<%BANNER%>

Item-Level Information Visibility

Permanent Link: http://ufdc.ufl.edu/UFE0024066/00001

Material Information

Title: Item-Level Information Visibility An Application of RFID
Physical Description: 1 online resource (95 p.)
Language: english
Creator: Zhou, Wei
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Information Systems and Operations Management -- Dissertations, Academic -- UF
Genre: Business Administration thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Being able to reveal product information at the item-level in a way that is fully automatic, instantaneous, and touchless, radio frequency identification (RFID) is emerging as the hottest information tracing technology in supply chain management. While industry practitioners and academic literature argue that RFID brings value by reducing labor cost, increasing sales, decreasing inventory cost, accelerating physical flow, and improves quality control, they are mostly based on case studies, acknowledging the fact that everything works out well because information visibility eliminates uncertainty. This dissertation investigates the beneficial properties and business applications of item-level information visibility in three different perspective from extant literature review: (1) value of item-level information, (2) knowledge based item-level Manufacturing and (3) item-level information sharing in oligopoly. In the first part, we model the benefits of item-level visibility as the result of reduced randomness, and as a function of the scale of the information system, the distribution of the sample space(s), the control variables and the production functions. This static model is extended for multiple period, which is simulated to verify the generality and robustness of the model. In the second part, we introduce an innovative concept of item-level manufacturing that is backed up by a knowledge-based adaptive learning system. We quantify the potential benefit of such manufacturing scheme. In the third part, we consider a homogeneous product market and the incentive for oligopolists to reveal item-level product information with their customers, by modeling it as a two-stage game. With a constant clearance discount rate, we derive pure strategy equilibria that are subgame perfect and demonstrate that complete information sharing is the unique Nash equilibrium of the game when the common demand is volatile and that no information revelation is the unique Nash equilibria when demand is not volatile. We show that the Nash equilibria is the same with a decreasing clearance discount rate and that neither complete information revelation nor zero information revelation is consistent with an equilibrium with an increasing discount rate. Results are similar in a duopoly non-homogeneous product market scenario.
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 Wei Zhou.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Piramuthu, Selwyn.
Local: Co-adviser: Bandyopadhyay, Subhajyoti.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0024066:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024066/00001

Material Information

Title: Item-Level Information Visibility An Application of RFID
Physical Description: 1 online resource (95 p.)
Language: english
Creator: Zhou, Wei
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Information Systems and Operations Management -- Dissertations, Academic -- UF
Genre: Business Administration thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Being able to reveal product information at the item-level in a way that is fully automatic, instantaneous, and touchless, radio frequency identification (RFID) is emerging as the hottest information tracing technology in supply chain management. While industry practitioners and academic literature argue that RFID brings value by reducing labor cost, increasing sales, decreasing inventory cost, accelerating physical flow, and improves quality control, they are mostly based on case studies, acknowledging the fact that everything works out well because information visibility eliminates uncertainty. This dissertation investigates the beneficial properties and business applications of item-level information visibility in three different perspective from extant literature review: (1) value of item-level information, (2) knowledge based item-level Manufacturing and (3) item-level information sharing in oligopoly. In the first part, we model the benefits of item-level visibility as the result of reduced randomness, and as a function of the scale of the information system, the distribution of the sample space(s), the control variables and the production functions. This static model is extended for multiple period, which is simulated to verify the generality and robustness of the model. In the second part, we introduce an innovative concept of item-level manufacturing that is backed up by a knowledge-based adaptive learning system. We quantify the potential benefit of such manufacturing scheme. In the third part, we consider a homogeneous product market and the incentive for oligopolists to reveal item-level product information with their customers, by modeling it as a two-stage game. With a constant clearance discount rate, we derive pure strategy equilibria that are subgame perfect and demonstrate that complete information sharing is the unique Nash equilibrium of the game when the common demand is volatile and that no information revelation is the unique Nash equilibria when demand is not volatile. We show that the Nash equilibria is the same with a decreasing clearance discount rate and that neither complete information revelation nor zero information revelation is consistent with an equilibrium with an increasing discount rate. Results are similar in a duopoly non-homogeneous product market scenario.
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 Wei Zhou.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Piramuthu, Selwyn.
Local: Co-adviser: Bandyopadhyay, Subhajyoti.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0024066:00001


This item has the following downloads:


Full Text

PAGE 1

1

PAGE 2

2

PAGE 3

3

PAGE 4

MyappreciationandthanksisdirectedtoDr.Piramuthu,Selwynforhisinspirationandhisyearsofpatienceandguidanceofthisdissertation.Withouthimthiswouldnothavebeenpossible.IwouldliketothankmycochairDr.Bandyopadhyay,Subhajyoti,whohasalwaysbeenbymysideoeringcontinualsupportthatfueledmyresearchengineovertheyears.MygratitudealsogoestomyadvisorsDr.PraveenPathakandDr.RichardRomanoforalltheirvaluableadviceandhelpinmyresearch. 4

PAGE 5

page ACKNOWLEDGMENTS ................................. 4 LISTOFFIGURES .................................... 7 ABSTRACT ........................................ 8 CHAPTER 1RFIDANDITEM-LEVELINFORMATIONVISIBILITY ............ 10 1.1Introduction ................................... 10 1.1.1OverviewofRadioFrequencyIdentication .............. 10 1.1.1.1ReviewofAIDCtechnologies ................ 11 1.1.1.2Tags,receiver,&informationsystemsforRFID ...... 14 1.1.2BusinessApplications .......................... 17 1.1.3RecentDebates ............................. 18 1.2ResearchQuestion ............................... 19 1.3LiteratureReview ................................ 21 1.3.1SomeGeneralDiscussions ....................... 21 1.3.2AreaTargetedDiscussions ....................... 23 1.3.2.1Qualitycontrol ........................ 23 1.3.2.2Inventorymanagement .................... 23 1.3.2.3Retailing ............................ 24 1.3.2.4Supplychainmanagement .................. 24 2VALUEOFITEM-LEVELINFORMATIONVISIBILITY ............ 25 2.1TwoExamples ................................. 25 2.2SingleItem ................................... 27 2.3MultipleCases ................................. 31 2.4ControlFunction ................................ 34 2.5MultiplePeriod:DynamicProgrammingApproach ............. 35 2.5.1FiniteTime ............................... 35 2.5.2InniteTime ............................... 37 3KNOWLEDGEBASEDITEM-LEVELMANUFACTURING .......... 39 3.1Introduction ................................... 39 3.2ManufacturingImprovementStrategies .................... 42 3.2.1TotalQualityManagement ....................... 43 3.2.1.1HistoryofTQM ........................ 43 3.2.1.2TQMinmanufacturing .................... 44 3.2.2Just-in-TimeProduction ........................ 45 3.2.3DesignforManufacturability ...................... 46 3.2.4LeanManufacturing ........................... 46 5

PAGE 6

........................... 47 3.3AFrameworkforItem-LevelManufacturing ................. 48 3.3.1TraditionalMassManufacturing .................... 48 3.3.2ManufacturingwithItem-levelInformation .............. 49 3.3.2.1Item-levelproductioncomponent .............. 50 3.3.2.2Knowledge-basedadaptivelearningcomponent ...... 51 3.3.3TaggedItemsinManufacturing .................... 56 3.3.4SimulationAnalysis ........................... 59 3.4Conclusions ................................... 61 4OTHERAPPLICATIONS .............................. 62 4.1CostBenetAnalysis .............................. 62 4.2IncompleteInformation ............................. 63 4.2.1VerticalIncompleteInformation .................... 63 4.2.2HorizontalIncompleteInformation ................... 65 5ITEM-LEVELINFORMATIONSHARINGINOLIGOPOLY .......... 66 5.1Introduction ................................... 66 5.2LiteratureReview ................................ 68 5.3TheModel .................................... 70 5.4DerivationofTheEquilibria .......................... 72 5.4.1ConstantClearanceDiscountRate ................... 73 5.4.2VariableClearanceDiscountRate ................... 75 5.5Non-homogeneousProduct ........................... 78 5.6ConcludingRemarks .............................. 79 APPENDIX AMATLABCODES .................................. 80 BCONSUMERBEHAVIORINITEM-LEVELREVEALMARKET ....... 84 B.1SymmetricDistribution ............................. 84 B.2SkewedDistribution .............................. 86 REFERENCES ....................................... 89 6

PAGE 7

Figure page 1-1Basicbarcodestructure. ............................... 12 1-2Typicalmoduleofmicroprocessorbasedcontactsmartcard. ........... 13 1-3AnRFIDsystem. ................................... 15 1-4Simpliedviewofdatatransferinlow-frequencypassiveRFIDtags. ....... 16 1-5TheRFIDclasses. .................................. 17 2-1Buildinganenginefromenginebodyandenginetop,withdierentspecs. ... 26 2-2Simplestscenario:chooseoneitemfromasetofnpossibilities. ......... 27 2-3Simpliedtwosets2-2example. ........................... 31 2-4MultipleCases .................................... 32 3-1Classicmassproduction ............................... 48 3-2Item-levelmanufacturing ............................... 49 3-3Itemlevelproduction ................................. 50 3-4Adaptiveknowledge-basedLearningframework .................. 52 3-5Benetofhavinginformationvisibilityforfabricationqualitycontrol. ...... 60 4-1VerticalIncompleteInformation. .......................... 63 4-2HorizontalIncompleteInformation. ......................... 64 7

PAGE 8

Beingabletorevealproductinformationattheitem-levelinawaythatisfullyautomatic,instantaneous,andtouchless,radiofrequencyidentication(RFID)isemergingasthehottestinformationtracingtechnologyinsupplychainmanagement.WhileindustrypractitionersandacademicliteraturearguethatRFIDbringsvaluebyreducinglaborcost,increasingsales,decreasinginventorycost,acceleratingphysicalow,andimprovesqualitycontrol,theyaremostlybasedoncasestudies,acknowledgingthefactthateverythingworksoutwellbecauseinformationvisibilityeliminatesuncertainty.Thisdissertationinvestigatesthebenecialpropertiesandbusinessapplicationsofitem-levelinformationvisibilityinthreedierentperspectivefromextantliteraturereview:(1)valueofitem-levelinformation,(2)knowledgebaseditem-levelManufacturingand(3)item-levelinformationsharinginoligopoly.Intherstpart,wemodelthebenetsofitem-levelvisibilityastheresultofreducedrandomness,andasafunctionofthescaleoftheinformationsystem,thedistributionofthesamplespace(s),thecontrolvariablesandtheproductionfunctions.Thisstaticmodelisextendedformultipleperiod,whichissimulatedtoverifythegeneralityandrobustnessofthemodel.Inthesecondpart,weintroduceaninnovativeconceptofitem-levelmanufacturingthatisbackedupbyaknowledge-basedadaptivelearningsystem.Wequantifythepotentialbenetofsuchmanufacturingscheme.Inthethirdpart,weconsiderahomogeneousproductmarketandtheincentiveforoligopoliststorevealitem-levelproductinformationwiththeircustomers,bymodeling 8

