




版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
ExtensibleandScalableTimeTriggeredSchedulingfor
AutomotiveApplicationsWeiZhengAdvisor:ProfessorAlbertoSangiovanni-VincentelliAgendaMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkProblemModelMathematicalModelImplementationAlgorithmEvaluationProjectMotivationHardReal-timeEmbeddedSystemsareubiquitouslyusedtodayinsafetycriticalcommercialapplicationsVerificationofcomplexsystemsistimeandresourceintensiveForfasttime-to-marketExtensibleandScalablesystemsPowerTransmissionUnit-6-linesperday-3000ppmresidentialdefects-5monthsvalidationtimeFABIOROMEO,Magneti-MarelliDAC,LasVegas,June20th,2001X-by-wireDesignFlowFunctionalModelSoftwareModelPhysicalArchitectureModelMappingControlalgorithmdesignPlantModeldesignFaultModelFunctionalSimulationAllocateFunctiontoTasksTaskandtheirWCETSignalsMiddlewareOSAllocatingtaskstoECUAllocatingsignalstoBUSECUarchitectureNetworkarchitectureSimulationcapturingcomputationalconstraintsTTbehavioralsimulationFunctionalityArchitectureSchedulingVirtualPrototypeRefinementINPUTPROCESSINGSYSTEMCOORDINATIONINPUTFAULTMANAGEMENTDISTRIBUTEDCONTROLAGREEMENTBY-WIRECONTROLNODESTATEOFHEALTHOUTPUTPROCESSINGOUTPUTFAULTMANAGEMENTTASKATASKCTASKATASKBTASKDSOURCE:GMFunctionality:AllocateFunctiontoTaskSystemArchitecture...ECUAPP.TTTASKSComm.controllerBUSDRIVEROSEKSCHEDULERBUSDRIVEROSEKTIMEDISPATCHERFT.TTTASKSAPP.OSEKTASKS......14012345681079111213STATICSEGDYNAMICSEGSYMBOLWINNITCOMMUNICATIONCYCLE14012345681079111213...AgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkIdentifyasetofmetricstocaptureextensibilityandscalabilityApplythesetofmetricsinadesignEvaluatetheeffectivenessofthesetofmetricsSpecifically,wewantto:StudyahardrealtimeembeddedsystemsintheautomotivedomainFocusontheschedulingaspectofsystemdesignCharacterizeextensibilityandscalabilityinschedulingApplythesetofmetricsinaschedulingalgorithmEvaluatetheeffectivenessoftheapproachwithindustrialcasestudyIdentifyaSetofMetricsFormallyDescribeMetricApplyMetricsToDesignEvaluateResult.MetricsProblemProblemStatementAgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkPreviousWork
StaticcyclicschedulinghasbeenextensivelyresearchedClassicalschedulingtheoryusemetricssuchasMinimizingsumofcompletiontimeMinimizingschedulelengthMinimizingresourceForrealtimesystems,deadlineisaddedasaconstraintEmphasisshiftedtofindingfeasiblesolutionswhileMinimizingend-to-enddelayMinimizingcommunicationcostClosestproblemconceptcomesfromPaulPop,etalClosestproblemformulationcomesfromArminBender,etalPreviousWork
PaulPop,etal,wroteaboutincrementaldesignUselistschedulingapproachtoobtainavalidscheduleUseaheuristictodistributeslackinthesystemMissingseveralimportantcomponents
PreemptionisnotconsideredResultingscheduleisnotsuitableforfuturetaskwithurgentdeadlineMessageslackisnotdistributedExtensibilityisnotconsideredArminBender,etal,usedmathematicalprogrammingformappingandschedulingWorkismotivatedbysoftware-hardwareco-designObjectiveistoobtainschedulefeasibilitywhileMaximizingPerformanceMinimizingresourceResearchDirectionFocusonoptimallyutilizeredundanciesinschedulesforextensibilityandscalabilityIdletimeandslacksaretraditionallyincorporatedinhardrealtimeembeddedsystemsschedulestoincreasesystemrobustnessWeshouldutilizetheseredundanciesto:TolerateincrementaldesignchangesAccommodatenewtaskstobeaddedinfutureproductupdatesD12_2D34D12_1T2_1T2_2T4T1_2T3T1_1Time01234BusECU2ECU1T5_1T5_2Idle[ECU1,T5_2]IdleTimeDataSlacksSlack[D12_2,T2_2]AgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkExtensibilityMotivation
