<%BANNER%>

Wireless Sensor Network for HVAC Control

Permanent Link: http://ufdc.ufl.edu/UFE0045567/00001

Material Information

Title: Wireless Sensor Network for HVAC Control
Physical Description: 1 online resource (53 p.)
Language: english
Creator: Subramany, Rahul
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: co2sensor -- environmentalsensor -- hvaccontrol -- motiondetection -- vav -- wirelesssensors
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Advanced control algorithms when used in building HVAC systems have a large potential of saving energy. Implementing such control techniques require the addition of sensor/sensing systems in all conditioned rooms within a building. This thesis attempts to develop a wireless sensor network that can provide the sensor data required by these advanced control algorithms. The proposed system will be low cost, easy to fabricate and can be easily deployed in existing buildings. The design and development of the wireless sensor network is discussed first. The wireless sensor node that has all the sensors required for advanced building HVAC control is developed first. The wireless sensor node has a microcontroller that collects sensor data, processes it and communicates it using a wireless transmitter. This data is received at another wireless receiver. The wireless receiver is attached to a microprocessor that communicates the received data to a plug computer. The plug computer time stamps the data and saves it to a database. The control algorithms can access this data when required. The developed wireless sensor network is tested and its scalability is studied.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Rahul Subramany.
Thesis: Thesis (M.S.)--University of Florida, 2013.
Local: Adviser: Barooah, Prabir.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045567:00001

Permanent Link: http://ufdc.ufl.edu/UFE0045567/00001

Material Information

Title: Wireless Sensor Network for HVAC Control
Physical Description: 1 online resource (53 p.)
Language: english
Creator: Subramany, Rahul
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: co2sensor -- environmentalsensor -- hvaccontrol -- motiondetection -- vav -- wirelesssensors
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Advanced control algorithms when used in building HVAC systems have a large potential of saving energy. Implementing such control techniques require the addition of sensor/sensing systems in all conditioned rooms within a building. This thesis attempts to develop a wireless sensor network that can provide the sensor data required by these advanced control algorithms. The proposed system will be low cost, easy to fabricate and can be easily deployed in existing buildings. The design and development of the wireless sensor network is discussed first. The wireless sensor node that has all the sensors required for advanced building HVAC control is developed first. The wireless sensor node has a microcontroller that collects sensor data, processes it and communicates it using a wireless transmitter. This data is received at another wireless receiver. The wireless receiver is attached to a microprocessor that communicates the received data to a plug computer. The plug computer time stamps the data and saves it to a database. The control algorithms can access this data when required. The developed wireless sensor network is tested and its scalability is studied.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Rahul Subramany.
Thesis: Thesis (M.S.)--University of Florida, 2013.
Local: Adviser: Barooah, Prabir.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045567:00001


This item has the following downloads:


Full Text

PAGE 1

WIRELESSSENSORNETWORKFORHVACCONTROLByRAHULSUBRAMANYATHESISPRESENTEDTOTHEGRADUATESCHOOLOFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENTOFTHEREQUIREMENTSFORTHEDEGREEOFMASTEROFSCIENCEUNIVERSITYOFFLORIDA2013

PAGE 2

c2013RahulSubramany 2

PAGE 3

Tomyparents 3

PAGE 4

ACKNOWLEDGMENTS IwouldliketoexpressmysinceregratitudetomyadviserDr.PrabirBarooahforguidingmethroughthecourseofmyresearch.HehasbeenagreatsourceofsupportandencouragementsincethedayIstartedworkingwithhim.Hehasalwaysbelievedinmeandhashelpedmebecomeanindependentthinker.AspecialthankstoDr.TimothyMiddelkoop,forprovidingkeyadvise.IthankmycolleaguesSiddarthGoyal,YashenLinandChendaLiaofortheirvaluabletime.Iheartilythankmyparents,mysisterandmyfriendsfortheirlove,supportandencouragement. 4

PAGE 5

TABLEOFCONTENTS page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 7 LISTOFFIGURES .................................... 8 ABSTRACT ........................................ 9 CHAPTER 1INTRODUCTION .................................. 10 1.1Motivation .................................... 10 1.2SystemRequirement .............................. 11 1.3FinalSystemLayout .............................. 15 2WIRELESSSENSORNODE ............................ 16 2.1HardwareSelection ............................... 16 2.1.1MicroprocessorandRadio ....................... 16 2.1.2Sensors .................................. 18 2.1.3CO2Sensor ............................... 18 2.1.4HumiditySensor ............................. 19 2.1.5PIRSensor ................................ 20 2.1.6TemperatureSensor ........................... 20 2.2DetailedDesign ................................. 20 2.2.1Prototype ................................ 20 2.2.2PowerSupply .............................. 21 2.2.3PCBDesign ............................... 22 2.2.4CaseDesign ............................... 23 2.3SoftwarefortheEndPoint ........................... 24 3ACCESSPOINT ................................... 33 3.1Hardware .................................... 33 3.2Software ..................................... 33 3.2.1sJoinSemBranch ............................ 35 3.2.2sPeerFrameSemBranch ......................... 36 3.2.3WritingtotheSerialPort ........................ 37 4TESTINGANDVALIDATION ........................... 39 4.1BaseStation ................................... 40 4.2Testing ...................................... 41 4.3Deployment ................................... 45 4.4CostAnalysis .................................. 45 5

PAGE 6

4.5Shortcomings .................................. 46 5CONCLUSIONANDFUTUREWORK ...................... 49 5.1Summary .................................... 49 5.2FutureWork ................................... 49 REFERENCES ....................................... 51 BIOGRAPHICALSKETCH ................................ 53 6

PAGE 7

LISTOFTABLES Table page 2-1ComparisonofArduino+ZigbeeandeZ430-Rf2500 ................. 17 2-2ComparisonofCOZIRandK-30Sensors ...................... 19 2-3Powerrequirementsforvariousnodecomponents ................. 22 2-4InstructionsetforCO2sensor. ............................ 30 4-1WirelessSensorNodecost .............................. 47 5-1WirelessSensorNodeVersion2 ........................... 49 7

PAGE 8

LISTOFFIGURES Figure page 1-1SystemOverview ................................... 14 1-2WirelessSensorNetwork ............................... 15 2-1TexasInstrumentseZ430-RF2500 .......................... 18 2-2Prototype ....................................... 21 2-3PrintedCircuitBoardSchematic .......................... 23 2-4Case-CADDrawing .................................. 24 2-5CaseandCover .................................... 25 2-6EndPointowchart ................................. 26 3-1AccessPointowchart ................................ 33 4-1WirelessSensorNode-Deployedonawall ..................... 39 4-2BaseStation ...................................... 41 4-3Deploymentplan:PughHallFloor2 ........................ 46 8

PAGE 9

AbstractofThesisPresentedtotheGraduateSchooloftheUniversityofFloridainPartialFulllmentoftheRequirementsfortheDegreeofMasterofScienceWIRELESSSENSORNETWORKFORHVACCONTROLByRahulSubramanyMay2013Chair:PrabirBarooahMajor:MechanicalEngineering AdvancedcontrolalgorithmswhenusedinbuildingHVACsystemshavealargepotentialofsavingenergy.Implementingsuchcontroltechniquesrequiretheadditionofsensor/sensingsystemsinallconditionedroomswithinabuilding.Thisthesisattemptstodevelopawirelesssensornetworkthatcanprovidethesensordatarequiredbytheseadvancedcontrolalgorithms.Theproposedsystemwillbelowcost,easytofabricateandcanbeeasilydeployedinexistingbuildings. Thedesignanddevelopmentofthewirelesssensornetworkisdiscussedrst.ThewirelesssensornodethathasallthesensorsrequiredforadvancedbuildingHVACcontrolisdevelopedrst.Thewirelesssensornodehasamicrocontrollerthatcollectssensordata,processesitandcommunicatesitusingawirelesstransmitter.Thisdataisreceivedatanotherwirelessreceiver. Thewirelessreceiverisattachedtoamicroprocessorthatcommunicatesthereceiveddatatoaplugcomputer.Theplugcomputertimestampsthedataandsavesittoadatabase.Thecontrolalgorithmscanaccessthisdatawhenrequired. Thedevelopedwirelesssensornetworkistestedanditsscalabilityisstudied. 9