PAGE 9

9

PAGE 10

InthischapterwewillgothroughthetechnicalbackgroundofRFIDandsomeofitscompetingtechnologiesintheareaofautomaticidentication.WewilldiscussthebusinessapplicationsandbenetsofthesecontemporarytechnologiesandintroducesomerecentdebatesregardingRFID.Wewillnalizethischapterbyraisingtheresearchquestion,followedbyacomprehensiveliteraturereview. 10

PAGE 11

AstreamofAIDCtechnologiesthatincludesbarcode,RFID,magneticstripes,smartcardsandbiometricsinvolvesaprocessofrecognizingobjects,receivinginformationabouttheobjectsandtransmittingthedataintocomputersystemswithoutanyhumaninvolvement.AnotherstreamofAIDCtechnologiesthatincludesOCR,OMR,voicerecognitionandMachineVisioninvolvesaprocessofobtainingexternaldata,particularlythroughanalysisofimages,soundsorvideos. InordertobetterunderstandRFIDanditsvalue,whichisthemajorissueinthisthesis,wewillbrieyreviewsomeoftherelevantAIDCtechnologies,theircharacteristicsandtheirapplicationsinthenextsection. Barcodesprovideasimpleyetinexpensivemethodofencryptingtextinformation,allowingdatatobecollectedrapidlyandwithextremeaccuracy.Abarcodeconsistsofaseriesofparallel,adjacentbarsandspaces.Symbologies,whicharepredenedbarandspacepatterns,areusedtoencodesmallstringsofcharacterdataintoaprintedsymbol 11

PAGE 12

Basicbarcodestructure. thatcanbethoughtofasaprintedtypeoftheMorsecodewithnarrowbarsrepresentingdots,andwidebarsrepresentingdashes.Therearemanydierentbarcodesymbologies,orlanguages.Eachsymbologyhasitsownrulesforencodingcharacters,printing,decodingrequirements,anderrorchecking.Internationalstandardscoverthecommonuseofbarcode,printqualitymeasurementsandequipment. Thebasicstructureofabarcode(Figure1-1)consistsofaleadingandtrailingquietzone,astartpattern,oneormoredatacharacters,optionallyoneortwocheckcharactersandastoppattern,whichusuallycontainshierarchicalinformationsuchasamanufacturerIDnumberandanitemnumber. 12

PAGE 13

Typicalmoduleofmicroprocessorbasedcontactsmartcard. Figure1-2showsatypicalmoduleofamicroprocessorbasedsmartcardthatconsistsofamicroprocessor,memory,andinterfaces.Thepurposeofhavingamicroprocessoronthesmartcardisforsecurityreasons.Thehostcomputerandcardreaderactually"talk"tothemicroprocessor.Smartcardsmayhaveupto8kofRAM,346kofROM,256kofprogrammableROM,anda16bitmicroprocessor,accordingtointernationalstandards. 13

PAGE 14

14

PAGE 15

AnRFIDsystem. Therearealsosemi-passivetagswherethebatterypowersthemicrochipbutthedevicecommunicatesbydrawingpowerfromthereader.Figure1-5showstheveclassesofRFIDtags.Thereisalsoawidevarietyofshapes,sizesandprotectivehousingsforRFIDtags.Sometagsarewrappedincreditcardsizedpackagesandsomecanbeinjectedbeneaththeanimalskin.Thesmallestdevicescommerciallyavailablemeasure0:4mm0:4mmandarethinnerthanasheetofpaper,whiletheperchipcostcanbeaslowas1ccurrently. 15

PAGE 16

Simpliedviewofdatatransferinlow-frequencypassiveRFIDtags. Chawathe,Krishnamurthy,Ramachandran,andSarma(2004),gure1-6,suggestedalayeredarchitechtureformanagingRFIDdata.ThelowestlayerconsistsofRFIDtags(locatedonobjectssuchascasesandpallets).Thenextlayerconsistsoftagreaders.Theinterfacebetweenthesetwolayersistheso-calledRFIDAirInterfaceandtheRFIDprotocolsforthisinterfacespecifythelow-leveldetailssuchasanti-collisiontechniques(similartothoseusedbyothernetworkingtechnologies).Thethirdlayerofthe 16

PAGE 17

TheRFIDclasses. architectureisresponsibleformappingthelow-leveldatastreamfromreaderstoamoremanageableformthatissuitableforapplication-levelinteractions. Asmartcardisasmallcardcontainingelectroniccircuitsandmemorychipsthatisabletostoreandencryptinformation.Smartcard-enhancedsystemsareinusetodaythroughoutseveralkeyapplications,includinghealthcare,banking,entertainmentandtransportation.Tovariousdegrees,allapplicationscanbenetfromtheaddedfeaturesandsecuritythatsmartcardsprovide.Itcanalsobemadecontact-less.Oneofthe 17

PAGE 18

Radiofrequencyidentication(RFID)isarelativelyoldAIDCtechnologywhichwasrstcommerciallydevelopedin1940s.RFIDhasfounditsimportanceinawiderangeofmarketsrangingfromlivestockidenticationtoAutomatedVehicleIdentication(AVI)systems,becauseofitscapabilitytotrackmovingobjects.RFIDareeectiveinmanufacturingenvironmentswherebarcodelabelscouldnotsurvive.ThemanyRFIDbusinessapplicationincludes:assettracking,manufacturing,supplychainmanagement,retailing,paymentsystems,securityandaccesscontrol.TherearemanyothercreativeusesofRFIDwhilemoreandmorenewRFIDapplicationscomeintosighteveryyear. RTLS,beinganapplicationofRFID,isasystemofndingthepositionofassets,usingactiveRFIDtages.ItsmajorbusinessapplicationisinassettrackingwithanexpectedglobalrevenueexceedingUS$1.6billionby2010(RFIDupdate2005).RFDC,providingsimilarfunctionalitiesasRFID,isnormallyusedtospeeddataacquisitioninreceiving,pick/putaway,inventory,shipping(verication),salesareamanagement,portablepoint-of-sale,andqualitycontrol.DespitethefullyrangedfunctionalitiesofRFDC,it'simpossibletomaintainanRFDCsysteminalargescalebecauseofitshighcost. 18

PAGE 19

19

PAGE 20

It'scriticalforcompaniestoknowtheinstantaneousstatusofitemsinasupplychain,processesitemshavegonethrough,andthehistoryofmovementofitemsduringtransactions.Anitem'sinstantaneousstatusincludesitsuniqueidentity,preciselocation,physicalstatus,andspecialkeyfeatures.Aneectiveandecientinformationtracingsystemenablesacompanytorapidlyinterveneintargetedsituations,consequentlyreducingoperationalcostandincreasingproductivity.Sahin,DalleryandGershwin,(2002)givesalistofpotentialbenetsofRFIDtechnologyonsupplychainprocesses,whichinclude:1.reductioninlaborcosts;2.increaseinstoresellingarea;3.accelerationofphysicalows;4.reductioninprotlosses;5.moreecientcontrolofthesupplychainduetoincreasedinformationaccuracy;6.betterknowledgeofcustomerbehavior;7.betterknowledgeofout-of-stocksituations;8.reductionofdeliverydisputes;9.bettermanagementofperishableitems;10.bettermanagementofreturns;11.bettertrackingofqualityproblems;12.bettermanagementofhumanconsumedproductrecallsandcustomersafety;13.improvedtotalqualitycontrol. TheessencebehindthepassionofRFIDinsupplychainmanagementisitscapabilitytoprovideinformationvisibility.ManyofthebenetsorpotentialbenetsofusingRFIDcanbeexplainedbyincreasedcertainty(reduceduncertainty),directlyresultingfrominformationvisibility.Thisincreasedcertaintyimprovessupplychaincoordination,reducesinventory,increasesproductavailability,improvestotalquality,providesbettermanagementofperishableitemandreturns,andsoon.Inthischapter,weusethefactorofcertaintyasoneofthekeyroletomodelthevalueofRFIDinformationvisibility. AmajorityofliteratureonRFIDdiscussthevalueofRFIDbyusingcasestudyanalysis(Dutta,LeeandWhang2007;Delen,HardgraveandSharda2007;Doeer,GatesandMutty2006;Karkkainen2007;Lee,Peleg,Rajwat,SarmaandSubirana2005). 20

PAGE 21

21

PAGE 22

JoglekerandRosenthal(2005)examinetherationalebehindthe"crawl-before-run"(experimentation)strategyforRFIDadoptionandinvestigatethenecessitytorunexperimentations.Lacy(2005)discussesWal-Mart'spioneeringadoptionofRFIDandthefactofotherretailers'hesitationinRFID.O'Connor(2005)showsimprovedprocesseciencyasthebiggestfactorinuencingthedecisiontodeploytheRFIDtechnology.Sahin,Dallery,andGershwin(2002)provideaframeworkforidentifyingprinciplesandfunctionalitiesofatraceabilitysysteminthecontextofaglobalsupplychain.Usingassessmentcriteriaobtainedbythisanalysis,theyevaluatetheperformanceofbarcodeandRFIDsystems.TheyconcludebydescribingbenetsthatRFIDtechnologycanprovidetosupplychainprocesses,whicharemeasuredasreductionincostsandimprovementinthecustomerservicelevel. 22

PAGE 23

1.3.2.1Qualitycontrol 23

PAGE 24

MostexistingliteratureinRFIDtagsmention/discussthemindetail.However,acomprehensivequantitativemodelofreduceduncertaintyasadirectresultofRFIDinformationvisibilityismissinginextantliterature.Fromanextensivesurveyofliteratureinthisarea,thelackofmentionofthisspecicadvantageleadstothenaturalconclusionthatthiseitherhasnotbeenattemptedattheindustrylevelortheliteraturehasnotcaughtupwithrecentindustrialtrends. 24

PAGE 25

Inthischapterwequantifythebenetofitem-levelinformationvisibilityinascenariowheremultiplefactorsdetermineanoutput.Werstdiscusstwosimpleexamplesinthecontextandmanufacturingandretailingrespectively,followedbyageneralmodel.Theresultsshowthatitem-levelinformationgeneratesnon-negativebenet,whichagreeswithBlackwell'sTheorem(1951).Wendtheupperboundandothercharacteristicsofthebenetfunction. Supposethatanenginebuilderneedstobuildanenginewithasetoftwoenginebodiesandasetoftwoenginetops.Thetwoenginebodieshavethesamepartnumberexceptthatthere'salittledierenceinthesizeofthehosethatlinkstheenginetop.Thetwoenginetopsareofthesamepartnumberexceptthedierenceinthewidthoftheconnectortothebody.Acertainlevelofmatchingduetothesizeoftheconnectorsontheenginebodyandtopyieldsacertainlengthofusagelifeoftheengine.Nowlet'sassumethatthehosesizeonthetwoenginebodiesareH1=10cmandH2=10:2cm;theconnectorwidthonthetwoenginetopsareC1=10:2cmandC2=10:4cm.WealsoknowthatthelifeofusageisL(H1;C1)=L(H2;C2)=10years,L(H1;C2)=L(H2;C1)=9years. Withoutknowingdetailedinformationofeachcomponent,theenginebuilderwouldhavetorandomlychooseonecomponentfromeachset,andtheexpectedlifeofusageoftheengineisE[L]=E[E(HjC)]=9:5years.Ifheknowstheexactspecsofallthe 25

