdehsiernabelne.v Wirohnemn eenntvdireosncmripentito dneisscrniopttiaovna iisla nbolet aovraiislatboloe eoxrp ise ntosoiv eex,ptheensdiveve,e tlhoep edresvneeloepdetros nbueeildd tiot
bbuyitlhde imt bsyel tvheesm. selves.
HHoowweevveerr,, ffoorr mmaannyy rardaidoi oprporpoapgaagtiaotnio enxpexeprtesr tthset phreeppraerpatairoant ioofn thoef etnhveireonnvmireonntm meondteml ios dmeolries
cmhoarlleecnhgainllge ntghianng tahnaanlyaznianlgy ztihneg rthadeiroa dpieorpfoerrmfoarnmcaen. cBee.sBideseisd, ems,omsto srterseesaeracrhc hrerseusultlst spprreesseenntteedd iinn
aaccaaddeemmiicc ccoonnfefererenncecsesr eruesuesteh ethsaem seamenev ireonnvmiroenntmtoenatn atloy zaentahleyizrew tohrekira nwdocrakn naontdte sctatnhneorto btuesstt ntehses
raonbduasctncuesrsa caynfdo arcdciuffrearceyn tfoern dviifrfoenrmenetn etns.vironments.
FFoorr ththisisp uprpuorpseo,steh, etWheS NW-DSPNC-MDPtoCoMls ettoinocllsuedt esinaclfuudnectsi ona alfcuonmctpioonnaeln tctohmatpcoanneanutt otmhaatt iccaallny
areuctoonmsatrtuiccatltlyh erethcoirndstdriumcte nthsieo nthfirrodm di2mDeinmsiaogne sfr(oFmig u2rDe 4im).aItgeexs t(eFnidgsurthe e4a).l gIto reixthtemndpsr othpeo saeldgoirnit[h3m5]
panrodpuosseesds uibn- s[a3m5]p lainndg aunsdesra nsudbo-msafmeaptluinregs ealnedct ioranntdecohmn iqfeuaetsufroer isteelreacttiivoenl etaercnhinngiqfuroesm fiomr aigteesr.atTivhee
leesatirmniantged frcoomnfi imdeangceesv. Talhuee eosftiemacahtecdla csosnffoirdeeancche pviaxleulec oafn ebaechre cilnatsesr pforer teeadchas paixperlo cbaanb bileit yredinistterripbruettieodn
auss ian gprsoobfatbmilaitxyt rdainstsrfiobrumtiaotnio unsitnogc aslocfut lmataext etxratunrsefolramyoautitopno ttoe nctailaclus.late texture layout potentials.

Figure 4. Work flow of the 3D outdoor environment reconstruction.
Figure4.Workflowofthe3Doutdoorenvironmentreconstruction.
The work flow of the method is shown in Figure 4. Meaningful object classes are recognized
The work flow of the method is shown in Figure 4. Meaningful object classes are recognized
through a machine learning mechanism, which consists of a training phase followed by an evaluation
throughamachinelearningmechanism,whichconsistsofatrainingphasefollowedbyanevaluation
phase. The training phase uses an image database through which the features of images are learned
phase.Thetrainingphaseusesanimagedatabasethroughwhichthefeaturesofimagesarelearnedand
and discriminated for the concerned types of objects. After an image is properly recognized, the
discriminatedfortheconcernedtypesofobjects. Afteranimageisproperlyrecognized,thealgorithm
algorithm automatically assigns different volumetric information according to scale ratio and
automaticallyassignsdifferentvolumetricinformationaccordingtoscaleratioandempiricalmaterial
empirical material for each object, as it can be seen in Figure 4.
foreachobject,asitcanbeseeninFigure4.
4.1.2. RF Simulation and Connectivity Matrix
4.1.2. RFSimulationandConnectivityMatrix
This component has two main functions within the WSN-DPCM platform: to allow the developer
ThiscomponenthastwomainfunctionswithintheWSN-DPCMplatform: toallowthedeveloper
to check node connectivity for a given placement in the application field and to generate the input data
tochecknodeconnectivityforagivenplacementintheapplicationfieldandtogeneratetheinput
Sensors 2016, 16, 804 10 of 40
Sensors2016,16,804 10of40
for the topology reduction and network simulation components of the WSN-DPCM platform. Its main
features have been recently presented at a conference [36] and are here better detailed.
dataforthetopologyreductionandnetworksimulationcomponentsoftheWSN-DPCMplatform.
The component calculates the received signal strength for each node when each of the other
Itsmainfeatureshavebeenrecentlypresentedataconference[36]andareherebetterdetailed.