PAGE 10

CHAPTER1INTRODUCTION 1.1Motivation Buildingsconsumenearly34%ofallenergyusedintheUnitedStates.Heating,VentilationandCooling(HVAC)systemsinbuildingsaccountforasignicantchunk(33%ofthisenergyuse[ 6 ].HVACsystemsincommercialbuildingsarecommonlyoperatedonascheduleassumingmaximumoccupancyduringnormalworkinghoursandareswitchedoorrunatreducedlevelsatnightorduringperiodsofanticipatednonoccupancy.Runningonascheduleishighlyinecientinmostscenariosastheenergyrequiredtoheat,coolorventilateabuildinglargelydependsonthenumberofoccupants.Observationsofactualbuildingoccupancyhavefoundaverageoccupancyinocebuildingsrepresentatmostathirdoftheirdesignoccupancy,evenatpeaktimesoftheday[ 4 ].15%to25%ofHVACenergyusecanbereducedjustbysettingventilationratesbasedonoccupancy[ 8 ].ThusthereisgreatpotentialtoconserveenergybymodifyingHVACsystemstooperatebasedonoccupancy(demand)ratherthanonaschedule. AcosteectivewayofoptimizingexistingHVACsystemstoconserveenergyistouseadvancedcontrolstrategiesandsensingsystems.Accurateinformationaboutthetrueoccupancyofabuildingiskeyfortheseadvancedcontrolstrategiestobeeective.Anumberofrecentpapershaveusedoccupancydatafromsensingsystemsasinputstoadvancedcontrolalgorithms[ 8 10 13 15 ].Thesehaveshowedthatoccupancybasedadvancedcontrolalgorithmsarecapableofimprovingtheenergyeciencyofbuildings. Allthesecontrolalgorithms,needsensorsthatprovidedataregardingtheIndoorAirQuality(IAQ)-temperature,humidityandCO2levels,insidebuildingsandoccupancyinformation.MonitoringIAQlevelsinsidebuildingsiskeytomaintainingthehealthandsafetyoftheoccupantsandtocomplywithvariousstandardsandcodes.ANSI/ASHRAEStandard62[ 3 ]providesguidelinestomaintainadequateindoorairqualityinbuildings.Advancedcontrolalgorithmsworkbyconstantlychangingparameterssuchasairow 10

PAGE 11

rates,temperaturesetpoints,ventilationratesetc.basedonbuildingoccupancylevels.ThiscanhaveanadverseaectonIAQlevelsandsomonitoringIAQinsidebuildingsisvital.SensorsarethusneededtomonitorCO2,temperature,humidity,airowrateetc. Retrottingexistingbuildingswithsensorsisadicultandexpensiveprocess.Onesolutionistheuseofwirelesstechnology.Goingwirelessremovesalldicultiesinvolvedinwiringthesensorstoacentralstationandsignicantlyreducescost.DesignanddevelopmentofaWirelessSensorNetwork(WSN)capableofprovidingthesensordataforadvancedHVACcontrolapplicationsisprovidedinthisthesisreport. 1.2SystemRequirement CommercialHVACsystemswithaVariableAirVolume(VAV)reheatcongurationtypicallyhaveairhandlingunits(AHU)thatconditiontheairto55F.ThisconditionedairisthensuppliedtoVAVboxes.EachVAVboxcansupplyairtooneormorethanoneroomdependingonthecongurationofthesystem.Airat55FissuppliedtoallVAVboxes.Theytypicallyhavesimplecontrolsystemsthatcontroltheamountofreheatandmaintaincomfortabletemperaturesetpointsinsidetheroom.Commonly,asimplerule-basedfeedbackcontrolisusedthatdoesnotuseanyreal-timeoccupancydata.Thecontrollerdeterminestheamountofairandthetemperatureatwhichithastobesuppliedtotheroomtomaintaintemperatureatpre-determinedsetpoints.TomaintainadequateIAQlevelsspeciedbyANSI/ASHRAEStandard62[ 3 ],mostoften,aminimumowrateisensured.Thustheroomisquiteoftenoverventilatedattimeswhenitisunoccupied.Also,thetemperatureismaintainedstrictlytoensurecomfort,causingwastageofenergyintheformofreheatattheVAVlevel. DemandControlVentilation(DCV)systems,arecommonlyusedtolowerHVACenergybyensuringthatjusttherightamountofconditionedairissuppliedtothebuilding[ 5 14 17 ].DCVsystemshaveusedCO2sensorstopredictoccupancylevelsinbuildingsforover12years[ 17 ].Buildingcodesgenerallydictatetheminimumamountoffreshairthathastobeprovided.BuildingswithoutDCVusuallyhaveventilationsystemsthat 11

PAGE 12

operateataxedratebasedonanassumedoccupancy.Hencequiteoften,thereismorefreshaircomingintothebuildingthanisnecessary.ThecostsassociatedwithconditioningtheexcessaircanbeavoidedbyusingthedierenceinCO2levelsbetweenindoorandoutdoortopredicttheoccupancyofthebuildingandthenalterventilationratesaccordingly.CO2sensorsarevitalsuchsystemsasthecanbeusedtogetapictureofindoorCO2concentrationsandthusoccupancy. PIRsensorscanbeusedtogettheoccupancyproleofbuildingsinwhichroomshavesingleoccupants.Itcanalsodetectmotionandcanhencebeusedforawidevarietyofapplicationsincludinglighting,securityetc.AkeyadvantageofusingaPIRsensoroverothermotiondetectiontechnologiesisthatitseldofviewcanbecontrolledbytheuser.Thishelpsavoidfalsedetectionsduetodisturbancessuchaspeoplemovingincorridorsetc.DeterminingtheoccupancylevelsinbuildingsusingonlyPIRsensorsiseectivewhenthebuildinghasroomswithsingleoccupancy.However,insuchascenario,studieshaveshownthatthePIRsensorisquiteaccurateatdeterminingwhetheranoccupantispresentornotinaroom.[ 19 ]hasshownthattheprobabilitywithwhichthePIRsensordetectsthepresenceorabsenceofanoccupantinasinglepersonoceis:P(sensordetectspresence|occupantpresent)=0.75P(sensordetectsabsence|occupantabsent)=0.99 APIRsensorcandetectthepresenceorabsenceofpeopleaslongastheyaremoving.Dependingonthesensitivityofthesensor,smallmovementofthelimbsofapersonwhoissittinginachairmaybeenoughtotriggerdetection.Hence,toaccuratelydetectthepresenceofanindividualwhoisworkinginanocespace,weneedaPIRsensorwithhighsensitivity.HenceacombinationofPIRandCO2sensorscanbeeectiveinpredictingoccupancylevelswithinbuildings.ControllingHVACsystemsbasedonoccupancycanhaveasignicantimpactonIAQandcomfortlevelsifthesearenotmonitoredcontinuously.BuildingcodeswillbeviolatedifaDCVsystemdoesnotsupplyenoughairtomaintaindesiredCO2levels.Occupantcomfortandsafetygetsaected.Excessiveheating/coolingcanalsocauseuncomfortandhencemonitoringIAQlevelsandtemperaturearevital. 12

PAGE 13

ASHRAEStandard55statesthattheRelativeHumidity(RH)insidebuildingsshouldberoughlywithin25%to60%inwinterand20%to60%insummer,dependingontemperature[ 2 ].AnyHVACcontroltechniqueshouldbecapableofmonitoringthehumiditylevelscontinuouslyforcompliancewithASHRAE55.Moreover,bacteria,viruses,fungi,andsomepestssuchasdustmitesbegintothriveinhumidityoutsidethisrange[ 20 ].Hencemonitoringthehumidityisnecessaryincommercialbuildings. ThecorefunctionalityofanyHVACcontrolsystemistokeeptheclimateinsidethebuildingwithinaspeciedrange.IncommercialbuildingsthatuseVAVsystemstoregulateairowandtemperature,knowledgeofthetemperatureineachzonecanbeusedtocontroltheVAVbox.Temperaturefeedbackallowsthecontrollertoconserveenergyandprovidemorecomfortfortheoccupants.KnowledgeofthetemperatureineachindividualroomwillbeextremelyhelpfulforsystemswhichuseoccupancyforHVACcontrol. Asmentionedearlier,advancedcontrolsystemscanbeemployedonVAVterminalreheatsystems.ThesecontrolalgorithmsuserealtimeoccupancydatatosupplytheminimumamountofconditionedairthatisrequiredtomaintainthestipulatedIAQandthermalcomfort.Controlalgorithmsrequiresensordataatvariousfrequencies.Forrealtimeoccupancyprediction,dataatthehighestfrequencyisdesired.Thewirelesssensornodemustbecapableoftransmittingsensordatafrequentlytoenableaccuratepredictionofchangingoccupancylevels.TheobjectiveofthisthesisistodevelopaWSNthatcanprovidethedatarequiredfortheseadvancedcontrolsystems.TheschematicoftheWSNimplementedinabuildingwithaVAVterminalreheatsystemisshowninFigure 1-1 WeidentifythefollowingrequirementsthatarevitalfortheWSNtobeeective: 1. Temperature,humidity,CO2andoccupancydatashouldbeavailableforeachroom.TemperaturedataisnotcriticalsinceallterminalVAVboxeshavetemperaturesensors. 2. Continuoulsyprovidedataatthedesiredintervalswithoutinterruptions. 13

PAGE 14

Figure1-1.SystemOverview 3. Thewirelesssensornodesdeployedinindividualroomsshouldhaveaservicelife(withoutanymaintenance)ofatleast1yearbarringfailure. 4. InstallationoftheWSNshouldnotrequireanyradicalredesignorreworkontheexistingbuildingstructure.TheWSNshouldberetrotcapableandshouldbedeployableatminimalcostandtime. 5. Hardwarecostsshouldbelowwithoutsacricingonaccuracyandprecision. 6. Theentiresystemshouldbediscreet.Thewirelesssensornodesshouldnotinterferewiththedesignedappealofinteriorspaces. 7. Thewirelesssensornodeshouldbeeasytofabricatewithlowmanufacturingandassemblycosts. 8. Componentswithinthesystemshouldbeeasilyserviceable,easytotroubleshootandmaintain. 14