PAGE 26

Buildinganenginefromenginebodyandenginetop,withdierentspecs. components,thelifeofusageoftheengineisL=maxfL(H1;C1);L(H2;C2);L(H1;C2);L(H2;C1)g=10 years.Ifwefurtherassumethatthebuilderneedstobuildtwoengines,thetotallifeofthetwoenginesis19yearswithoutinformationvisibility.Thetotallifeis20yearswithinformationvisibility. InthisexamplewendthatRFIDcanimproveproductqualityfromincreasedinformationvisibilityofmajorcomponents.Nowwewilldiscussanotherexampleinaretailingscenario.Supposethataelectronicsstoresellsanewmodelmp3playerbyplacinganumberoftheplayersonitsself-serviceshelf.Weassumethattheonlyoneclientinthestorehas50%possibilityofbuyingthemp3player.Thestorerestocksbyplacing19unitsofthemp3playersontheshelfeverymorning.Thenumberofremainingunitsisuniformlydistributedoverthesetf0;1;2;;19g.Iftheshelfisempty,theclientwon'tbuythemp3playerandthemanagerwillrestocktheshelfifsheknowsit. WithoutRFIDtagging,theexpectedsalesofthemp3playeris50%19 20=0:475unit.WithRFIDinformationvisibility,theexpectedsalesis0:5unit.Inotherwords,thestorecouldsell0.025moreunitofthemp3playerasabenetofhavingRFID. 26

PAGE 27

Figure2-2. Simplestscenario:chooseoneitemfromasetofnpossibilities. Usingdescriptivestatistics,weassumethatinthesetfXjx1;x2;x3;xng,eachvariablexifollowsdistributionfx(x).Functiony=g(x)denotestheproductionasafunctionofx.ThedistributionofYishencefy(y)=fx(g1(y))g0(y).Weassumethatthedecisionmaker 27

PAGE 28

Knowingthedistributionofthemaximumproduction,wearenowabletoquantifythebenetsofhavingRFIDinformationvisibilityas=~OO,suchthat: where Inmostbusinessenvironmentswhatisinterestingisnotonlythebestoutcome,butalsoasetofthebestoutcomes.Thentheproblemsimplybecomesthesumofthebestkproductionsinrankingstatistics.Assumingthattheunderlyingrandomvariablesx1;x2;:::

PAGE 29

(k1)!(nk)!uk1(1u)nkfy(y)(2{4) Hence,thesumofthebestkis: (i1)!(ni)!ui1(1u)nify(y)dy=ZykXi=1n! (i1)!(ni)!ui1(1u)niyfy(y)dy(denej=i1)=Zynk1Xj=0(n1)!

PAGE 30

Therefore,=Zynk1Xj=0(n1)!

PAGE 31

Theorem3iseasilyprovablebyreplacingkwithnintheoriginalequation.Theorem3meansthatiftheexperimentexhaustsallthepossibilities,theresultwithinformationvisibilityandtheonewithoutvisibilityarenotdierent. Inasimplestcase,n=2andm=2.ThetwosetsarefXjx1;x2gandfYjy1;y2grespectively.Therearefourpossibleoutputsfz1=g(x1;y1);z2=g(x2;y2);z3=g(x1;y2);andz4=g(x2;y1)g.WithoutRFIDinformationvisibility,theexpectedoutputistheaverageoftheoutputsE[Z].Withvisibility,wecanidentifyeachcomponentsandarrangethepair-matchingtomaximizetheoutputmaxfZg. Figure2-3. Simpliedtwosets2-2example. Togeneralizethepreviousexample,let'sassumethatwehaveacaseofmmultiplesetsfXjX1;X2Xng.EachsetXifromfX1jx11;x12;x13x1n1g,fX2jx21;x22;x23x2n2gfXmjxm1;xm2;xm3xmnmghasnisamplesthathaveasamedistribution.Variables 31

PAGE 32

=RRRg(X1;X2Xm)yfX1;X2Xm(X1;X2Xm)dX1dX2dXm. Figure2-4. MultipleCases Withoutinformationvisibility,onesamplefromeachsetischosenrandomlytoproducetheoutcome.Hencetheexpectedoutcomeis: Withinformationvisibility,onesamplefromeachsetischosenifsuchacollectionproducesthemaximumpossibleoutcome,~O=maxfYg.BecauseYhasn1n2nmdierentpossiblevaluesintotal,thedistributionof~Ois It'sbecauseF(maxfYg)=Pr[y1y;y2y;;yn1n2nmy;]=Fy(y)n1n2nm

PAGE 33

where Let'sdeneN=n1n2nm.Iftheproductionincludesthebestkoutcomes,becomes: (k1)!(Nk)!uk1(1u)Nkfy(y)(2{8) Hence,thesumofthebestkis: (i1)!(Ni)!ui1(1u)Nify(y)dy=ZynFbinomial(N1;u)(k)yfy(y)dy

PAGE 34

Proof. Thefunction(X)dependsonthedecisionmaker'sbusinessstrategyanditmayvaryfromcasetocase.TheproductionfunctionhencebecomesY=g((X)). Afterall,weconcludethatthebenetofintroducingRFIDinformationvisibilityisafunctionoftheinformationscale,thedistributionofthesample,thenon-staticfunctionandtheproductionfunction. 34

PAGE 35

Proof. @n=@Ryyu0(nun11)dy @n=Zyyu0(un1+nun1ln(n1))dy0 2.5.1FiniteTime Inmostcasesinabusinessenvironment,suchasSCM,retail,orqualitycontrol,theutilityoveratimeperiodistheaccumulatedsumofutilitiesoverseparatetimesegments,ft0;t1;;tT1;tTg.Nowlet'sassumethatthetotalutilityistime-separable.Thatis:(X0;X1;;XT;v0;v1;;vT1)=0(X0;v0)+1(X1;v1)++T1(XT1;vT1)+S(XT) whereS(XT)isa"scrap"valuefunctionattheendoftheprogram,wherenofurtherdecisionsaremade.Let'sdene()asanintertemporalfunctionthatconnectsthestateandcontrolvariablessuchthatXT=T1(XT1;vT1) 35

PAGE 36

(2{11) subjectto: 1:XTk+1=Tk(XTk;vTk;G(vTk)) (2{12) 2:X0=~X0 3:vTk=Tk(XTk) (2{14) 4:vt2forallt=0;1;;T1 (2{15) Inconstraint4,isthefeasiblesetforthecontrolvariablesthatisassumedtobeclosedandbounded. Nowlet'slookattheaboveprobleminmoredetail,rstfromthelasttimesegmenttT1tTthatisasimple2periodproblem,andthenworkbackwards.Theoptimizationproblemoft0=T1is: maxvT1fT1(XT1;vT1)+S(XT)g(2{16) subjectto: 1:XT=T1(XT1;vT1) (2{17) 2:XT1=~XT1 3:vT1=T1(XT1) (2{19) Constraint3canbesubstitutedintoconstraint1andfurtherbackintotheobjectivefunctiontocharacterizethesolutionasavaluefunction:V(XT1;1)T1(XT1;T1(XT1))+S(T1(XT1;T1(XT1)) 36

PAGE 37

Thegeneralizedvaluefunctionwhent0=tTkis Aftergoingthroughthesuccessiveroundsofsingleperiodmaximizationproblems,eventuallyonereachestheproblemintimezero: subjectto: 1:X1=0(~X0;v0) (2{22) 2:X0=~X0 3:v0=0(~X0) (2{24) BecauseX0isgivenavalueattheoutsetoftheoveralldynamicproblem,wehavenowsolvedforv0asanumberthatisindependentoftheXs.It'seasytoworkoutX1,andhencev1fromthecontrolruleofthatperiod,andthenX2,v2...soonandsoforth.ThisprocesscanberepeateduntilalltheXiandvivaluesareknown.Thentheoverallproblemissolved. 37

PAGE 38

subjectto: 1:Xt+1=0(Xt;vt) (2{26) 2:X0=~X0 3:vt=(Xt) (2{28) whereisthediscountfactorand01.Bydening it'sthesametowriteequation3-25incurrentvalueas: (Sargent1987andStokey1989)showedthattheaboveiterationsstartingfromanyboundedcontinuousW0willcauseWtoconvergeasthenumberofiterationsbecomeslarge.Moreover,,theW()thatcomesoutofthisprocedureistheuniqueoptimalaluefunctionfortheinnitehorizonmaximizationproblem.BecauseW()isuniquelyassociatedwiththecontrolfunction(),ifwendtheoptimalcontrolrulewendthemaximumutilityoverthetime. 38

PAGE 39

RadioFrequencyIdentication(RFID)tagshavegainedwide-spreadpopularityinawidevarietyofapplicationdomains.Theirapplicationinthemanufacturingenvironment,however,stillremainsatalowlevel.SomeoftheimpedimentstoRFIDtag'sinroadinthisdomainincludeitsrelativeandassociatedcost,novelty,andsimplythelackofawarenessofitsbenecialaspects.Weconsidertheconceptofitem-levelinformationinamassmanufacturingcontextbyutilizingaknowledge-basedlearningsystemthatsupportssuchconcept.Weanalyzesomeofthebenetsofthismanufacturingconceptandcompareitwiththoseoftraditionalmassproductionscenario.Preliminaryresultsindicatethatmanufacturingwithitem-levelinformationissignicantlyadvantageouswhenthereisalargevarianceinthemanufacturingprocess.Wealsoshowthatthisbenetisbounded.Anexamplemanufacturingscenarioissimulatedtoverifyresultsobtainedthroughanalysis. AlthoughRFIDtagshavenumerousadvantagescomparedtobarcodes(Raza,Bradshaw,andHague1999;Shepard2005),theexactbenetsofRFIDinamanufacturingenvironmenthasnotbeenclearsinceitsintroductionduringWWII.AmajorityofexistingliteratureonRFIDapplicationsareinthesupplychainmanagementarea.Amajorityoftheseliteraturearecasestudiesorsimulationsofdomain-specicpossibleRFID 39

PAGE 40

Itisbecomingincreasinglycriticalformanufacturerstobeknowledgableaboutanitem'sinstantaneousstatus,theprocessesithasgonethrough,anditshistoryofmovementsduringtransactions.Anitem'sinstantaneousstatusincludesitsuniqueidentity,precisephysicallocation,physicalstatus,andspecialkeyfeatures.Sahin,DalleryandGershwin(2002)providealistofpotentialbenetsofRFIDtechnologyonsupplychainprocessesincluding(1)reductioninlaborcosts,(2)increaseinstoresellingarea,(3)accelerationofphysicalows,(4)reductioninprotlosses,(5)moreecientcontrolofthesupplychainduetoincreasedinformationaccuracy,(6)betterknowledgeofcustomerbehavior,(7)betterknowledgeofout-of-stocksituations,(8)reductionofdeliverydisputes,(9)bettermanagementofperishableitems,(10)bettermanagementofreturns,(11)bettertrackingofqualityproblems,(12)bettermanagementofproductrecallsandcustomersafety,and(13)improvedtotalqualitycontrol. Therehavebeenagrowingnumberofstrategiesthathavebeenaimedatimprovingthegeneralmanufacturingprocesssincethe1980sincludingTotalQualityManagement(Oakland1995),Just-in-Timeproduction(John1989),DesignforManufacturability(Venkatachalam,MellichampandMiller1993),leanmanufacturing(ShahandWard2003), 40