ToleratechangesofTaskWCETToleratechangesofDataWCTTImplementationMaintainBusSchedule
Maintainnon-involvedECU
schedules
MaintaininvolvedECU
scheduleswithoutreconfigurationApproach
Messageleft&RightslackMaxSumofallslacks
MinVarianceofallslacksScalability
AccommodateNEWtasksby
staticallyschedulingthemon
alegacysystem
Provideblocksofcomputation
timeforfuturecomputation
intensivetasks
Provideporosityinschedulesto
allowforfuturetaskswithtight
deadlinesECUidletimedistributionBusidletimedistributionEvenlydistributeallidle
timeModelCapturetheMetricsApplyingtheMetricsDevelopaformalrepresentationoftheproblemusingmathematicalprogrammingandsolveitusingexistingsolverModelingLanguage:AMPL
Advantage:obtainoptimalsolution
.costfunctionDisadvantage:computationallyintensive
suitableonlyformoderately
sizedproblemsAssumptions:HardrealtimedeadlinesStaticallyscheduledtaskswithdatadependencyDistributedandheterogeneousmulti-processorarchitectureTimetriggeredbuswithTDMAprotocolPreemptionallowedonECUswithnolevellimitsMulti-ratetasksupportwithadaptivetaskgraphexpansionFixedtaskallocationwithnotaskmigration
MathematicalModelMathematicalFormulation1Notations
ThesetofTasksThesetofECUThesetoftaskpairwithdatadependencyrunningonthesameECUThesetoftaskpairwithdatadependencyrunningondifferentECUThesetoftasknon-reachabletaskpairrunningonthesameECUThesetoftaskpairrunningonthesameECUThesetoftaskallocationforECUMathematicalModelMathematicalFormulation2
WCETReleaseTimePeriodIdletimeStartingtimeFinishingtimeTask:6-tupleparametervariableVector
WCTTLeftSlackRightSlackStartingtimeFinishingtime
Message:5-tupleparametervariableVector
ParametersandVariablesMappingfromTaskstoECUsTaskandMessageiftaskiismappedtoECUjotherwiseMathematicalModelMathematicalFormulation3ifthestartingtimeoftaskiprecedethestartingtimeoftaskjotherwiseiftaskiisnotpreemptedbytaskjotherwiseParametersandVariables(continue)IdletimeandIntegerVariablesifdatatransmittedfromtaskitotaskjprecedesdatatransmittedfromtaskktolotherwiseIdletimefortaski,ifiisnotthefirsttaskinofitsrunningECUFirstidletimeforeachECUkIdletimeforECUkbeforethesuperperiodMathematicalModelMathematicalFormulation4Subjecttothefollowingconstraints:ReleaseandDeadlineConstraints
ExecutionTime/TransmissionConstraintsPrecedenceConstraintsNon-negativeandIntegerConstraintsMathematicalModelMathematicalFormulation5Constraints(continued):MutualexclusionconstraintsIdletimeconstraintsMathematicalModelAgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkMultipleObjectiveCostFunctionExtensibility
ScalabilityJointlyExtensibility&ScalabilityAlgorithmD34T5_2D12_2T2_2T1_2T4T5_1T3T2_1T1_1Time01234BusECU2ECU1D12_1ExtensibleScheduleWCETChangesWCETIncreaseT6D34D12_2T2_2T1_2T4T3T2_1T1_1T5_1D12_1Time01234BusECU2ECU1ScalableSchedule:AddNewTaskT5_2T2_2D12_2T1_2T5_2T4T3D12_1T2_1T1_1Time01234BusECU2ECU1T5_1D34MinimizeEndtoEndLatencyTaskgraphexpansion(inaSUPERperiod)T1T4T3functionalityECU2ECU1FlexRayarchitectureMappingT5T1T3T2T1T2T4T5T5T2_2D12_2T5_2D34T4T3D12_1T2_1T1_1Time01234BusECU2ECU1T5_1T1_2CheckingSchedulabilityD34D12_2T2_2T1_2T4T3T2_1T1_1Time01234BusECU2ECU1T5_1T5_2D12_1ScalableScheduleT6T2D34T5_2D12_2T2_2T1_2T4T5_1T3T2_1T1_1Time01234BusECU2ECU1D12_1ExtensibleScheduleD34D12_2T2_2T4T1_2T2_1D12_1T3T1_1Time01234BusECU2ECU1T5_1T5_2JointlyConsideringExtensible