PAGE 15

1.3FinalSystemLayout TheWSNthatisdiscussedinthisthesisisshowninFigure 1-2 .Ithasastartopology.Individualnodescommunicatetoacentralhub.TexasInstrumentseZ430-RF2500waschosentoactasnodesandthehubinthisWSN.TheindividualnodesareattachedtosensorsandarecalledEndPoints(EP).ThecentralhubiscalledanAccessPoint(AP).TheAP'sareconnectedtoaplugcomputerorabasestationusingcables.SensordatafromtheEP'sarereceivedattheAP.ThebasestationlogsthedataontoaremotelyhosteddatabaseviatheInternet.Thecontrolalgorithmsthenaccessthisdatafromthedatabase.EachEPhassensorsthatcanmeasuretemperature,humidityandCO2levels.APassiveInfrared(PIR)sensorincludedinthewirelesssensornodedetectsmotion.Whentherearemultipleoccupantspresentinaroom,theCO2readingscanbeameasureoftheoccupancy. Figure1-2.WirelessSensorNetwork 15

PAGE 16

CHAPTER2WIRELESSSENSORNODE Thedesignandimplementationofthewirelesssensornodeisdiscussedhere.Wewentthroughacompleteproductdevelopmentcyclewhiledesigningthewirelesssensornode.Thenodeneedstohavesensorstomonitortemperature,humidity,presenceandcarbondioxideaswellastheabilitytoprocessandtransmitsensordatatoabasestation.Thuswewillhavetointegrateallthesensorstoamicroprocessorforprocessingsensordataandaradioforwirelesscommunication.Inthepreliminarydesignphase,welookatvariousoptionsavailableforthehardwarecomponentsandselecttheonesthatmeetourrequirementsofcostandspecications.WethendoadetaileddesignofthePrintedCircuitBoard(PCB)andthepackagingforthenode.Finally,thesoftwarefortheEPonboardthewirelesssensornodeisdiscussed.Fabricationofthewirelesssensornodeisoutsourcedtoalocalcompanyduetotherequirementofsurfacemounttechnologyforsolderingandthelargenumberofsensornodesthathavetobefabricated.Designofthenodeisconstrainedbytheneedforeasyassembly,disassembly,hardwarereplacementandeasytroubleshooting. 2.1HardwareSelection Multiplechoiceswereavailableforallthehardwarecomponentsusedinthewirelesssensornode.Thebestsuitedhardwarewasselectedaftercarefulconsiderationintermsofcostandfeatures. 2.1.1MicroprocessorandRadio TherstprototypewasmadeusinganArduinoBoard,aXbeetransmitterandaZilogPIRsensor. X-CTUsoftwareisusedtointeractwiththeXbeemodules.OneofthemoduleswassettoreceivingmodeandisconnectedtoaPCviaaUSBcable.TheotherXbeemodulewhichisapartoftheprototypeissettotransmitmode.TheXbeemodulesareconguredusingtheinstructionsgivenin[ 1 ]. 16

PAGE 17

Table2-1.ComparisonofArduino+ZigbeeandeZ430-Rf2500 DeviceArduino+XBeeeZ430-Rf2500 Cost$85$20Range100ft15-30ftMeshNetworkingCapability45TemperatureSensor54LowPower54Easytoprogram45CommunicationprotocolZigbeeSimpliciTI TheArduinoboardisprogrammedtoreadadigitalvaluefromthePIRsensorononeofitsI/Oports.Thedigitalreading(1correspondstomotiondetectedand0correspondstonomotiondetected)issenttotheXbeewirelessmodulesetfortransmission.ThereceivingXbeemodulereceivesthedataandsendsittoacomputerviatheserialport.ItwasnotedthattheXbeemodulesprovidedexcellentrangebasedonthehardwarethatwasbeingused.TheArduinoplatformiseasytoprogram,hasaverygoodonlinecommunityandhasenoughAnalog/Digitalportsforattachingperipherals/sensors.Theonlymajorhindranceisthecost.Thecostofthewirelessmodule(XBEE)alongwiththemicrocontroller(Arduino)iscloseto$85perWSN. TexasInstrumentsoersacompletewirelessdevelopmenttool-theeZ430-RF2500( 2-1 )for$20each.TheeZ430-RF2500,includesaMSP430microcontrollerandaCC2500radioalongwiththehardwarerequiredtodevelopanentirewirelessproject.ItcomesbuiltinwithSimpliciTI,aproprietarylow-powerstarnetworkstack,whichenablesrobustwirelessnetworksoutofthebox.ThehardwareincludestheMSP430F22x4whichcombines16-MIPS(millioninstructionspersecond)performance,a10-bitAnalog-Digital-Converter(ADC)pairedwithaCC2500multi-channelRFtransceiverandanintegratedtemperaturesensor[ 11 ].Theonlydownsidetothishardwareplatformisthatithaslimitedwirelessrange.Outofthebox,ithasarangeofonly30feet.Thiscanbea 17

PAGE 18

Figure2-1.TexasInstrumentseZ430-RF2500 majorhindrancewhentheWSNisusedinbuildingswithlargeoorareas.However,therangecanbeextendedbyreducingthecommunicationbandwidth. ThisArduinobasedprototypeestablishedthatitispossibletousewirelessmodulescoupledwithmicrocontrollerstotransmitsensordatawirelesslyhowever,theeZ430-RF2500waschosentobetheEPandAPintheWSNduetoitslowcostandabilitytomeetalltheotherrequirementsincludingrange. 2.1.2Sensors TheeZ430-RF2500platformsupportsanaloganddigitalI/O.Sensorsforsensinghumidity,CO2andtemperaturecommonlyhaveanalogoutputswhilePIRsensorsusuallygiveadigitaloutput.SensorswereselectedtocomplywiththelimitationsoftheeZ430-RF2500platform. 2.1.3CO2Sensor IndoorCO2levelsaretypicallybetween350to2500ppm[ 7 ].CommonlyusedCO2sensorshaderrorsrangingfrom75ppmtoover200ppm[ 9 ].EnergysavingsfromDCVbasedonoccupancydependontheaccuracyofthe CO2sensorsandsosacricingaccuracyforlowerhardwarecostsisnotlogical.ThebestchoiceforaCO2sensorcapableof 18

PAGE 19

Table2-2.ComparisonofCOZIRandK-30Sensors SensorCOZIRK-30 Cost$175$65Range0-2000ppm0-10000ppmPre-calibrated44Powerconsumption3.5mW200mWAccuracy 50ppm30ppmWarmuptime3Seconds120Seconds measuringCO2levelsupto2500ppmatacceptablelevelsofaccuracywastheSensairK30CO2sensor.TheK30sensorhasameasuringrangeof0to10000ppmatanaccuracyof30ppm.TheaccuracyoftheK30isbetterthanmostofthesensorstestedin[ 9 ].Itsupportsserialcommunicationandhastwoselectableanalogoutputs.Theonlydrawbackisthat,theK30isnotsuitableforlowpowerapplicationsasitoperateson150mAat5V.AnotheroptionisaCOZIRultralowpowerCO2sensor.Itneedsonly33mAat3.3V.However,itcanmeasureonlyupto2000ppmandismoreexpensiveat$170eachatthetimeofhardwareselection.TheK30wasselectedasitgivesbetteraccuracyandrangeatalowpriceofonly$65each.Itcomespre-calibratedanddoesnotneedrecalibrationduringitslifetime.ThisisamajoradvantageascalibrationdriftofCO2sensorshasbeenreportedasaseriousissue[ 9 ]. 2.1.4HumiditySensor Alotofhumiditysensorsareavailablecommercially.Humiditysensorsarebasedonresistive,capacitativeorthermalconductivitysensingtechnologies[ 16 ].CapacitivesensorsarethemostwidelyusedbecauseoftheirwideRHrangeandcondensationtolerance.Thekeyfeaturesrequiredinahumiditysensorforthewirelesssensornodeare:goodaccuracy,widerange,abilitytooperateatlessthan5V,analogorserialoutput.TheHIH4030humiditysensorfromHoneywellsatisesalloftheaboveanddoesnotneedanycalibration.Moreover,itcomesasasurfacemountdevicemakingitcheapandeasytouseonaPrintedCircuitBoard(PCB). 19

PAGE 20

2.1.5PIRSensor TherearealotofcommerciallyavailablePIRsensorswhichareusedformotiondetection.APIRsensormeasuresinfraredlightemittedbyobjectsintheenvironmentinitseldofview.Itgetstriggeredwhenaninfraredsourceofintensityhigherthanthebackgroundappearsandthendisappears,suchasahumanbeingmovinginfrontofawall.Thesensordetectsthischangeintheheatsignatureoftheenvironment.TheParallaxPIRsensoriscapableofdetectingmotionatdistancesofupto 30ft.Itcanoperateat5Vandhasanoptiontovarysensitivityifneeded.Thesensoralsohasanonboardledwhichlightsupforvisualfeedbackwhenmotionisdetected.Communicationisthroughadigitaloutput.Acompactformfactor,goodsensitivityandlowprice($10atthetimeofdevelopment)makestheParallaxPIRsensoranidealchoiceforthewirelesssensornode. 2.1.6TemperatureSensor TheEZ430comeswithatemperaturesensorwitharangeof)]TJ /F1 11.955 Tf 9.3 0 Td[(10Fto500F.Thisismorethansucientforindoorclimatemonitoring.Thetemperaturesensorisnotaccurateandneedstobecalibrated.SincemostoftheterminalVAVboxeshavetemperaturesensors,onlyafewroomswhichdonothavededicatedVAVboxeswillneedtemperaturedatafromwirelesssensornodes. 2.2DetailedDesign 2.2.1Prototype TherstprototypethatintegratedallthesensorstotheeZ430-RF2500wasdoneonabreadboard.AllsensorswereconnectedtotheirrespectiveportsontheeZ430-RF2500.TheMSP430wasashedwiththecodefortheEPinstructingittoreadandtransmitsensordata.Powerrequirementsforallperipheralsonthenodearenotsame.ThehumiditysensorandtheCO2sensoroperateat5VwhiletheeZ430-RF2500andthePIRsensoroperateat3:3V.Anarduinoboardwasusedtosupplythetwovoltages.TheprototypewastestedafterconguringanAccessPointtoreceivesensordatatransmittedbytheEP.AfterconrmingthattheeZ430-RF2500wascapableofreadingsensordataand 20