PAGE 41

Weinvestigateaninnovativeconceptofitem-levelmanufacturingdrivenbyaknowledge-basedlearningsupportsystem,enabledbyadvancedIDtracingtechnologysuchasRFID.Traditionalmassmanufacturingisbasedoncertainstandardsthataregenerallyestablishedduringthepre-manufacturingtestingprocess.Thesestandardsarethenappliedtoallproductionactivitiesuntilthereisaneedtoupdatethemduetochangesinworkingenvironmentorsystemmassbias.Itshouldbenotedthatinanymanufacturingprocess,thereisalwaysacertainleveloftolerancethatisconsideredtobewithinanacceptablerange.Clearly,thereisvarianceevenwithinasampleofobjectspassingtheacceptabletolerance-leveltest.Acknowledgingthepresenceofdisparityinmaterialqualityandworkingenvironmentacrosstime,wearguethatthroughutilizationofitem-levelinformationduringthemanufacturingprocess,rmshavethecapabilitytoproducehigherqualityproductsandtherebygenerateincreasedprot. WeconsiderascenariowhereaproductismanufacturedfromseveralRFID-taggedparts.Eachofthesetagscontaininformationaboutthehostpart,includingitsuniqueidentierandotherspecicationsofinterest.Basedonthesespecicationsandsomeperformancecriterion,themostappropriatesetofpartsisselectedtoformthenalproduct.Theproposedknowledge-basedframeworkaidsthisprocessbyutilizingdecisionrulesthatselectcomplementarypartsbasedontheirrespectivemeasuredspecications. TheimpetusbehindRFIDinsupplychainmanagementandmanufacturingarisesprimarilyfromitsinherentcapabilitytoprovideitem-levelinformationvisibility.Its 41

PAGE 42

Thischapterisorganizedasfollows.Abriefbackgroundintroductionofmanufacturingimprovementstrategiesisdiscussedinsection2.Wepresenttheitem-levelproductionframeworkandcompareittoatraditionalmassmanufacturingsettinginSection3.Wemodelthebenetofitem-levelmanufacturinginsection4.Section4alsoincludesresultsfromsimulationtoverifythemodelandtoshowtheeectivenessofthisnewconceptinmanufacturing.Section5concludesthischapterwithabriefdiscussionontheinsightsgarneredandtheirimplications. 42

PAGE 43

InJapan,TQMresultedinsuchmanagerialinnovationsasqualitycircles,equitycircles,supplierpartnerships,cellularmanufacturing,just-in-timeproduction,andhoshinplanning.AmericanrmsbegantotakeseriousnoticeofTQMaround1980,whensomeU.S.policyobserversarguedthatJapanesemanufacturingqualityhadequaledorexceededU.S.standards,andwarnedthatJapaneseproductivitywouldsoonsurpassthatofAmericanrms.Productivitytrendssupportedtheseassertions,leadingsomeopinionleaderstopredictthat-barringaradicalchangeinAmericanmanagementpractices-JapanandotherAsiancountrieswouldsoondominateworldtradeandmanufacturing,relegatingtheU.S.tosecond-tiereconomicstatus.Somehigh-proleAmericanrmssuchasFord,Xerox,andMotorolawereeasilyconvinced,havingalreadylostmarketsharetomore 43

PAGE 44

Itisimportanttorecordnotjustthemeasurementranges,butwhatfailurescausedthemtobechosen.Inthatway,cheaperxescanbesubstitutedlater(say,whentheproductisredesigned)withnolossofquality.OnceTQMhasbeeninuse,it'sverycommonforpartstoberedesignedsothatcriticalmeasurementseitherceasetoexist,orbecomemuchwider. Ittookpeopleawhiletodevelopteststoidentifyemergentproblems.Onepopulartestisa"lifetest"inwhichthesampleproductisoperateduntilitfails.Anotherpopulartestiscalled"shakeandbake"inwhichtheproductismountedonavibratorinanenvironmentaloven,andoperatedatprogressivelymoreextremevibrationand 44

PAGE 45

ThephilosophyofJITissimple-inventoryisdenedtobewaste.JITinventorysystemsexposethehiddencausesofmaintaininginventoryandarethereforenotasimplesolutionacompanycanadopt;thereisaradicallynewwayinwhichthecompanymustoperateinordertomanageitsconsequences.Theideasincludedinthiswayofthinkingoriginatefrommanydierentdisciplinesincludingstatistics,industrialengineering,productionmanagementandbehavioralscience.IntheJITinventoryphilosophythereareviewswithrespecttohowinventoryislookedupon,whatitsaysaboutthemanagementwithinthecompany,andthemainprinciplebehindJIT. QuickcommunicationoftheconsumptionofoldstockwhichtriggersnewstocktobeorderediskeytoJITandinventoryreduction.Thissaveswarehousespaceandcosts.Howeversincestocklevelsaredeterminedbyhistoricaldemand,anysuddendemandrisesabovethehistoricalaveragedemandwilldepleteinventoryfasterthanusualandcausecustomerserviceissues.SomehavesuggestedthatrecyclingKanbanfastercanalso 45

PAGE 46

Leanproductionisamulti-dimensionalapproachthatencompassesawidevarietyofmanagementpractices,includingjust-in-time,qualitysystems,workteams,cellularmanufacturing,suppliermanagement,etc.inanintegratedsystem.Thecorethrustofleanproductionisthatthesepracticescanworksynergisticallytocreateastreamlined,highqualitysystemthatproducesnishedproductsatthepaceofcustomerdemandwithlittleornowaste. 46

PAGE 47

Masscustomizationcanbedenedeitherbroadlyornarrowly.Thebroad,visionaryconceptpromotesMCastheabilitytoprovideindividuallydesignedproductsandservicestoeverycustomerthroughhighprocessagility,exibilityandintegration.MCsystemsmaythusreachcustomersasinthemassmarketeconomybuttreatthemindividuallyasinpre-industrialeconomies.MCsystemsarepositionedbelowthemaindiagonalofHayesandWheelwright'sproduct-processmatrix,i.e.havingmediumtohigh-volumeprocesstypessuchasmanufacturingcellsorassemblylinesthatareabletodeliverthehighproductvarietiesusuallyassociatedtofunctionalorxed-typeoperations. TheydeneMCasasystemthatusesinformationtechnology,exibleprocesses,andorganizationalstructurestodeliverawiderangeofproductsandservicesthatmeetspecicneedsofindividualcustomers(oftendenedbyaseriesofoptions),atacostnearthatofmass-produceditems.Inanycase,MCisseenasasystemicideainvolvingallaspectsofproductsale,development,production,anddelivery,formingafull-circlefromthetimeofcustomerselectingorplacingtheordertoreceivingthenishedproduct. ThejusticationforthedevelopmentofMCsystemsisbasedonthreemainideas.First,newexiblemanufacturingandinformationtechnologiesenableproductionsystemstodeliverhighervarietyatlowercost.Second,thereisanincreasingdemandforproductvarietyandcustomization(evensegmentedmarketsarenowtoobroadastheynolongerpermitdevelopingnichestrategies).Finally,theshorteningofproductlifecyclesand 47

PAGE 48

3.3.1TraditionalMassManufacturing Figure3-1. Classicmassproduction Partsthatformthenalproductsfollowstandardsobtainedduringthetestingphaseandaretreateduniformlyatthemassproductionstage.Acknowledgingthequalityorspecicationdisparityinsimilarparts,wendthatignoringuniqueitem-levelspecicationinformationresultsinaninferiorproductionprocesscomparedtothescenariowhereitem-levelinformationisconsideredinthemanufacturingprocess.Weanalyticallyshowthisinsection3.ReduceduncertaintyasadirectresultofRFIDitem-levelinformationvisibilitybringsdirectimprovementinmanufacturingaswediscussinthefollowingsection. 48

PAGE 49

Figure3-2. Item-levelmanufacturing Theconsideredadaptiveknowledge-basedframeworksupportsadaptivedecision-makingtochangesintheuserpreferences,thequalityoftheproductanditscomponents,andtheenvironment.Thedynamicsofsuchachangemaybesuchthatsomeareamenabletoaproactivestanceorevenareactivestancefromadecision-makingperspective. 49

PAGE 50

Figure3-3. Itemlevelproduction TheProblemSolvercomprisesasetofdecisionsupporttoolsthatcomputeanddeliversolutionstoroutinestructuredproblemswhereallnecessaryinputsaredeterministicallyknowntofairlysophisticated`intelligent'toolsthatpro-activelyseektoprovideappropriatesupportformakingdecisionsinsemi-structuredorevenunstructuredenvironments.Dynamicenvironmentsthatareessentiallycharacterizedbyuncertaintiesinseveraldimensionsnecessitateareasonably`smart'decisionsupporttool.Thesedecisionsupporttoolsarerequiredtoprovideorassistingenerating`good'decisionsinreal-time. Problem-solvingcapabilityisanessentialcharacteristicofanadaptiveknowledge-basedsystemsinceitisarequirementforsupportingdecision-makingsituations.Comparedtohumans,therelativespeedatwhichcomputersareabletosolveproblemsaremeasuredinmultipleordersofmagnitude.Thisisbeneciallyutilizedintheconsideredadaptive 50

PAGE 51

EssentialcharacteristicsoftheProblem-solvingcomponentincludetheabilitytoupdateitsknowledge-baseusinginputfromtheAdaptiveLearningcomponent,appropriatelyinvokingnecessaryknowledgefromitsknowledge-baseandusingittoaddresstheinputdecisionmakingproblemfromtheenvironmentwiththeProblem-solver,andprovidethemostappropriatesolutionoutputforagivencombinationofexistingknowledgeandproblemofinterest. TheProblemSolveridentiesandimplementsmanufacturingdecisionsthatincludecomponentmatching,processadjustment,environmentsettingandmachineadjustmentwhichcoverthefourfacetsof4M1E-Man,Machine,Method,MaterialandEnvironment.Byconsideringtheminor,albeitsignicant,variationsfromcorespecicationsofinstancesofanitem,themanufacturerisabletomakeaccuratedecisionstoimproveoutputperformancewiththeexactspecicationsofindividualitem-levelinformation. 51

PAGE 52

Adaptiveknowledge-basedLearningframework Thiscomponentisresponsibleforpro-activelykeepingtheknowledge-basefrombecomingstale.Thisisprimarilydonethroughindirectlymonitoringthequalityofthe 52

PAGE 53

Thetwosetsofinputthatthissub-componentreceivesincludesolutiontothedecision-makingproblem,whichisessentiallythequalityofthemanufacturedproduct,anditsspecicationsincludinganyrelatedtolerancefactors.Theseinputsaremappedtodetermineanydeviationsthatareaddressablethroughmodicationstotheknowledge-base.Theknowledge-baseisthenincrementallymodiedtoreectthisadditionalknowledge.Whenthedeviationsareduetoafreakcircumstance(e.g.,themachinerybeinginterruptedduetosomeunforeseenreasons),asolutionaddressingthisdeviationmayormaynotbeincorporatedintheknowledge-base.Therationalebehindthisissimplythefactthattheknowledge-baseneedstobecompactforittorespondinstantaneously,andanyirrelevantorunnecessaryinformationdoesnotwarrantbeingincorporatedintheknowledge-base. 53