nodes in the network is transmitting (one at a time) using an electromagnetic solver, or RF tool. It
Thecomponentcalculatesthereceivedsignalstrengthforeachnodewheneachoftheothernodes
receives as input a digital description of the application scene and the technical specifications of the
inthenetworkistransmitting(oneatatime)usinganelectromagneticsolver,orRFtool. Itreceivesas
transmitting antenna. The former is provided using a file in Keyhole Markup Language (KML)
inputadigitaldescriptionoftheapplicationsceneandthetechnicalspecificationsofthetransmitting
format that describes the buildings, and a raster file that describes the topography of the terrain, i.e.,
antenna.TheformerisprovidedusingafileinKeyholeMarkupLanguage(KML)formatthatdescribes
the Digital Terrain Model (DTM), if available. The format employed for building description
thebuildings, andarasterfilethatdescribesthetopographyoftheterrain, i.e., theDigitalTerrain
simplifies data import from various sources, such as local authorities, Google Earth or Google Maps,
Model(DTM),ifavailable. Theformatemployedforbuildingdescriptionsimplifiesdataimportfrom
or reconstructions from aerial photography using the component described in Section 4.1.1. Relative
varioussources,suchaslocalauthorities,GoogleEarthorGoogleMaps,orreconstructionsfromaerial
permittivity and conductivity of building walls and terrain can be added to improve simulation
photographyusingthecomponentdescribedinSection4.1.1. Relativepermittivityandconductivityof
accuracy, otherwise the simulator uses sensible defaults based on area type (historical, residential,
buildingwallsandterraincanbeaddedtoimprovesimulationaccuracy,otherwisethesimulatoruses
business district). The transmitting antenna is described by its position, radiated power, pointing
sensibledefaultsbasedonareatype(historical,residential,businessdistrict). Thetransmittingantenna
direction, radiation diagram and polarization.
isdescribedbyitsposition,radiatedpower,pointingdirection,radiationdiagramandpolarization.
The solver is based on a 3D space analysis similar to that described in [22,23] using a ray tracing
Thesolverisbasedona3Dspaceanalysissimilartothatdescribedin[22,23]usingaraytracing
algorithm that considers direct, reflected and diffracted rays. In particular, reflections are computed
algorithmthatconsidersdirect,reflectedanddiffractedrays. Inparticular,reflectionsarecomputed
using Geometrical Optics (GO) and diffractions are evaluated using the Uniform Theory of
uDsiifnfrgacGteioonm (eUtrTicDal).O Tphtiec su(sGerO )maanyd adlisfofr ascetlieocnt saa r“efaesvta lmuaotdeed”u tshinagt othnelyU ancifcoorumntTsh feoorr ythoef Ddiirfefrcatc tainodn
(rUefTleDct)e.dT hraeyuss, etro mavaoyidal msoossetl ceoctmap“uftaasttiomnoadl elo”atdh awthoinclhy ias cdcuoue ntots dfiofrfrtahceteddi rreacytsa. ndreflectedrays,to
avoidThmeo rsetscuolmtinpgu etaleticotrnoamllaogandewtich fiicehldis isd pureotvoiddeifdf roanc treedguralayrs .2D grids (“layers”) located on surfaces
at weTllh-deerfeisnueldti nhgeieglhectstr aobmoavgen tehteic gfireoludnids p(orro vabidoevde othner erogoufl,a irf 2tDheg grridids p(“olianyte irss ”o)nl otocpat eodf ao nbusiuldrfiancge)s.
aAtcwcoerldl-idnegfilyn,e da h3eDig mhtaspa bios voebtthaeingerdo,u wndhi(cohr asbhoovwest htheer ofoiefl,dif ltehveeglsr idprpoodiuncteids obnyt othpeo ftraanbsumildititningg).
Aanctceonrndain ginly ,thae3 Dcomnsaipdeirseodb tsacienneed., Twhhei cmhashpo iws sstthoerefide lidn lgeveoeltsifpf rfoodrmucaetd; bity ctahne tbrea ndsimspitltaiynegda nbtye ntnhae
iWnethbTeocpon ussidere riendtesrcfaencee .(GThUeI)m aanpd isopsttoiornedalilny gsueopteirffimfopromseadt; iotnc aan Gboeodgilsep Elaayrethd bimyatghee W(ase bsThoopwuns ienr
iFnigteurrfaec 5e)(, GaUndI)/oarn dit ocpanti obnea lplyasssuepde troim thpeo sceodnonnecatiGviotyo gmleaEtrairxt hmiomdauglee.( aTshseh loawttenri,n bFaisgeudr eon5 )t,haen dfi/eoldr
imtcaapns pberopdauscseedd btoy tthhee ceolencntreocmtivaigtynemtiac tsroixlvmero, dcoumle.pTuhtees laanttde rs,tboareses dthoen fitehled fiveallduems arpadsiparteodd ubcye edabchy
tahneteenlencat r(onmodagen) eatti cthseo llvoecra,tcioomnsp ouft easlla ontdhesrto arnestetnhneafise (lndovdaelus)e sinr aad “iactoendnbeycteivaicthy amnatetrninxa”.( nTohdise )maatttrhixe
lcoacna tbioen esaosiflyal lcootnhveerratendte innn aas b(innoadreys )minatarix“c iofn rneeccetiivveirty fimeladt rtihx”re.sThhoilsdm leavtreilxs caarne bdeeefianseildy. cNonovne-zreterdo
ienleambeinntasr yofm thater imxaiftrriexc eciavne rbfiee dldistphlraeysehdo lbdyl ethveel sGaUreI adse fiarncesd c.oNnonne-czteinrog ethleem ceonrtrsesopfothnedminagt rnioxdceasn, baes
dshisopwlany iend Fbiyguthree 6G. UIasarcsconnectingthecorrespondingnodes,asshowninFigure6.

(a) (b)
FFiigguurree 55.. 22DD ggrraapphhiiccaall ddiissppllaayy ooff tthhee ccoommppuutteedd ffiieelldd rraaddiiaatteedd bbyy oonnee ooff tthhee nnooddeess ((aa)) tthhaatt ccaann bbee
ssuuppeerriimmppoosseedd ttoo GGooooggllee EEaarrtthh iimmaaggee ooff tthhee aarreeaa ((bb))..