PAGE 21

Figure2-2.Prototype transmittingit,thedesignforaprintedcircuitboard(PCB)capableofsupportingthesensorsandtheeZ430-RF2500wasdone. 2.2.2PowerSupply Onceaworkingprototypewasestablished,apowerauditwasperformedtodeterminethepowersupplyrequirementsofthewirelesssensornode.Ideally,thewirelesssensornodeshouldoperateoneasilyavailableAAorAAAbatteries.Thebatteriesshouldlastforatleastayearsothatthewirelessnodesatisestherequirementofbeinglowmaintenance.However,theCO2andthehumiditysensorsneeda5VsupplywhiletheeZ430-RF2500andthePIRsensorsneed3.3V.Thewirelesssensornodetransmitsdataeverysecondsothereisnotimetoputanyofthesensorsinalowpowersleepmode.Thismakesthenodepowerintensiveandrulesoutthepossibilityofusingcommonlyavailablebatteries. TheCO2sensordrawsthehighestcurrent.ThepowerrequirementsofthevariouscomponentsinthenodeareshowninTable 2-3 A10,000mAHbatterypackthatchargesviaUSBcanpowerthenodeforonly127.2hours(5days).Thisbatterypackcosts$45onamazon.Thebatterypackistooexpensive 21

PAGE 22

Table2-3.Powerrequirementsforvariousnodecomponents ComponentVoltage(V)Current(mA) PIRSensor3.38.9HumiditySensor50.5 CO2540eZ430-RF25003.35.1 andtheoperatinglifeconstantforthenodeusingsuchabatterypackisreallylow.Usingbatteriesforsuchapowerintensiveapplicationisnotpossibleandsothewirelesssensornodewasdesignedtorunonwallpower.Poweroutletsareeasilyaccessibleinmostbuildings. OnewayofrunningthewirelesssensornodeobatteriesistooperateitwithouttheCO2sensor.Thisreducesthepowerrequirementandthenodecanrunforaround466hours(20days)ona10,000mAHbattery.Hencethepowerjackonboardthewirelesssensornodewaschosentobecompatiblewithbatterypowerpacksforoperatingthenodetemporarilyinplaceswithoutwalloutlets. 2.2.3PCBDesign AminimalisticapproachwasusedtodesignthePCBboardtokeepthewirelesssensornodecompactandlight.ThePCBwasdesignedusingproprietarysoftwarefromExpressPCB.ExpressPCBisalowcostPCBmanufacturerthathasintegratedPCBdesignandorderingofPCBboardsintoaconvenientsoftwarepackage.ThePCBusedforthewirelesssensornodeisdoublesided.Thedesignspecicationsfortheboardare: 1. 3:3Vand5Vpowersupply 2. ProgrammingheadersforashingtheMSP430 3. ScrewholesforfasteningthePCBtothecase 4. PinholesforconnectingCO2,PIRandeZ430-RF2500 5. Surfacemountslotforhumiditysensor A5Vwalladaptersuppliespowertotheboard.ApowerjackmakesthispoweravailableonthePCB.A3:3Vregulatorisusedtostepdownthevoltage.Theregulator 22

PAGE 23

needsa1Fcapacitorattheinputanda10Fcapacitorattheoutput.TheoutputfromtheregulatoriswiredtotheeZ430-RF2500andthePIRsensor.TheCO2sensorandthehumiditysensorarepowereddirectlyfromthepowerjack.ThepinsontheeZ430-RF2500thatareusedforashingcodeontheMSP430arealsomadeavailableontheboard.ThismakesiteasytoashtheMSP430withoutremovingitfromthePCB.SincetheCO2sensorsharesthepinsontheeZ430-RF2500usedforprogramming,makingitavailableexternallyimprovesaccessibilityfortroubleshootingtheeZ430-RF2500.AschematicofthePCBisshowninFigure 2-3 Figure2-3.PrintedCircuitBoardSchematic 2.2.4CaseDesign ThePIRsensoronboardthewirelesssensornodeneedsdirectlineofsighttodetectmotion.Thenodethereforeshouldbepositionedtohavemaximumvisibilitywiththeoccupant.Asthenodewillbedeployedinplacessuchasoceworkstations,roomwallsetc.,itshouldappearpleasing.Thepackagingforthewirelesssensornodewasdesignedtomakeitasdiscreetaspossible.Thematerialchosenforthecasingisacrylonitrilebutadiene 23

PAGE 24

styrene(ABS).ABSisoneofthemostcommonlyusedpolymersformakingelectronicpackaging.Itisstrong,durableandlight.Tokeepcostsdown,anexistingmoldfromaninjectionmoldingvendorwasmodiedtomeetourspecications.VentilationholesallaroundtheperipheryofthecaseensurethattheIAQsensorshavegoodaccesstotheairinsidetheroom.TheCADdrawingforthecaseusedisshowninFigure 2-4 .ThecasefabricatedasperthedesigninFigure 2-4 isshowninFigure 2-5 Figure2-4.Case-CADDrawing 2.3SoftwarefortheEndPoint TheEPisaneZ430-RF2500programmedtoreadsensordata,processit,convertitintohexadecimalandtransmititusingtheCC2500radio.TheEPactsasanodeinawirelessnetworkwithastartopology.TheEP'stalkdirectlytotheAccessPoints.CommunicationishandledbytheSimpliciTInetworkprotocol.SimpliciTIisalowpower 24

PAGE 25

Figure2-5.CaseandCover radio-frequency(RF)protocoltargetingsimplesmallRFnetworks(<100nodes).TheprotocolrunsoutoftheboxonTI'sMSP430microcontrollers.SimpliciTIsupportsonlyapeer-to-peernetworktopology.However,hardwarelimitationslimittheimplementationofmeshnetworksontheEZ430platformandhenceSimpliciTIideal. TheeZ430-RF2500isprogrammedusingembeddedCprogramming.TexasInstrumentsCodeComposerStudio(CCS)providesanIntegratedDevelopmentEnvironment(IDE)andthecompilerforconvertingCcodeintobinary.TheowchartfortheoperationsbeingperformedattheEPisshowninFigure 2-6 25

PAGE 26

Figure2-6.EndPointowchart main EP:ccontainsthermwarebuiltintotheEPforourapplication.Thecodeexecutionbeginswithasysteminitialization.ThefollowingcodeshowstheEP'ssysteminitializationprocedure: 1 voidmain(void) 2 f 3 WDTCTL=WDTPW+WDTHOLD; 4 addr tlAddr; 5 /Initializeboard)]TJ /F11 9.963 Tf 8.62 0 Td[(specifichardware/ 6 BSP Init(); 7 /InitializeTimerAandoscillator/ 8 BCSCTL3j=LFXT1S 2;//LFXT1=VLO 9 TACCTL0=CCIE;//TACCR0interruptenabled 10 TACCR0=12000;//~1sec 11 TACTL=TASSEL 1+MC 1;//ACLK,upmode 26

PAGE 27

BSP Init()SimpliciTIAPIcallinitializesthehardwareontheeZ40-RF2500.ItestablishescommunicationbetweentheMSP430andtheCC2500radioandinitializestheLEDs/switchesontheboard.Thefollowingotherinitializationsarealsoperformed: TheDCOandMCLKaresettorunat8MHz. Timer Aissettotriggerinterruptsat1secondintervals. Theuniversalserialcommunicationinterface(USCI)UARTisinitializedtocommunicatewiththeCO2sensor. Afterallthesystemcomponentshavebeeninitialized,theEPtriestoconnecttoanAPthatislisteningforconnections.TheLED'sareturnedontoindicatethatjoiningisyettooccur.TheTimerA0willwaketheCPUupeverysecondtotryandestablishaconnection.Ifaconnectionismade,boththeLED'sareturnedo.ThefollowingcodeshowshowtheEPtriestoestablishaconnectionwithalisteningAP. 1 while(SMPL SUCCESS!=SMPL Init(0)) 2 f 3 BSP TOGGLE LED1(); 4 BSP TOGGLE LED2(); 5 bis SR register(LPM3 bits+GIE);//LPM3withinterruptsenabled 6 g 7 /LEDsonsolidtoindicatesuccessfuljoin./ 8 BSP TURN OFF LED1(); 9 BSP TURN OFF LED2(); 10 /UnconditionallinktoAPwhichislisteningduetosuccessfuljoin./ 11 linkTo(); Ifasuccessfulconnectionisestabilshed,theprotocolissuesahandlecalledtheLinkID.ThiswillbeuniqueforeachEPonthenetworkjustlikeanaddress.IndividualpacketsfromeachEPareidentiedattheAPusingtheLinkID.ThelinkTo()functionhandlessensordatacollectionandcommunicationoncetheEPhasattainedaworkinglinktoanAP.TimerA0wakesuptheCPUeverysecondtoenablesensordatacollectionand 27