PAGE 54

Althoughlearningbyitselfcanbeaccomplishedthroughseveralmeans,wefocusonmachinelearningasthemodeoflearningintheconsideredframework.Theprimarymotivationbehindthisisthenaturalandseamlesswayinwhichsuchalearningcanbeincorporatedwiththerestoftheframeworktoachieveimprovedperformanceresults.Anyoftheseveralexistingsupervisedmachinelearningalgorithmssuchasdecisiontrees,decisionrules,feed-forwardneuralnetworks,geneticalgorithms,etc.couldbeusedinthiscomponent.Dependingonthedomainofinterest,morespecicallyonthedatacharacteristicsincludingdatatypes(e.g.,numeric,alphanumeric,symbolic)andinteractionsamongthemselvesinthedomainofinterest,anappropriatealgorithmcanbeselected.Forexample,somealgorithmssuchasthosethatareusedinfeed-forwardneuralnetworkworkbetterwithreal-valueddata,whilesomeotherssuchasthoseusedininducingdecisiontreesworkbetterwithsymbolicdataingeneral. 54

PAGE 55

BeingapartoftheAdaptiveLearningcomponent,theLearningsub-componentextensivelyinteractswiththePerformanceEvaluationsub-component.TheinteractionbetweentheLearningandPerformanceEvaluationsub-componentsisiterativeastheyarebothsynergisticallyrelatedtogether.OutputfromthePerformanceEvaluationsub-componentdeterminesandtriggers,toagreatextent,thetimingandextentoftheLearningsub-componenttoaccomplishitsgoalsoflearningthemostappropriateknowledgeinatimelymanner.EssentialcharacteristicsoftheLearningsub-componentincludetheabilityto(1)concisely,accurately,andquicklylearntheconceptsofinterest,(2)acceptnecessaryinputdata,and(3)generatelearnedconceptsinaformthatisrequiredofthenextcomponentintheframework.ThePerformanceEvaluation 55

PAGE 56

Thefollowingexampleillustratesthisscenarioinamanufacturingcontext.ConsiderthecaseofanLCDpanel,whichcomprisesatleastthreecomponentsincludingbase,liquidcrystal,andcover.AssumethatamanufacturingplanthasveassemblylinesthatproducethesametypeofLCDpanels.Moreover,assumethatalltheindividualcomponentsmeetcertainbase-levelqualitycriteria,whichisgivenbythefollowinglist. 0. datessincemanufacturing<365days 0. percentageofsurfaceatness>90% 0. materialpurity>94% 0. dateofmanufacturing<365days 0. materialpurity>97% 56

PAGE 57

0. dateofmanufacturing<730days 0. percentageofsurfaceatness>97% 0. materialpurity>90% WithoutRFIDtags,allthefactorycandoistorandomlypickaninstanceofeachcomponentandassembleanLCDpanelononeofthevemanufacturinglines.WithRFIDitem-levelinformationvisibility,however,thefactoryhasmoreinformationabouteveryindividualsamplecomponentsuchthatabasesamplecarriesinformationlikeBase(life=180days;S:F:=97%;M:P:=95%).Themanufacturerisnowabletodistinguishamongtheindividualmixofcomponentsthatgointoproducinganalproduct.Thisitem-levelinformationandthemeanstoutilizesuchinformationinthemanufacturingprocessdidnotexistbeforetheintroductionofRFIDtagtechnology.WithRFIDtaggedcomponents,aknowledge-basedsystemcanidentifypatternsthatgenerallyresultinbetterquality,fasterthroughput,lesswastage,etc.Thesepatternsthatresultinabetterproductcanbeacombinationofcertainparameters,orexclusionofcertainparameters,suchasQuality8>>>>>>>>>>>>>><>>>>>>>>>>>>>>:>9:5=10;ifBase(L<180days;S:F:>97%;M:P:>95%)<6=10;ifBase(L>270days;S:F:<95%;M:P:<94:5%)&Cover(L>540days;S:F:<98%;M:P:<92%)>7=10&<9:5=10;ifBase(L>180days;S:F:>97%;M:P:<95%) Knowingtheserulesandtheitem-levelinformation,themanufacturerthusisabletoimproveproductqualitybypro-activelypursuingdecisionsthroughappropriatedecisionaidssuchasanadaptiveknowledge-basedsystem. Wemodelthebenetsofthisself-learningitem-levelmanufacturingschemecomparingtotheclassicmassproductionthatiswithoutsuchinformationandintelligence. 57

PAGE 58

Let'sconsidermpartsfXjX1;X2Xmgthatjointlydetermineaproduction.EachpartXiconsistsofnisamplesthatsharethesamedistributionsuchas:fX1jx11;x12;x13x1n1g,fX2jx21;x22;x23x2n2gfXmjxm1;xm2;xm3xmnmg.VariablesX1;X2XmfollowjointdistributionfX1;X2Xm(X1;X2Xm).ProductionfunctionisdenedasY=g(X1;X2Xm;)withcdf: Thesumofthebestkproductionis: (i1)!(Ni)!ui1(1u)Nify(y)dy =ZynFbinomial(N1;u)(k)yfy(y)dy Hencethebenetofintroducingitem-levelRFIDinformationvisibilityisafunctionoftheinformationscale,thedistributionofthesample,andtheproductionfunction,suchthat: (3{4) =Zy(NFbinomial(N1;u)(k)k)yfy(y)dy Basedonthisstudy,weobservethefollowingforthisdomain: Result2.Thebenetofimplementingitem-levelmanufacturingisupperboundedlooselybyRy(nk)yfy(y)dyandboundedtightlybyRy(ne2(nk)2

PAGE 59

Result4.Thebenetofimplementingitem-levelmanufacturingmonotonicallyincreaseswithmanufacturingscale. Result5.Ifalltheresourcesareused,thebenetofhavingitem-levelinformationvisibilityiszeroifthere'sonlyonepart;thebenetispositiveiftherearemultipleparts. 59

PAGE 60

Benetofhavinginformationvisibilityforfabricationqualitycontrol. Weassumethatinthefunction=(k;X;G(X)),eachfactorsbeingdenedas:X1UniformorNormalwithvariance1X2UniformorNormalwithvariance2G(x1;x2)=ex1x2 60

PAGE 61

Weconsidereditem-levelinformationprovidedbyincorporationofRFIDtagsinamanufacturingsetting,andshowedthatitcanbebeneciallyusedtoimproveproductionquality.Whereaspreviouslymanufacturerswereonlyabletoutilizeclass-levelinformationformanufacturingandassembly,RFIDtagsenableitem-levelmanufacturingwithmorefocusontheidiosyncrasiesofeachindividualitemasitrelatestootheritemsthatareaggregatedtogethertoformanalproduct.Althoughthetraditionalmanufacturingprocessresultedinnishedproductsthatsatisedtoleranceconstraintsforeachspecication,item-levelvisibilityenablesimprovingtheirquality.Moreover,therearesituationswheretwo(or,more)partsthatarewellwithinacceptablespecicationtolerancelevelscouldresultinanishedproductthatisunacceptableduetomismatchamongtheindividualparts.Suchsituationsleadtounnecessarywastageof`good'partsthatcouldhavebeensalvagedotherwisewiththepresenceofitem-levelinformation.Duetoitsrecentpopularity,RFIDtagsandtheirapplicationsinamanufacturingsettinghasnotreceivedmuchattentionfromresearchers.Ourstudyisastepinthisdirectionofreducingsuchwastageaswellasimprovingtheoverallqualityofmanufacturedaswellasassembledproducts. 61

PAGE 62

InthisChapter,weextendthecurrentdiscussiontopotentialresearchapplications,suchasthecostbenetanalysisandtheproblemofincompleteinformationcoverage. =(n)c(n) (4{1) =Zyyu0(nun11)dycnb Bytakingtherstorderderivativeonn,@=@n,weareabletondtheoptimalscaleoftheinformationsystemthatwouldbringthemaximumbenetofhavingitem-levelinformationvisibility. Ifnln(n1)=c1ornn1=ec1,@ 62

PAGE 63

Theapplicationsofincompleteinformationnotonlyappliestomanufacturing.Inchapter5,weshowthatinthecontextofcongestedmarket,thereexistsuniqueNashequilibriaforoligopoliststostrategicallyrevealpartialoftheinformationforahomogeneousproduct(verticalincomplete)orforsubstitutableheterogeneousproduct(horizontalincomplete). VerticalIncompleteInformation. Herewedescribetheincompleteinformationproblemwhennoteverycomponentinthebusinesschainisinformationtraceable.Theinformationincompletenessmaybecausedbythefactthatnotallupper-streampartnersareequippedwithRFIDinfrastructuresothatonlyaportionoftheproductsacquiredhaveinformationvisibility(Figure4-1).Informationincompletenessmayalsobecausedbythefactthatsomeofthecomponentsdon'thaveinformationvisibilityhorizontallyinthesystem(Figure4-2). 63

PAGE 64

Figure4-2. HorizontalIncompleteInformation. fromSection3,wendtheabsolutevalueofpartialinformationinahorizontalonecomponentexampleas where~ndenotesthemagnitudeofavailableinformation.Let'srecallthatnisthemagnitudeoforiginalinformation,includingbothvisibleandinvisibleinformation.Thenn~nisthesizeoftheinvisibleinformation.Byassumingthattheorderstatisticsofthesampleissymmetricwendthatthevalueofinformationvisibilitycomesfromthersthalfof~n.Itimpliestheoverallbenetofhavinginformationvisibilityisthesameasequation(33)ifk~n 64

PAGE 65

wherethegiaresingle-dimensionalfunctions,thekiarepositiveweighting(scaling)constants,andlargervaluesofgaremorepreferable. Withunknowninformationthatwedenoteas~x,thedierencebetweenfullinformationrevelationandtheincompleteinformationisY~Y=XXj2~XkXi=1kjE[g[Xj;i:n]] wherekXi=1E[Ai:n]=ZAnFbinomial(n1;u)(k)AfA(A)dA 65

PAGE 66

Weconsiderahomogeneousproductmarketandtheincentiveforoligopoliststoshareitem-levelproductinformationwiththeircustomers.EnabledbyRFIDtechnology,eachrmhastheoptiontorecordandrevealitem-levelinformationofaproportionofitsproduct.Eachrmrstdecidesitsproductionplanandthendecidesitslevelofinformationrevelationinatwo-stagegame.Withaconstantclearancediscountrate,wederivepurestrategyequilibriathataresubgameperfectanddemonstratethatcompleteinformationsharingistheuniqueNashequilibriumofthegamewhenthecommondemandisvolatileandthatnoinformationrevelationistheuniqueNashequilibriawhendemandisnotvolatile.Furthermore,weshowthattheNashequilibriaisthesamewithadecreasingclearancediscountrateandthatneithercompleteinformationrevelationnorzeroinformationrevelationisconsistentwithanequilibriumwithanincreasingdiscountrate.Resultsaresimilarinaduopolynon-homogeneousproductmarketscenario. Inthischapterweinvestigatetheincentivesforitem-levelinformationsharingfromrmstoconsumersinamarketwithduopolistssellinghomogeneousproducts.Incontrasttoearlierpapers(NovshekandSonnenschein1982,vives1984andGal-or1985)whereinformationsharingisaboutunknowncommondemand,or(Gal-or1986)where 66

PAGE 67

RFID(RadioFrequencyIDentication)isatrackingsystemthatusestags(siliconchipsimplantedinaproductoritspackaging)tocommunicatewithareader.RFIDtagscanbeusedtostoreandretrieveproductinformationatanitem-levelinawaythatisfullyautomatic,instantaneous,andtouchlessandcouldbeusedtotrackanyobjectfromcandybarstobigscreenTVs.Despiteitslimitedprocessingpowerandstoragecapacity,theitem-levelinformationitcanstoreisdramaticallyhigherthanthoseusingcompetingtechnologiessuchasbarcode(Raza,Bradshaw,andHague1999;Shepard2005).Unlikebarcodethatprovidescategorical-levelinformation,RFIDtechnologyfacilitatesdistinguishingindividualinstancesofproductsbyassigningauniqueelectronicproductcode(EPC)toeachofthem. Retailersandmanufacturersarenowabletoprovideitem-levelinformationforalmostanyproduct.However,theyarefacedwithrealisticissuessuchas1.whethertheywillbenetfromsharingsuchinformation,2.ifsotowhatextent,and3.whatstrategyshouldbefollowedtosharethisitem-levelinformationwhenvariousoligopolistscompetewithahomogeneousproduct. Ahomogeneousproductisdenedasaproductfromanindustryinwhichoutputsfromdierentrmsareindistinguishable,however,notwohomogeneousproductsareindeedthesame.Moreprecisely,theterm"homogeneous"describesagroupofproductsthatfollowacertainstatisticalcriteria,whichispre-denedbyanindustrytoconvenemarketingsales.Qualityvarianceofdierenthomogeneousproductsdiersbyawidemargindependingontheindustry.Ifitem-levelproductinformationisavailable,itbringsadditionalvaluetothebuyerbyprovidingtheopportunitytochoosebetterproductsandmakingbettercontroldecisions(Zhou2008).Otherbenetsincludelocation 67

PAGE 68

Armmightenjoyabettermarketpositionbyvirtueofprocessingtheabilitytorevealitem-levelproductinformationbysellingmoreofitsabove-averageproductsthanhiscompetitors.Thisdynamicisespeciallysalientwhensupplyexceedsanticipatedcommondemand.Inthisresearchweassumethatrmsdon'tmanipulateorselectivelyrevealinformationalthoughit'sdoableandbenecialtothermsundercertaincircumstances(CrawfordandSobel1982;Greezy2005).Abuyeralwaysbenetsfromadditionalinformation(Blackwell1953)aordedbythechoicetopickthebestaboveaverageproducts. Thischapterisorganizedasfollows.Section2presentsabriefoverviewofrelevantliterature.Section3containsinformationaboutthemodeldescription,assumptionsandsetup.Derivationofequilibriaappearsinsection4,alongwithsomeanalysisanddiscussion.Section5concludesthechapterwithabriefdiscussionontheinsightsgarneredandtheirimplications. CrawfordandSobel(1982)developamodelofstrategicinformationtransmissioninwhichaninformedagenttransmitsinformation(possiblynoisy)totheprincipalwhotakesanactionthatdeterminesthewelfareofboth.Theyshowthattheprincipal'sequilibriumexpectedutilityriseswhentheagent'spreferencesaremoresimilar,assuming 68

PAGE 69

Weconsiderstrategicitem-levelproductinformationtransmissionfromthermtotheconsumer,facilitatedbymoderntracingtechnologysuchasRFID.Mostliteratureinoligopolygametheorywithstrategicinformationtransmissiondealwithinformationassociatedwithnon-publicinformationoncommondemand(Gal-or1985)orunknownprivatecosts(Gal-or1986).Weassumethattheinformationsenderdoesn'tmanipulatetheinformationnordoesheselectivelychooseinformationwithinacertainrangeinfavorofhim.KrishnaandMorgan(2001)studythemodelofexpertiseinwhichperfectlyinformedexperts,whoarebiased,transmitinformationtoadecisionmakerwhoseactiondecidethewelfareofallandshowthattheexpertwithholdssizableinformationfromthedecisionmakerinaone-expertscenario.Greezy(2005),inhisdiscussionofdeceptionininformationtransmission,showsthattheaveragepersonprefersnottolieandbydoingsoincreaseshispayoonlybyalittlebutgreatlyreducestheother'spayo.Inthisresearchwedon'tconsiderthemoralgamesthathavebeenstudiedbefore. TheresearchquestioninthischapterisenabledbyRFIDthatasanemergingtracingandidenticationtechnologyhasnumerousadvantagescomparedtotraditionalbarcodes(Raza,Bradshaw,andHague1999;Shepard2005).However,theexactbenetsofRFIDinretailingandsupplychainmanagementhasn'tbeenveryclearsinceitsintroductionduringWWII.MostexistingliteraturesintheareaofRFIDapplicationsarecasestudiesorsimulationsofdomain-specicpossibleRFIDimplementations,mostlyintheeldsofinventorymanagementandreplenishment,supplychainoperations,andretailing.LeeandOzer(2005)investigatethevalueofRFIDinasupplychain.Dutta,Lee,andWhang 69

PAGE 70

Thepriordistributionofe,whichisindependentandwithmeanzero,isknowntobothrms.emayfollowadierentdistributionaccordingtothemarketcharacteristicsofinvolvedindustry.Q=Q1+Q2,whereQi;i=1;2denotesthequantityproducedbyrmi. Atthebeginningofatimeperiod,eachrmdecidestheproductacquisition/manufacturingplanonquantity:Qiandtheportionoftheproductswithitem-levelinformationrevealed:i.ThenumberoftaggedunitsequalsiQandthenumberofunitswithouttagsequals(1i)Q.Tagswithitem-levelinformationcannolongerbeplacedoncetheproductisaftermanufacturing(acquisition).Weassumethatthecostoftaggingnegligible.ItisareasonableassumptiongiventheunitcostofRFIDtobe6ccomparedtoabottleof$10shampooora$500computer.Wealsoassumethatallthebuyersarerationalandbuythebestproductthatisavailable. Attheendofthetimeperiod,theactualdemandisrealizedandbothrmsmayover-sellorunder-sell.Ifarmhasunsoldproduct,hiQiwherehidenotestheproportion 70

PAGE 71

Thegameinourmodelconsistsoftwostage.Attherststageeachrmmakesitsmanufacturingoracquisitionplan.Atthesecondstageeachrmdecideshowmuchitem-levelinformationshouldberevealed.Thelevelofinformationrevelationischosendependentupontheoutputplanandthedistributionofthecommondemand.WederivepurestrategyNashequilibriathataresubgameperfectandinvestigatetheincentivestoshareitem-levelproductinformationinpossiblescenariossuchasnon-constantdiscountingrateandnonsymmetricproductqualitydistribution. Theincentiveforrevealingitem-levelinformationareinvestigatedwhenthevolatilityofcommondemandrangesfromextremevolatiletofreezinglystable.Wedemonstratethatcompleteinformationrevelation(i=1;i=1;2)isadominantstrategywhendemandisvolatileandnoinformationsharing(i=0;i=1;2)isadominantstrategywhendemandisnotvolatile. 2.Productsarehomogeneousandhavethesamestatisticalcharacteristics; 3.Atthebeginningofatimeperiod,eachplayeri;i=1;2decidestheproductacquisition/manufacturingplanonquantity:Qiandtheportionoftheproductswithitem-levelinformationrevealed:i,sothenumberoftaggedunitsequalsiQandthenumberofunitswithouttagsequals(1i)Q.Tagswithitem-levelinformationcannolongerbeplacedoncetheproductisacquired/manufactured.ProductperunitcostisdenotedasCi.Costoftaggingisassumedtobenegligible. 4.ThepriceissetasafunctionofexpectedaccumulateddemandP=APiQiwithunsoldproductdiscountconsidered.Accumulateddemandinatimeperiodisstatistically 71

PAGE 72

5.Attheendofatimeperiod,ifthereareunsoldproducts,unsoldproductsqi;i=1;2willbeclearedoutwithadiscount; 6.Consumerbehavior:Consumersbuythebestaboveaverageproductwithknown-item-levelinformationrst.Ifalltheaboveaveragetaggedproductsaresoldout,consumersbuytheuntaggedproductsuntiltheyaresoldout.Ifboththeaboveaveragetaggedandtheuntaggedproductsaresoldout,consumerswillbuythebestofthebelowaveragetaggedones; wherehi(Qi;i)denotestheproportionofplayeri'sunsoldproductanddenotestheclearancediscountrate.Theunsoldproportionofrmi'soutputis (Qii+Qjj)I.(Q1+Q2)1 2(1Q1+2Q2)jQjiQj+i(Qi+Qj) 2(1Q1+2Q2)(Q1+Q2)(1i)(Qi+Qj)+1 2Qj(ij) (1i)Qi+(1j)QjII.otherwise(5{4) 72

PAGE 73

(hi)Qi Firmichoosesitsdecisionrulei()andsubsequentlyQi()tomaximize(3),givendecisionruleschosenbytheotherrms.Thequantity/informationpairf(Q1;1);(Q2;2)gisaNashequilibriumifforeachrmi,(Qi;i)solves max0Qi<1;0i1i[(Qi;i);(Qj;j)] (5{6) =max0Qi<1;0i1[Qip(1)hi(Qi;i)Qi][A(Qi+Qj)]CiQi Theorem7. 2,theuniqueNashequilibriumofthetwostagegameisi=0andQi=A Proof. UnderconditionI: 2(1Q1+2Q2)(5{9) Playeri'sunsoldportionofhisproductis: (Qii+Qjj)(5{10) (Qii+Qjj)2>0(5{11) Thepayofunctionisstrictlydecreasingfunctionofi.HenceunderconditionI,i=0isadominantstrategyforeachrm. UnderconditionII: 2(1Q1+2Q2)(Q1+Q2)1 2(1Q1+2Q2)(5{12) 73

PAGE 74

2Qj(ij) (1i)Qi+(1j)Qj(5{13) 2Qj+1 2Qi] [(1i)Qi+(1j)Qj]2(5{14) Thepayofunctionisstrictlydecreasingfunctionofiif<1 2.HenceinconditionII,i=0isalsoadominantstrategyforeachrm. TherstorderconditiononQiis@[[Qip(1)Qi](AQiQj)CiQi] Throughasimilarlyprocedureforrmjwehave: Therefore, (5{17) (5{18) AccordingtotheTheorem7noinformationrevelationisadominantstrategyforeachrmwhenthevolatilityofthecommondemandislow. 2,theuniqueNashequilibriumofthetwostagegameisi=1andQi=A Proof. When1 2,therstorderconditiononiisstrictlydecreasing,accordingto(12).Hencei=1isadominantstrategyforeachrm. 74