PAGE 28

transmission.Aag-sSelfMeasureSem,insidetheTimerinterruptroutinekeepstrackofwhethersensordataneedstobemeasuredornot. OncethetimerraisesthesSelfMeasureSemag,thesensorreadingsareattainedstartingwiththetemperaturesensor.ThetemperaturesensorismappedtooneoftheADCports(Pin2)ontheeZ430-RF2500.Areferencevoltageof1.5Visused.Thereferenceisallowedtosettleforatleast30sbeforetheanalogportissampled.Themeasuredvalueisthensavedinanarray.Thefollowinglinesofcodeshowhowthetemperatureisreadviaananalogport: 1 /Gettemperature/ 2 ADC10CTL1=INCH 10+ADC10DIV 4;//TempSensorADC10CLK/5 3 ADC10CTL0=SREF 1+ADC10SHT 3+REFON+ADC10ON+ADC10IE+ADC10SR; 4 /Allowrefvoltagetosettleforatleast30us(30us8MHz=240 cycles) 5 delay cycles(240); 6 ADC10CTL0j=ENC+ADC10SC;//Samplingandconversionstart 7 bis SR register(CPUOFF+GIE);//LPM0withinterruptsenabled 8 results[0]=ADC10MEM;//Retrieveresult 9 ADC10CTL0&=~ENC; 10 11degC=(((temp)]TJ /F11 9.963 Tf 14.05 0 Td[(673)4230)/1024);// temp calibration 12 if((tempOffset)!=0xFFFF) 13 f 14 degC+=(tempOffset); 15 g Theanalogvalueisconvertedintotherequiredtemperaturevalue.Anyerrorsinthemeasurementarecorrectedusingaonepointcalibration.Thecalibrationconstantisrepresentedasanosetandisdierentfordierentsensors.Thecalibrationconstanthastobedeterminedempirically.ItisthenupdatedonthecodebeforetheeZ430-RF2500is 28

PAGE 29

ashed.Assumingazerothorderderivative,thetemperatureiscalibratedas:^T=TADC+offset (2{1) ThePIRsensorgivesadigitalo/p.Anoutputhighcorrespondstomotiondetected.Thedigitalpinremainsatoutputlowifnomotionisdetected.OneoftheADCpins(Pin2.2)ontheeZ430-RF2500issetasadigitalinput.Avalueof1or0isassignedtoavariablebasedontheoutputatthispin.ThefollowinglinesofcodeillustratedatacollectionfromaPIRsensor: 1 //GetPIR 2 intvalue; 3 P2DIR&=~BIT2;//SetP2.2toinput 4 value=(P2IN&BIT2); Thehumiditysensoralsogivesananalogoutput.ItismappedtoADCPin8.Itrequiresareferencevoltageof2.5V.Thevaluemeasuredisconvertedintotherequiredhumidityvalueandisstoredinavariable. 1 ADC10CTL0=SREF 1+ADC10SHT 3+REFON+ADC10ON+ADC10IE+REF2 5V; 2 delay cycles(240); 3 ADC10CTL0j=REF2 5V; 4 5ADC10CTL1=INCH 3+ADC10DIV 3;//InputSelctandClockDiv 6 P2SEL=0x08;//SelectPin2.3 7 ADC10AE0=0x08;//ADCLowBit(A3) 8 ADC10AE1=0x00; 9 ADC10CTL0j=ENC+ADC10SC;//Starttosample 10 bis SR register(CPUOFF+GIE);//LPM0w/int 11 results[3]=ADC10MEM;//StoreResult 12 ADC10AE0=0;//ResetSelectionBits 13 ADC10AE1=0; 14 humidity=(int)((((((results[3]/1023.0)2.5))]TJ /F8 9.963 Tf 8.54 0 Td[(.45)100.0)/3))]TJ /F8 9.963 Tf 8.13 0 Td[(15); //DerivedfromDatasheetInfo 29

PAGE 30

ReadingsfromtheCO2sensorareavailableinanalogorserialformat.Duringtesting,itwasnotedthattheanalogreadingswerenotveryreliableasthesignalprocessingalgorithmbuiltintotheCO2sensorworksecientlyonlywhenthedataisreadinserialformat.PollingtheCO2sensorviaserialrequiresafreeserialportontheeZ430-RF2500.TheeZ430-RF2500hasonlyoneserialportwhichisusedtoashtheMCU.ThisportisavailableoncetheMCUisashedandcanbeusedtoreadfromtheCO2sensor.PollingtheCO2sensorviatheserialportinvolvessendingasetofhexadecimalvaluestothesensor.ItthenrespondswiththedesiredCO2reading.Thecommandsin 2-4 needtobesendtotheCO2sensortoinitiatearesponsefromthesensor[ 18 ]. Afterreceivingtheinstructionlist,theCO2sensormeasurestheCO2valueandtransmitsitviatheserialport.ThetransmitteddatawillincludetheaddressoftheCO2sensoralongwiththemeasuredCO2levels.Themeasuredvaluewillbeinhexadecimalformatwithahighbyteandalowbyte.Theresponsehasthefollowingformat: Table2-4.InstructionsetforCO2sensor. AddressoftheSensorReadinitiatecommands 0xFE0x440x000x080x020x9F0x25 TheMSP430MCUcanreadtheresponseusinganinterruptserviceroutine[ 12 ].Thefollowinglinesofcodeshowhowtheinterruptserviceroutineisimplemented. 1 /USCIAinterruptserviceroutine/ 2 #pragmavector=USCIAB0RX VECTOR 3 interruptvoidUSCI0RX ISR(void) 4 f 5 if(UCA0RXBUF==0xFE)nnCheckifcorrectdataisbeingsent 6 f 7 bufpos=0; 8 buffer[bufpos]=0xFE; 9 g 30

PAGE 31

10 buffer[bufpos]=UCA0RXBUF; 11 bufpos++; 12 g TheinterruptserviceroutinegetsactivatedwhentheCO2sensorsendstherstpacketoftheresponse.EachpacketgetsstoredintheRXbueroftheMSP430'sUARTandistransferredontoatemporaryvariablesothatthemorepacketscanbewrittenintoit.Thetemporaryvariable,representstheresponsefromtheCO2sensor. Onceallthesensordataiscollected,itisconvertedintohexadecimalformatmadereadyfortransmission.Thefollowinglinesofcodeshowhowthesensorreadingsfortemperature,humidityandCO2areformattedandstoredinatemporaryvariable(msg). 1 //Temperaturesensor 2 temp=results[1]; 3 volt=(temp25)/512; 4 msg[0]=degC&0xFF; 5 msg[1]=(degC>>8)&0xFF; 6 msg[2]=volt; 7 8//HumiditySensor 9 msg[5]=humidity&0xFF;//Packagethehumiditydataintohighandlow bytes 10 msg[6]=(humidity>>8)&0xFF; 11 12//CO2sensor 13 if(buffer[0]==0xFE) 14 f 15 msg[3]=buffer[3]; 16 msg[4]=buffer[4]; 17 g 31

PAGE 32

ThetemporaryvariableissenttotheaccesspointusingtheSMPL SendOptSimpliciTIAPI.TheLinkIDfortheEPassignedbythenetworkissentalongwiththetransmittedmessage. 32

PAGE 33

CHAPTER3ACCESSPOINT 3.1Hardware TheAccessPointisaez430-RF2500boardthatrelaysdatabetweentheendpointandthebasestation.TheAPinterfaceswiththebasestationoverUSBasaserialdevice.AUSBtoserialconverterconguredtooperateat3:3VVCCand3:3VI/Oenablesserialcommunication.TheCOMportsettingsare9600;N;8;1(baud,parity,databits,stopbits). 3.2Software TheAPactsasadatahubinawirelessnetworkwithastartopology.AlltheEndPointsactasnodesandtalkdirectlytotheAP.CommunicationishandledbytheSimpliciTInetworkprotocol.ThesequenceofoperationoftheAPisshowninFigure 3-1 Figure3-1.AccessPointowchart 33