PAGE 75

2(1Q1+2Q2)(Q1+Q2)(5{19) Playeri'sunsoldportionis: (iQi+jQj)2(5{21) therstorderconditiononQiis@[[Qip(1)Qi](AQiQj)CiQi] Similarlyforrmj Qj=AQi Therefore, (5{24) (5{25) Theorem8showsthatcompleteinformationrevelationisadominantstrategyforeachrmwhenthevolatilityofthecommondemandishigh. Theorem9. 2,if@ @h0,theuniqueNashequilibriumofthegameis1=2=0;if@ @h>0,thereexistsnoequilibriumwhere1=2=2[0;1]. Proof. 75

PAGE 76

wherep0,A(Qi+Qj)0andhi0.When<1 2,wehaveh0i>0andhi0.Sotheequilibriaoninformationsharingisi=j=0. Iftheclearancediscountrateincreaseswiththequantityofunsoldproduct,f=khi:2(0;1)g.Therstorderconditiononiis Andthesecondorderconditionis (Qii+Qjj)2>0@2hi (Qii+Qjj)4<0 Thus,hi=1 2kmaximizesrm'spayofunction.Nowlet'sassumethereexistsanequilibriumandhi6=,thenhi=1 2kisthenecessaryconditionforauniqueequilibriumforrmi.Followingthesameprocedurewecanobtainthenecessaryconditionforrmj.ThenwehavehiQi+hjQj=1 2k(Qi+Qj)=(Qi+Qj),whichsignies1 2k=orhi=,whichcontradictstheassumption. Incontrasttothecasewhentheclearancediscountrateisdecreasing,completelyrevealproductinformationorrevealnothingisnotnecessarilyadominantstrategywhenclearancediscountrateisanincreasingfunctionoftheproportionofunsoldproduct. 2,if@ @h0,theuniqueNashequilibriumofthegameis1=2=1;if@ @h>0,thereexistsnoequilibriumwhere1=2=2[0;1].

PAGE 77

2,h0i<0,sotherstorderderivativeonthepayofunctionispositiveif@ @h0,accordingto(24).Therefore,theuniqueequilibriaisi=j=1. When>1 2,wendthat@i[(Qi;i);(Qj;j)]=@i=0if 2k(5{29))jQjiQj+i(Qi+Qj) 2k)i=(12k)jQj 2kcontradictstheexistenceofauniqueNashequilibrium. @h>0,neithercompletenorzeroinformationrevelationisconsistentwithanequilibrium. Proof. ~hi=8>>>><>>>>:ii(Qi+Qj) (Qiii+Qjjj)I.(Q1+Q2)1 2(1Q1+2Q2)jjQjiiQj+ii(Qi+Qj) 2(1Q1+2Q2)(Q1+Q2)(1ii)(Qi+Qj)+1 2Qj(iijj) (1ii)Qi+(1jj)QjII.otherwise(5{30) 77

PAGE 78

Thepayofunctionofrmiis: FirmichoosesitsdecisionrulePi()andsubsequentlyi()tomaximize(29),givendecisionruleschosenbytheotherrms.Theprice/informationpairf(P1;1);(P2;2)gisaNashequilibriumifforeachrmi,(Pi;i)solves max0Pi<1;0i1i[(Pi;i);(Pj;j)] (5{33) =max0Qi<1;0i1[Qip(1)hi(Pi;i)]PiCiQi Proof. 78

PAGE 79

Sincehi(i)followsthesameformulathatwefoundinprevioussection,wecanprovethesametheoremswithnon-homogeneousproductinasimilarproceduredescribedbefore. Theanalysispresentedhereleavesunansweredmanyinterestingquestionsintheeldofitem-levelinformationrevelationandsharing.Oneinterestingprobleminsignallingisarm'sbestinformationtransmissionstrategyifitcanselectivelyrevealinformationtomaximizeitsownutility. Asanimmediateextensiontothischapter,weareworkingonthegamesofinformationrevelationinahorizontallyandverticallydierentiatedmarket.Thisresearchalsohaswideapplicationsinsupplychainmanagementwhenunsoldmerchandizeareusuallystoredininventoryforuseinthenextperiodratherthansimplybeingclearedoutatadiscount. 79

PAGE 84

2(1Q1+2Q2) Playeri'sunsoldportionofthehisproductis: (Qii+Qjj)(B{1)@hi=i=(Qii+Qjj)(Qi+Qj)Qii(Qi+Qj) (Qii+Qjj)2=Qjj(Qi+Qj) (Qii+Qjj)2>0 (Qii+Qjj)2=iQj(ji) (Qii+Qjj)2 2(1Q1+2Q2)(Q1+Q2)1 2(1Q1+2Q2) playeri'sunsoldportioncanbedescribedas:hi="(1i)Qi(Qi+Qj)1 2(iQi+jQj) 2iQi#=Qi=(1i)(Qi+Qj)1 2(1i)jQj+1 2i(1j)Qj 2Qj(ij) (1i)Qi+(1j)Qj

PAGE 85

2Qj) [(1i)Qi+(1j)Qj]2+(1i)Qi(Qi+Qj)+1 2QiQj(ij) [(1i)Qi+(1j)Qj]2=(1j)Qj(Qi+Qj)+(1j)Qj1 2Qj+1 2QiQj(1j) [(1i)Qi+(1j)Qj]2=(1j)Qj[(Qi+Qj)+1 2Qj+1 2Qi] [(1i)Qi+(1j)Qj]2>0,if<1 2<0,if>1 2=0,if=1 2 2(1Q1+2Q2)(Q1+Q2)Playeri'sunsoldportioncanbedescribedas:hi=QiiQi(1)(Qi+Qj) (iQi+jQj)2=jQj[Qj+(Qi+Qj)Qi] (iQi+jQj)2=jQj[(1+)(Qi+Qj)] (iQi+jQj)2<0 85

PAGE 86

(iQi+jQj)2=i(jQj(jQjiQj+iQj) (iQi+jQj)2=iQj(1+)(i+j) (iQi+jQj)2 2(1Q1+2Q2) Playeri'sunsoldportionofthehisproductis: (Qii+Qjj+i+j)(B{2)@hi=i=(Qii+Qjj)(Qi+Qj)Qi(i+i)(Qi+Qj) (Qii+Qjj)2=(QjjQii)(Qi+Qj) (Qii+Qjj)2>0 (Qii+Qjj)2=iQj(ji) (Qii+Qjj)2 2(1Q1+2Q2)(Q1+Q2)1 2(1Q1+2Q2) 86

PAGE 87

2(iQi+jQj) 2iQi#=Qi=(1i)(Qi+Qj)1 2(1i)jQj+1 2i(1j)Qj 2Qj(ij) (1i)Qi+(1j)Qj@hi=i=[(1i)Qi+(1j)Qj]((Qi+Qj)+1 2Qj) [(1i)Qi+(1j)Qj]2+(1i)Qi(Qi+Qj)+1 2QiQj(ij) [(1i)Qi+(1j)Qj]2=(1j)Qj(Qi+Qj)+(1j)Qj1 2Qj+1 2QiQj(1j) [(1i)Qi+(1j)Qj]2=(1j)Qj[(Qi+Qj)+1 2Qj+1 2Qi] [(1i)Qi+(1j)Qj]2>0,if<1 2<0,if>1 2=0,if=1 2 2(1Q1+2Q2)(Q1+Q2)Playeri'sunsoldportioncanbedescribedas:hi=QiiQi(1)(Qi+Qj)

PAGE 88

(iQi+jQj)2=jQj[Qj+(Qi+Qj)Qi] (iQi+jQj)2=jQj[(1+)(Qi+Qj)] (iQi+jQj)2<0@hi=Qi=(iQi+jQj)ii(jQjiQj+i(Qi+Qj)) (iQi+jQj)2=i(jQj(jQjiQj+iQj) (iQi+jQj)2=iQj(1+)(i+j) (iQi+jQj)2

PAGE 89

[1] Abell,P.andQuirk,C.2002.\ePC/RFIDandItsImminentEectontheSupplyChain,"AMRResearchReport. [2] Alexander,K.,Birkhofer,G.,Gramling,K.,Kleinberger,H.,Leng,S.,Moogimane,D.,Woods,M.2002.\Focusonretail:ApplyingAuto-IDtoimproveproductavailabilityattheretailshelf,"WhitePaperAuto-IDcenter. [3] Alexander,K.,Gilliam,T.,Gramling,K.,Grubelic,C.,Kleinberger,H.,Leng,S.,Moogimane,D.,Sheedy,C.2002.\ApplyingAuto-IDtoReduceLossesAssociatedwithShrink,"MITAuto-IDCenterWhitePaperIBMBusinessConsultingServices. [4] AndersonConsulting.1996.\Wheretolookforincrementalsalesgains:Theretailproblemofout-of-stockmerchandise,"TheCoca-ColaRetailingResearchCouncil. [5] Atali,A.,Lee,H.L.andOzer,O.2004.\IftheInventoryManagerKnew:ValueofRFIDunderImperfectInventoryInformation,"RWorkingPaperStanfordUniversity. [6] Bellman,R.E.2003.\DynamicProgramming,"DoverPublications,Incorporated Blossom,P.2005.\LevelsofRFIDmaturity,"RFIDJournalJan.(1). [8] Cachon,G.,M.Fisher.2000.\Supplychaininventorymanagementandvalueofsharedinformation,"ManagementScience46(8). [9] Chappell,G.,Durdan,D.,Gilbert,G.,Ginsberg,L.,Smith,J.andTobolski,J.2002a.\Auto-IDintheBox:TheValueofAuto-IDTechnologyinRetailStores,"MITAuto-IDCenterWhitePaper. [10] Clark,R.2004.\RFIDForecastUpdatedafterallthoseMandates,"RFID.comMarch16. [11] Clark,R.2004.\InventoryTopsRFIDBenetsListinWeb-BasedPool,"RFID.comSeptember16. [12] DeHorantius,N.,Mersereau,A.,Schrage,L.2006.\Retailinventorymanagementwhenrecordsareinaccurate,"ForthcominginManufacturing&ServiceOperationsManagement. [13] Delen,D.,B.C.Hardgrave,R.Sharda.2007.\RFIDforbettersupply-chainmanagementthroughenhancedinformationvisibility,"ProductionandOperationsManagement16(5)613-624. [14] DoC.2005.\Radiofrequencyidentication:Opportunitiesandchallengesinimplementation,"DepartmentofCommerce1-38. [15] Doer,K.H.,Gates,W.R.,Mutty,J.E.2006.\AhybridapproachtothevaluationofRFID/MEMStechnologyappliedtoordinanceinventory,"InternationalJournalofProductionEconomics1-16. 89

PAGE 90

Dutta,A.,Lee,H.L.,Whang,S.2007.\RFIDandoperationsmanagement:Technology,valueandincentives,"ProductionandOperationsManagement16(5). [17] Economist.2003.\TheBestThingsincetheBar-Code,"Feb6. [18] Ferrer,G.,M.E.Ketzenberg.2004.\Valueofinformationinremanufacturingcomplexproducts,"IIETransactions36265-277. [19] Fleisch,E.,Tellkamp,C.2005.\Inventoryinaccuracyandsupplychainperformance:Asimulationstudyofaretailsupplychain,"InternationalJournalofProductionEconomics95(3)373-385. [20] Gaukler,G.M.,Seifert,R.W.,Hausman,W.H.2007.\Item-levelRFIDintheretailsupplychain,"ProductionandOperationsManagement16(1)65-76. [21] Gavirneni,S.,R.Kapuscinski,S.Tayur.1999.\Valueofinformationincapacitatedsupplychains,"ManagementScience45(1)16-24. [22] Hardgrave,B.C.,Miller,R.2006.\ThemythsandrealitiesofRFID,"InternationalJournalofGlobalLogisticsandSupplyChainManagement1(1)1-16. [23] Heese,H.S.2007.\Inventoryrecordinaccuracy,doublemarginalizationandRFIDadoption,"ProductionandOperationsManagement16(5)542-553. [24] Iglehart,D.L.,andMorey,R.C.1972.\InventorySystemswithImperfectAssetInformation,"ManagementScience18B388-B394. [25] Inderfurth,K.2004.\Analyticalsolutionforasingle-periodproduction-inventoryproblemwithuniformlydistributedyieldanddemand,"CentralEuropeanJournalofOperationsResearch12117-127. [26] Inderfurth,K.2005.\IncorporatingdemandandyielduncertaintyinadvancedMRPsystems,"LogistikManagement-InnovativeLogistikkozepteDeutscherUniversitaets-Verlag,Wiesbaden,Germany. [27] Jogleker,N.,S.Rosenthal.2005.\ExperimentationwithRFIDusageinsupplychains,"POMSChronicle12(1)10-11. [28] Kang,Y.,Gershwin,S.B.2005.\Informationinaccuracyandinverntorysystems:Stocklossandstockout,"IIETransactions37(9)843-859. [29] Karaer,O.,Lee,H.L.2007.\ManagingthereversechannelwithRFID-enablednegativedemandinformation,"ProductionandOperationsManagement16(5). [30] Karkkainen,M.(J.W.).2007.\IncreasingeciencyinthesupplychainforshortshelflifegoodsusingRFIDtagging,"InternationalJournalofRetail&DistributionManagement31(10)529-536. 90

PAGE 91

KurtSalmonAssociatesInc.1993.\EcientConsumerResponse:EnhancingConsumerValueintheGroceryIndustry,"FoodMarketingInstitute,WashingtonD.C.. [32] Lacy,S.2005.\RFID:PlentyofMixedSignals,"BusinessWeekOnlineJan31. [33] Lee,H.L.,Peleg,B.,Rajwat,P.,Sarma,S.andSubirana,B.2005.\AssessingtheValueofRFIDTechnologyandtheEPCStandardforManufacturers,"EPCGlobalWhitePaper. [34] Lee,H.L.,So,R.,andTang,C.S.2000.\ValueofInformationSharinginaTwo-LevelSupplyChain,"ManagementScience46(5)626-643. [35] Lapide,L.2004.\RFID:What'sinitfortheforecaster,"TheJournalofBusinessForecastingMethods&Systems23(2)16-19. [36] Larson,J.,Bradlow,E.,andFader,P.2005.\SAnexploratorylookatsupermarketshoppingpaths,"InternationalJournalofResearchinMarketing22,395-414. [37] Lee,H.L.,O.Ozer.2007.\UnlockingthevalueofRFID,"ProductionandOperationsManagement16(1)40-64. [38] McFarlane,D.,She,Y.2003.\Theimpactofautomaticidenticationonsupplychainoperations,"TheInternationalJournalofLogisticsManagement14(1)1-17. [39] METROGroup.2004.\RFID:BenetsforRetailersandSuppliersthroughEectiveDeploymentandRoll-Out,"Report. [40] Moradpour,S.,Bhuptani,M.2005.\RFIDeldguide:Deployingradiofrequencyidenticationsystems,"SunMicrosystemsPressNewYork,NY. [41] Murphy-Hoye,M.,H.L.Lee,J.B.Rice.2005.\Areal-worldlookatRFID,"SupplyChainManagementReview9(5)18-26. [42] Ngai,E.,W.T.,Cheng,C.E.,Lai,K.H.,Chai,Y.F.Choi,Y.S.,Sin,R.K.Y.2007.\DevelopmentofanRFID-basedtraceabilitysystem:Experiencesandlessonslearnedfromanaircraftengineeringcompany,"ProductionandOperationsManage-ment16(5)554-568. [43] O'Connor,M.C.2005.\EcienciesdriveRFIDadoption,"RFIDJournal. [44] O'Connor,M.C.2006.\RFIDgetspersonal,"RFIDJournal. [45] Pisello,T.2004.\TheThreeRsofRFID:Rewards,RiskandROI,"Technology-Evaluation.comSep28. [46] Rekik,Y.,Sahin,E.,Dallery,Y.2005.\AnalysisofthebenetsofAuto-IDtechnologyinimprovingretailshelfavailability,"TechnicalReportLGI-EcoleCentraleParis,France. 91

PAGE 92

Sahin,E.,Dallery,Y.andGershwin,S.2002.\Performanceevaluationofatraceabilitysystem.Anapplicationtotheradiofrequencyidenticationtechnology,"Systems,ManandCybernetics,2002IEEEInternationalConferenceonOct. [48] Song,S.,andZipkin,P.1996.\Inventorycontrolwithinformationaboutsupplyconditions,"ManagementScience421409-1419. [49] Sorensen,H.2003.\Thescienceofshopping,"MarketingResearch1530-35. [50] Trybula,W.J.1994.\Buildingsimulationmodelswithoutdata,"ProceedingsoftheIEEEInternationalConferenceofSystems,Man,andCybernetics209-214. [51] Want,R.2004.\RFID:Akeytoautomatingeverything,"ScienticAmericanJan56-65. [52] Ware,L.C.2004.\Theimpactofradiofrequencyidenticatioinonsupplychainfacilities,"JournalofFacilitiesManagement3(3)226-239. [53] Whitacker,J.,Mithas,S.,Krishnan,M.S.2007.\AeldstudyofRFIDdeploymentandreturnexpectrations,"ProductionandOperationsManagement16(5). [54] Yano,C.A.,Lee,H.L.1995.\Lotsizingwithrandomyields:Areview,"OperationsResearch43(2)311-334. [55] Zipkin,P.H.2000.\FoundationsofInventoryManagement,"McGrawHill,Boston2000. [56] Roberts,C.M.2006.\Radiofrequencyidentiction(RFID),"Computers&Security2518-26. [57] Landt,J.2005.\ThehistoryofRFID,"IEEEPotentials24(4)8-11. [58] Sarma,S.,Engels,D.2002.\TheReaderCollisonProblem,"2002IEEEInterna-tionalConferenceonSystems,2. [59] Sarma,S.2001.\Towardstheve-centtag,"TechnicalReportMIT-AUTOID-WH-006Auto-IDLabs. [60] Want,R.2006.\AnIntroductiontoRFIDtechnology,"PervasiveComputing,IEEE5(1)25-33. [61] Angeles,R.2005.\RdTechnologies:Supply-ChainApplicationsandImplementationIssues,"InformationSystemsManagement22(1)51-65. [62] Weinstein,R.2005.\RFID:ATechnicalOverviewandItsApplicationtotheEnterprise,"ITProfessional7(3)27-33. [63] Garnkel,S.andRosenberg,B.2005.RFID:Applications,Security,andPrivacyAddison-WesleyProfessional 92

PAGE 93

Chen,L.andNath,R.2004.\Aframeworkformobilebusinessapplications,"InternationalJournalofMobileCommunications2(4)368-381. [65] Inoue,S.,Hagiwara,S.andYasuura,H.2006.\SystematicerrordetectionforRFIDreliability,"ARES'06:ProceedingsoftheFirstInternationalConferenceonAvailability,ReliabilityandSecurity,280-286. [66] Li,S.,Visich,J.,Khumawala,B.andZhan,C.2006.\Radiofrequencyidenticationtechnology:applications,technicalchallengesandstrategies,"SensorReview26(3)193-202. [67] Kay,E.2003.\WhatsthenextstepforRFID,"FrontlineSolutions21-25. [68] Avoine,G.,Dysli,E.andOechslin,P.2005.\ReducingTimeComplexityinRFIDSystems,"Proceedingsofthe12thAnnualWorkshoponSelectedAreasinCryp-tography291-306. [69] Wang,F.andLiu,P.2005.\TemporalmanagementofRFIDdata,"Proceedingsofthe31stinternationalconferenceonverylargedatabases1128-1139. [70] Gonzalez,H.,Han,J.,Li,X.andKlabjan,D.2006.\WarehousingandanalysismassiveRFIDdatasets,"22ndInternationalConferenceonDataEngineering83. [71] Chawathe,S.S.,Krishnamurthy,V.,Ramachandran,S.andSarma,S.2004.\ManagingRFIDdata,"Proceedingsofthe30thVLDBconferenceToronto,Canada. [72] Sargent,T.1987.DynamicMacroeconomicTheory,HarvardUniversityPress. [73] Stokey,N.,Lucas,R.E.,andPrescott,E.1989.Recursivemethodsforeconomicdynamics,HarvardUniversityPress. [74] Carlsson,C.,T.Jelassi,andP.Walden.\IntelligentSystemsandActiveDSS."HICSS,5,1998,pp.4-8. [75] Culberson,J.C.,\OntheFutilityofBlindSearch:AnAlgorithmicViewof`NoFreeLunch'."EvolutionaryComputation,6,1998,pp.109-127. [76] Igel,C.,andM.Toussaint.,\OnClassesofFunctionsforwhichNoFreeLunchResultsHold."InformationProcessingLetters,86(6),2003,pp.317-321. [77] Simon,H.A.TheNewScienceofManagementDecisions.HarperandRow,NewYork,1960. [78] Wolpert,D.H.,andW.G.Macready,\NoFreeLunchTheoremsforSearch."TechnicalReportSFI-TR-05-010,SantaFeInstitute,SantaFe,NewMexico,1995. [79] John,M.1989.\LevelSchedulesforMixed-ModelAssemblyLinesinJust-in-TimeProductionSystems."ManagementScience,35(2),pp.192-207 93

PAGE 94

OaklandJ.S.1995.\Totalqualitymanagement."FuelandEnergyAbstracts,36(4),pp.304-304(1) [81] Shah,R.andWard,P.T.2003.\Leanmanufacturing:context,practicebundles,andperformance."JournalofOperationsManagement,21(2),pp.129-149 [82] Silveira,G.D.,Borensteinb,D.andFogliatto,F.S.2001.\Masscustomization:Literaturereviewandresearchdirections."InternationalJournalofProductionEconomics,72(1),pp.1-13 [83] Venkatachalam1,A.R.,Mellichamp,J.M.andMiller,D.M.1993.\Aknowledge-basedapproachtodesignformanufacturability."JournalofIntelligentManufacturing,4(5),pp.355-366 [85] E.Koer,Z.W.Kmietowicz,A.D.Pearman.1984.\DecisionMakingwithLinearPartialInformation(L.P.I.)."TheJournaloftheOperationalResearchSociety,35(12),pp.1079-1090 [85] CraigW.KirkwoodandRakeshK.Sarin.1985.\RankingwithPartialInformation:AMethodandanApplication."OperationsResearch,35(1),pp.38-48 94

PAGE 95

95