PAGE 34

main AP:ccontainsthermwarebuiltintotheAPforourapplication.ThecodeexecutionbeginswithasysteminitializationthatisalmostidenticaltotheEP'sinthenetwork.ThefollowingcodeshowstheAP'ssysteminitializationprocedure: 1 voidmain(void) 2 f 3 bspIState tintState; 4 /Initializeboard/ 5 BSP Init(); 6 7/InitializeTimerAandoscillatorReference:Glitovsky/ 8 BCSCTL3j=LFXT1S 2;//LFXT1=VLO 9 TACCTL0=CCIE;//TACCR0interruptenabled 10 TACCR0=12000;//~1second 11 TACTL=TASSEL 1+MC 1;//ACLK,upmode 12 13/Initializeserialport/ 14 COM Init(); BSP Init()SimpliciTIAPIcallinitializesthehardwareontheEZ40.ThiscallinitializesthecommunicationbetweentheMSP430andtheCC2500radioandtheLEDs/switchesontheboard.Thefollowingotherinitializationsarealsoperformed: TheDCOandMCLKaresettorunat8MHz. Timer Aissettotriggerinterruptsat1secondintervals. Theuniversalserialcommunicationinterface(USCI)UARTisinitializedtocommunicatewiththePCComport. Oncehardwareinitializationiscomplete,theSMPL Init(sCB)function,initializesthenetwork.ThesCBparameterisafunctionpointerthatisexecutedwithintheinterruptserviceroutine(ISR)uponpacketreceptionbytheAP.Oncenetworkinitializationiscomplete,boththeLEDsareturnedontoindicatethattheAPiswaitingforanEPtojoin. 34

PAGE 35

ThesCBcallbackfunction,ltersthereceivedpacketaccordingtoitslinkIDtoidentifythesourceoftransmissionanddistinguishtheEPjoinrequestfromadatapackettransmissionfromanEPthathasalreadyestablishedaconnectiontothenetwork.AlinkIDof0identiesajoinrequest.UponacceptanceofanEPjoinrequest,theAPenumeratesnewmemberstothenetworkandassignsincrementalIDsfrom0x01to0x1D.Thepossibleenumerationvaluesfrom0x01to0x1DareadesignnetworkconstraintfortheSimpliciTIprotocolthatallowsupto30devicestobelinkedtotheAP.AlinkIDfrom0x01to0x1DidentiesthereceptionofapacketfromoneofthenetworkEP's.Forourapplication,themaximumnumberofconnectionsisrestrictedto8. AccordingtothelinkID,thesCBcallbackfunctionidentiesandincrementstherespectivesPeerFrameSem[aphore]orsJoinSem[aphore]forhandlingtheprogram'smainloop.Thesetwosemaphores,controltheprogramowafternetworkinitialization. 3.2.1sJoinSemBranch ThesJoinSemsemaphoreissetwhenanEPcallsitsSMPL Init()function.Ajointothenetworkisactuallyasideeectofinitialization,asthereisneveranactualcallmadethatrequestsanetworkjoin.WhenthesJoinSem[aphore]hasbeensetintheAP'scallbackfunctionandaslongastheAPhasmadelessthanitsmaximumnumberoflinks,theAPwillcallitsSMPL LinkListen()function.Asthefunctionparameter,thelinklistenfunction,takesapointertothelinkIDthatwillbeusedtocommunicatewiththelinkedEP.SMPL LinkListen()isablockingcall,meaningitwillnotreturnuntilasuccessfullinkhasbeencreated,soitisimportantthattheSMPL LinkListen()callbemadeonlywhentheuserknowsthatanotherdevicehasmadethelinkrequestusingtheSMPL Link().Onasuccessfullinkcreation,thesJoinSembranchincreasesthenumberofdevicesthattheAPrecognizesaspartofthenetworkandunlocksthesJoinSem[aphore]foranotherdevicetoset.ImplementationofthesJoinSembranchisshowninthefollowinglinesofcode: 1 35

PAGE 36

2if(sJoinSem&&(sNumCurrentPeers
PAGE 37

6 for(i=0;i
PAGE 38

thefunctionTXString.ThisfunctionusespointerstolltheTXbuer.ThefollowinglinesofcodeshowtheimplementationofTXString: 1 voidTXString(charstring,intlength) 2 f 3 intpointer; 4 for(pointer=0;pointer
PAGE 39

CHAPTER4TESTINGANDVALIDATION ThewirelessnodesweremanufacturedasperthedesignpresentedinSection 2 .ThedesignandtheperipheralssourcedfromvariousvendorswereprovidedtoJustInTimeManufacturing-acontractmanufacturingcompany.Therstbatchof5nodeswasfabricatedtoestablishthemanufacturingcostandtoobtainsamplesfortesting.Thedicultiesinmanufacturingtheboxwerenotedandthedesignwastweakedtoremovethese.Atotalof70nodeswerefabricatedwiththenewupdateddesign.ThenishedwirelesssensornodewhendeployedonawallisshowninFigure 4-1 Figure4-1.WirelessSensorNode-Deployedonawall FortestingtheWSNweneedabasestationthatcanlogdatafromtheAPontoadatabase.Section 4.1 detailsthebasestationthatwasusedfortestingtheWSN. 39

PAGE 40

4.1BaseStation ThemainroleofthebasestationistoreceivedatafromtheAP,timestampitandwritethedatatoaremotelyhosteddatabaseviatheInternet.Forthispurposeweneedaplugcomputercapableofrunningascriptthatcanreaddatafromtheserialport,processitandwriteittoadatabase.TheDreamplugfromMarvellisacompactplugcomputerthatiscapableofrunningDebianLinux.Itcomespreloadedwiththeoperatingsystem.Ithas802:11b=gWiFiadapter,twoEthernetport,twoUSBportsand4GBofinternalmemory.Thankstoitsruggednessandcompactformfactoritcanbeconcealedwhendeployedinbuildings.TheDreamplugischosenasthebasestation. AFTDIserialtoUSBconverterthathasa3:3VserialI/Oanda3:3VpowersupplyisusedtoconnecttheAPtothebasestation.ApythonscriptisexecutedonthebasestationtologdatacominginfromtheAP.Thepythonscriptstartswithinitialization.ThedatabaseconnectionandserialcommunicationwiththeAPareestablished.ThefollowinglinesofcodewrittenbySiddarthGoyalillustratetheinitializationprocess. 1 2db=postgresql.open("pq://postgres:therock@10.245.16.227/postgres")nn Databaseconnectionopened 3 ports=["/dev/ttyUSB0","/dev/ttyUSB1","/dev/ttyUSB2","/dev/ttyUSB3"]nn Portstocheckforconnections 4 #ports=["/dev/ttyUSB0","/dev/ttyUSB2"] 5 nnserialportsettings 6 byte size=8 7 rate=9600 8 parity=0 9 stopbit=1 10 debug=True ThepyserialAPIisusedtohandleserialcommunication.EachlinerepresentsdatafromoneoftheEP's.Theline.decodefunctionisusedtoprocesseachdataset.Thesensor 40

PAGE 41

Figure4-2.BaseStation readingsarecommaseparatedandtheline.decodefunctionsplitsthedatasetintoindividualsensorreadings.Thedataischeckedforconsistencyandisthenwrittenintothedatabaseusingthedb.executeroutine. 4.2Testing TestswerecarriedouttoensuretheecacyandrobustnessoftheWSN.LongtermsustainabilityiskeyfortheWSNasmaintenanceandreworkcostsarereallyhighandtimeconsuming.TheHVACsystemhastoperformwithzerodowntimeandhencetheWSNfailingevenforashorttimewillhaveastrongimpactonenergysavings,occupantcomfortandhealth. TherstroundoftestswasperformedtodeterminetheindoorrangeoftheeZ430-RF2500radio.Theresultswerenotveryencouraging.TheeZ430-RF2500isaplatformthatisoptimizedforlowpowerapplicationsandhencetherangewaslimitedto 41

PAGE 42

lessthan15feet.Thiswasnotacceptableasbasestationsareexpensiveandhencedeploymentcostswillnotbejustiable. Asignicantimprovementinrangecanbeattainedbydecreasingthedatarate.TheSimpliciTIprotocolhasadatarateof250kbpsbydefault.SuchahighdatarateisnotrequiredforourapplicationaseachdatapacketbeingtransmittedbytheEP'sislessthan1.6KBandthesensordataisbeingtransmittedonceeverysecond.Adatarateof2kbpswillbesucient.Thedatarateforthecommunicationchannelisdeterminedbyparameterssetinmr defs.hwithintheSimpliciTIprotocol.Themodieddenitionsareshownbelow: 1 2#define mrfi LENGTH FIELD SIZE 1 3 #define mrfi ADDR SIZE 4 4 #define mrfi MAX PAYLOAD SIZE 20 5 6#define mrfi RX METRICS SIZE 2 7 #define mrfi RX METRICS RSSI OFS 0 8 #define mrfi RX METRICS CRC LQI OFS 1 9 #define mrfi RX METRICS CRC OK MASK 0x80 10 #define mrfi RX METRICS LQI MASK 0x7F 11 12#define mrfi NUM LOGICAL CHANS 4 13 #define mrfi NUM POWER SETTINGS 3 14 15#define mrfi BACKOFF PERIOD USECS 250 16 17#define mrfi LENGTH FIELD OFS 0 18 #define mrfi DST ADDR OFS ( mrfi LENGTH FIELD OFS + mrfi LENGTH FIELD SIZE ) 19 #define mrfi SRC ADDR OFS ( mrfi DST ADDR OFS + mrfi ADDR SIZE ) 42

PAGE 43

20 #define mrfi PAYLOAD OFS ( mrfi SRC ADDR OFS + mrfi ADDR SIZE ) 21 22#define mrfi HEADER SIZE (2 mrfi ADDR SIZE ) 23 #define mrfi FRAME OVERHEAD SIZE ( mrfi LENGTH FIELD SIZE + mrfi HEADER SIZE ) 24 25#define mrfi GET PAYLOAD LEN (p)((p))]TJ /F13 9.963 Tf 6.33 0 Td[(>frame[ mrfi LENGTH FIELD OFS ] )]TJ ET q .398 w 39.15 -171.94 m 42.14 -171.94 l S Q q .398 w 43.43 -171.94 m 46.42 -171.94 l S Q BT /F8 9.963 Tf 47.11 -172.14 Td[(mrfi HEADER SIZE ) 26 #define mrfi SET PAYLOAD LEN (p,x)st((p))]TJ /F13 9.963 Tf 6.33 0 Td[(>frame[ mrfi LENGTH FIELD OFS ]=x+ mrfi HEADER SIZE ;) BychangingtheparametersintheCC2500congurationle,thedataratewassetto2kbps.Withthereviseddatarate,wewereabletoboosttheindoorrangeto50-70feet.Thisincreasedrangecaneasilycover7-8roomsinatypicalocebuilding. AfterimprovingtherangeoftheeZ430-RF2500platform,anotherroundoftestswereperformedtoestablishtheirlongtermsustainability.4EP'swereconnectedtoabasestationthroughanAP.Thebasestationwasexecutingthepythonscriptthatlogsdataontothedatabase.Itwasnoticedthatthethenetworkfailedafter55hours.Whenthetestwasrepeatedwiththesameconguration,thenetworkagainfailedbutnowin69hours.WhenthebasestationwasreplacedwithaLinuxcomputertovisualizetheincomingdatainrealtime,itwasnoticedthatthedatastringbeingreadbythepythonscripthadformattingerrorsandwasnotconsistent.Sucherrorscanleadtomemoryleakageswithinpythonandcanbeareasonforthefailureofthenetwork.TheseformattingerrorshoweverwerenotseenwhentheAPwasconnectedtoawindowsmachine.ThedatawhenvisualizedinwindowsusingaserialportreadersoftwaresuchasTermite,wasconsistentanderrorfree.Torearmthattheproblemwasonthebasestationpluspythonside,theentirenetworkwasagaintestedusingawindowsplatformtovisualizedatabeing 43

PAGE 44

transmittedbytheAP.Thenetworkhowever,failedagaininalmostthesametime(51hours).WecannowsafelyassumethatthereasonforfailureisattheAPorbasestation. ChangestotheformatinwhichtheAPtransmitsdatadidnotsolvetheissue.TheAPemploysatimerthatisusedtoreadthetemperaturesensoronboardtheeZ430-RF2500.Thereisanothertimerthattakescareofreceivingwirelessdataandcommunicationthroughtheserialport.Ahypothesiswasthatthepresenceoftwotimers(oneisdependentontheother)maybeleadingtomemoryissuesontheresourcelimitedeZ430-RF2500platform.TheeZ430-RF2500actingastheAPdoesnotneedtosensetemperature.TheonlyfunctionoftheAPistoreceivewirelessdataandtoforwardittotheserialport.WecansafelydisablethetimerthatcontrolsthetemperaturereadingandreportingontheAPwithoutanysignicantimpactontheWSN.TheAPnowhasonlyonetimerthattakescareofreceivingandforwardingwirelessdata. AnotherhypothesiswasthatthepythonscriptwasresettingtheAPasboththeAPandthebasestationsharedthesameserialbuer.ItwasproposedthattheAPIusedtoreadfromtheserialportinpythonwascausingmemoryissues.AstheAPwrotedataintothisbuer,memoryoverowontheserialbuercanstietheAPfromwritingtothesamebuer.Torectifythisissue,thepythoncodewasmadetoresetevery24hours. TheWSNwasagaintestedusingthesingletimerAPconnectedtoabasestationrunningaselfresettingpythonscript.TheWSNworkedwithoutanyhiccupsfor30days.ThistestestablishesthelongtermviabilityoftheWSN. EachAPontheWSNcanconnecttoamaximumof8EP's.Thenumber8islimitedbytheSimpliciTIprotocolduetotheeZ430-RF2500platformslimitedhardware.Thislimitationcanbeamajorhindranceifthereisaneedtoscalethethenetwork.Wefacethischallengeasweplantodeployandtesta70nodenetworkinabuildingattheUniversityofFloridacampus.Atleast8AP'swillbeneededtoservicethisbuilding.TolimitinterferenceandtoensurethattheEP'scommunicateonlywiththeAPtowhichithasbeenassigned,theAP'sweresettooperateatdierentfrequencychannels.Thebase 44

PAGE 45

frequencyusedis2424.998MHz.Eachchannelisspaced249.938MHzapart.EP'ssettooperateonchannel1canonlycommunicatewiththeAPthatislisteningonchannel1.Thenetworkcanthusbeeasilyscaledwithlimitedinterference. 4.3Deployment ThewirelesssensornetworkwillbeusedtoimplementanadvancedcontrolalgorithmontheHVACsystematabuilding(PughHall)ontheUniversityofFloridacampus.70wirelesssensornodeswillbeconnectedto10basestationstocovertheentireoorareaofthebuilding.AschematicofthelocationofthebasestationsandthewirelesssensornodesonFloor2ofthebuildingisshowninFigure 4-3 4.4CostAnalysis ThecostofdeployingandoperatingawirelessnetworkforHVACcontrolshouldbeminimalasmostHVACdesignershavelimitedfundsforretrots.Alsoaninvestmenttoreduceenergycostsshouldhavepaybackperiodsof3yearsorless.TheentiredesignoftheWSNwasperformedkeepinghardwarecoststoaminimum.Thedesignofthewirelesssensornodeandthenetworkwasdonebygraduatestudentsearning$20anhour.Thedesignanddevelopmentofthenodestook12weeks.Thefabricationofthenodesisoutsourcedtoalocalcompanywhichcharges$76.68eachforsolderingandassemblingallthecomponentswithinthenode.TestingandvalidationoftheWSNrequired15weekstakingthedevelopmenttimetoatotalof27weeks.Agraduatestudenttypicallydedicateshalfofhis20hourweektoresearchandsoovera27weekperiod,atleast270hourswouldhavebeenspentondevelopingtheWSN.ThelaborcostforR&Dwillthusbecloseto$5400.Consideringonlythecostofhardwareandfabrication,eachnodecosts$214.90.ThecomponentwisepricebreakupisshowninTable 4-1 .Pricescanbefurtherreducedforlargequantitiesbyoptimizingthemanufacturingsetupandobtainingbulkquantitydiscountsonthecomponents. 45

PAGE 46

Figure4-3.Deploymentplan:PughHallFloor2 Thedreamplugsbeingusedasbasestations,costs$169each.Eachdreamplug,needsanAP(eZ430-RF2500)thatcosts$20.40andaserialtousbFTDIcablethatcosts$14.60.Thuseachbasestation+APintheWSNcosts$204. 4.5Shortcomings Therearequiteafewareasforimprovement.Duringtesting,thenetworkhasdemonstratedstabilityandrobustnessbutlongtermviabilityoftheWSNinitscurrent 46

PAGE 47

Table4-1.WirelessSensorNodecost ItemCost($) eZ430-RF250020.40PIRSensor9.34HumiditySensor9.45 CO2Sensor65.00Capacitor1 F0.11Capacitor10 F0.12PowerJack0.853.3VRegulator15VPoweradapter5.27Case16.55Cover1.12PCB9.01Assembly76.68 TotalCost214.9 hardwareandsoftwareformfactorisyettobeveried.Someofthemajorpitfallsandsuggestionsforimprovementareprovidednext. WhentheWSNgoesthroughapowercycle(possiblyduetoablackout),theAP'swaitforconnectionsandtheEP'skeeptryingtoestablishaconnectionwithawaitingAP.Onceaconnectionisestablished,theWSNperformsasdesired.However,afterestablishingconnection,iftheAP'sinthenetworkgothroughapowercycle,thenetworkwillfail.TheEP'scannotrecognizethattheAP'shavegonethroughapowercycle.ThenetworkprotocolhastoreassigntheLinkIDtoeachEPbyrepeatingtheconnectionprocedure.SimpliciTIdoesnothaveanybuiltinfunctionsthatautomaticallyperformthereconnectionprocedure.ThisproblemcanbesolvedbywritingafunctionresidingontheAPthathashandlestoreassignLinkID'stoeachEPthathaslostconnectivity. Loweringthedatarateto2kbpsgaveasignicantboosttotheindoorrangeoftheeZ430-RF2500platform.However,theindoorrangeisstillunder100ft.Lowerrangecallsfortheuseofmorebasestationsjusttocompensateforthelackofrange.Thiswillincreasethecostofdeploymentinlargebuildings.RangeisamajorlimitationoftheeZ430-RF2500asthisplatformhasbeenoptimizedforlowpower/lowcostapplications. 47

PAGE 48

UsingotherradiosinsteadoftheCC2500canresultinasignicantboostinrange.XbeeradiosrunningontheZigbeeprotocolareavailablecommerciallywitharangeofupto10km.However,theywillincreasethecostofeachwirelesssensornodeby$65. HardwarelimitationsoftheeZ430-RF2500preventtheimplementationofmeshnetworks.AlsotheyarenotcapableofrunningembeddedoperationsystemssuchasTinyOS.Asaresult,themicroprocessoronboardcannotbeprogrammedovertheair.Theyhavetobebroughtbacktothelabtobereprogrammed.DoingrmwareupgradesinalargebuildingwherealotofwirelesssensornodeshavebeendeployedisalaborioustaskinthecurrentcongurationoftheWSN. Althoughusingwallpowerisreliableandcheap,wallsocketsarenotalwaysavailable/accessibleinallrooms.Byoptimizingthehardwaretoconsumeverylittlepower,wecanrunthewirelesssensornodeoreadilyavailablelowcostbatteries.Lowpowerdevicesareusuallymoreexpensiveandlessecient.ForexampleanultralowpowerCO2sensor,costs$175whiletheoneusedinthewirelesssensornodecostsonly$65. TheEP'sinthenetworkareassignedLinkIDsbytheSimpliciTIprotocol.AlltheAP'sthevicinityofthenetworkwhichassignedtheIDcancommunicatewiththeEP's.ThereisnowayforanEPtodecidewhichAPitwantstoconnectto.ThisrestrictionmakesitdiculttopreventtheEP'sfromcommunicatingwithtwonearbybasestations.AseachAPcanconnecttoonly8EP's,wewishtokeepmultipleconnectionsataminimum. 48

PAGE 49

CHAPTER5CONCLUSIONANDFUTUREWORK 5.1Summary Inthisthesis,wehaveshownthedevelopmentcycleofaWirelessSensorNetworkforcollectingclimate,occupancyandIAQdatafromwithinbuildings.ThisdatawillthenbeusedbyadvancedcontrolalgorithmstooptimizetheoperationofVAVterminalreheatbasedHVACsystems.TheWSNconsistsofwirelesssensornodeswithEndPointsthatrelydatatoAccessPointsthatareconnectedtobasestations.Basestations,logthedataintoadatabaseviaEthernetorWiFi.TheWSNwastestedtoensurethatitmeetsallofthesystemrequirements.Testswereconductedtoestablishitslongtermsustainabilityandrobustness. 5.2FutureWork TheWSNwillbedeployedandtestedatPughHall,a40,000sqftLEEDcertiedbuildingontheUniversityofFloridacampus.ThebuildinghasaVAVterminalreheatHVACsystemandisidealfortestingthefunctionalityoftheWSN. Table5-1.WirelessSensorNodeVersion2 ItemCost($) Zigbee+MCU85PIRSensor9.34HumiditySensor9.45 CO2Sensor175.00Capacitor1 F0.11Capacitor10 F0.12PowerJack0.85BatteryPack45Case16.55Cover1.12PCB9.01Assembly76.68 TotalCost428.23 Infuture,wewishtoaddressmostoftheissuesraisedinSection 4 .Solvingtheseissueswouldrequirechangesinthehardwareandsoftwareofthecomponentsinvolvedandisnot 49

PAGE 50

trivial.ThecostofcomponentsforanupdatedwirelesssensornodeisshowninTable 5-1 .ThelowpowerCO2sensorwillenablethewirelesssensornodetolastupto8monthsona10,000mAHbatterypack.Thisisasignicantimprovementoverthecurrenthardware. Thenewwirelesssensornodewillallowmorethan8connectionsforeachAP.TherangeofeachAPwillalsoincreasesubstantiallyasZigbeeradioshaveindoorrangeofupto500ft.ThistranslatestofewernumberofbasestationsineachWSN.CostofeachbasestationcanalsobereducedbyusingaRaspberryPIinsteadofdreamplug.ARaspberryPIisalowcostplugcomputerandsatisesalltherequirementsforabasestation.ARaspberryPIcostsonly$25. 50

PAGE 51

REFERENCES [1] U.S.G.EnergyUseAdministration,\Electricityexplained-useofelectricity,"2010.[Online].Available: http://www.eia.gov/energyexplained/index.cfm?page=electricity use [2] M.J.BrandemuehlandJ.E.Braun,\Theimpactofdemand-controlledandeconomizerventilationstrategiesonenergyuseinbuildings,"ASHRAETransactions,vol.105,1999. [3] V.Erickson,M.A.Carreira-Perpinan,andA.Cerpa,\OBSERVE:Occupancy-basedsystemforecientreductionofHVACenergy,"in10thInternationalConferenceonInformationProcessinginSensorNetworks(IPSN2010),April2011,pp.258{269. [4] S.Goyal,H.Ingley,andP.Barooah,\Zone-levelcontrolalgorithmsbasedonoccupancyinformationforenergyecientbuildings,"inAmericanControlConference(ACC),2012,pp.3063{3068. [5] M.Morari,F.Oldewurtel,andD.Sturzenegger,\ImportanceofOccupancyInformationforBuildingClimateControl,"AppliedEnergy,Jan.2012.[Online].Available: http://control.ee.ethz.ch/index.cgi?page=publications;action=details;id=3984 [6] F.Oldewurtel,D.Gyalistras,M.Gwerder,C.Jones,A.Parisio,V.Stauch,B.Lehmann,andM.Morari,\IncreasingEnergyEciencyinBuildingClimateControlusingWeatherForecastsandModelPredictiveControl,"inClima-RHEVAWorldCongress,Antalya,Turkey,May2010. [7] AmericanSocietyofHeating,RefrigeratingandAir-ConditioningEngineers,Inc.,\ANSI/ASHRAEstandard62.1-2007,ventilationforacceptableairquality,"2007.[Online].Available: www.ashrae.org [8] J.R.Sand,\Demand-controledventilationusingco2sensors,"U.S.DepartmentofEnergy,EnergyEciencyandRenewableEnergy,Tech.Rep.,March2004. [9] N.NassifandM.Zaheer-Uddin,\SimulatedperformanceanalysisofamultizoneVAVsystemunderdierentventilationcontrolstrategies,"ASHRAETransactions,pp.617{629,Jan2007. [10] S.J.EmmerichandA.K.Persily,\State-of-the-artreviewofCO2demandcontrolledventilationtechnologyandapplication,NISTIR6729,"NationalInstituteofStandardsandTechnology,Tech.Rep.,July2001. [11] R.Subramany,C.Liao,andP.Barooah,\Performancecomparisonofsensingsystemsforbuildingoccupancymeasurement,"EnergyandBuildings,2012. [12] R.AmericanSocietyofHeatingandI.Air-ConditioningEngineers,\Ansi/ashraestandard55-1981,thermalenvironmentalconditionsforhumanoccupancy,"1992.[Online].Available: www.ashrae.org 51

PAGE 52

[13] A.TenWoldeandW.B.Rose,\Criteriaforhumidityinthebuildingandthebuildingenvelope,"inProceedingsofworkshoponcontrolofhumidityforhealth,artifactsandbuildings,November1993,pp.63{65. [14] \Arduinoinstructions."[Online].Available: http://www.ladyada.net/learn/arduino/ [15] T.Instruments,\ez430-rf2500user'sguide,"2007.[Online].Available: www.ti.com/lit/ug/slau227e/slau227e.pdf [16] C.Erdmann,K.Steiner,andM.Apte,\Indoorcarbondioxideconcentrationsandsickbuildingsyndromesymptomsinthebasestudyrevisitted:Analysisof100bulidigdataset,"inProceedings:IndoorAir2012. [17] W.Frisk,D.Sullivan,D.Faulkner,andE.Eliseeva,\Co2monitoringfordemandcontrolventilationincommercialbuildings,"CaliforniaEnergycommisssionreport,2011. [18] D.K.Roveti,\Choosingahumiditysensor:Areviewofthreetechnologies,"July2001.[Online].Available: http://www.ohmicinstruments.com/home/?p=179 [19] Sensair,\K-30CO2sensorcommunicationdocumentation."[Online].Available: http://co2meters.com/Documentation/Other/SenseAirCommGuide.zip [20] G.Litovsky,\Beginningmicrocontrollerswiththemsp430-tutorial."[Online].Available: http://www.glitovsky.com/Tutorialv0 4.pdf 52

PAGE 53

BIOGRAPHICALSKETCH RahulwillbeginhiscareerinJune2013,asaSeniorMechanicalEngineerwithLutronElectronicsinWestPalmBeach,wherehewillbeamemberofthenewproductdevelopmentteam.Hesharesthecompany'spassionanddedicationtodesignandbuildinnovativeproductsthatcanenrichliveswithoutaectingmothernature.RahulwillgraduatewithaMasterofSciencedegreeinmechanicalengineeringfromtheUniversityofFlorida,GainesvilleinMay2013.SinceOctober2011,hehasbeenaGraduateResearchAssistantintheDistributedControlSystemsLabattheUniversityofFloridaunderDr.PrabirBarooah.HisresearchisfocusedondevelopingsolutionsforthecontrolofHeatingandAirConditioningSystemsinsmartbuildings.Duringhismaster'sprogramhehasbeenaTeachingAssistantformultiplecourses.Hehasalsodoneresearchindataacquisitionusingquadcoptersandconditionmonitoringofmachinetools.RahulearnedhisBachelorofTechnologydegreeinmechanicalengineeringfromAmritaVishwavidyapeethamin2011.HeinternedatBharatHeavyElectricalsLimitedandToyotaKirloskarAutoPartswhilepursuinghisdegree.Hisundergraduatedissertationentitled,ConditionmonitoringofendmillcutterusingArticialIntelligencewassupervisedbyDr.Babudevasenapati. 53