andScalableAlgorithmExtensibilityandScalabilityofTimeTriggeredScheduling
AgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationSummaryConclusionFutureWorkAdvancedAutomotiveControlApplicationDesiredSpeedCurrentSpeedObjectDistanceandSpeedCurrentthrottlepositionDesiredbrakingforceDesiredThrottlepositionActuatebrakesActuateThrottleLeft-RearWheelspeedActuateThrottleActuateBrakeDesiredBrakingForceLateralaccelerationHand-wheelpositionYawRateRight-FrontWheelSpeedRight-RearWheelSpeedLeft-FrontWheelSpeedRoad-wheelforceHand-wheelpositionDesiredhand-WheeleffortDesiredroad-WheelangleForcefeedbackTodriverActuatesteeringRackmotorT10T14T12T13T11T8T7T9T24T23T22T21T19T20T18T17T16T15T2T6T5T4T3T1ApplicationsandcorrespondingtaskgraphrepresentationsAdaptiveCruiseControlTractionControlElectricPowerSteeringArchitectureandTaskAllocationP1T4,T11P3T10,T20P2T3,T12,T22FlexRayThrottlepositionSteeringRackForceObjectDistanceBrakeThrottleValueHand-WheelmotorSteeringRackMotorAccel-aratorHand-WheelangleCarSpeedRFWheelSpeedRRWheelSpeedLFWheelSpeedLRWheelspeedProcessorswithtasksallocatedcSensorsaredirectlyconnectedtothebusActuatorsaredirectlyconnectedtothebusImplementationDescribetheMetricsFormalizetheMetricsGetSchedulingResultEvaluateResult.MetricsCasestudyAutomaticAMPLdatafilegenerationAMPLmodelwithcostfunctionandconstraintsCPLEXsolverAutomaticGantgraphgenerationT2_2D12_2T1_2T5_2T4T3D12_1T2_1T1_1Time01234BusECU2ECU1T5_1D34.Self-developedprojectinfrastructure.Off-the-shelf
projectinfrastructureImplementationInfrastructureAgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationSummaryConclusionFutureWorkCostFunctionEvaluationMulti-objectivecostfunctionisanabstractionMathematicalprogrammingformulationhaslimitedsemanticsExtensibilityandscalabilitymetricsaretoocomplexDescribedinfull,thecostfunctionwouldbetoocomputationallyexpensiveMustdetermineifthecostfunctionabstractioneffectivelyrepresentsthemetricsUsetheresultsofCPLEXsolverExtractrealslackandidletimedistributionsbasedonprecisedefinitionofthemetricsCompareresultswiththeschedulewithoutextensibilityandscalabilityoptimizationEvaluationTraditionalSchedulingResultOptimizingforEndtoEndLatencyEvaluationOptimizedSchedulingResultOptimizingforExtensibilityandScalabilityEvaluationMetricsEvaluationOursetofmetricsisoneabstractionoftheextensibilityandscalabilityconceptMustdetermineifourmetricseffectivelyhandlesincrementaldesignchangesIncrementalDesignScenario:BasicACCStop-N-GoACCAdditionofanewAdaptiveCruiseControlfeaturePredictdesiredspeedbasedon:DigitalmapinformationForwardlookingvisionsensorRequiresadditionoftasksandmessagesSomeexistingtaskswillneedmorecomputationtimeEvaluationAdaptiveCruiseControlIncrementalDesignChanges:AddnewDigitalMapComputationtaskonP1MorecomplexalgorithminT10(DesiredSpeedControl)DesiredSpeedControlrequiresnewinputfromHandWheelSensorDesiredThrottleControlrequiresnewinputfromForwardVisionSensorEvaluationT10T14T12T13T11T8T7T9T_addT19DesiredSpeedCurrentSpeedObjectDistanceandSpeedCurrentthrottlepositionDesiredbrakingforceDesiredThrottlepositionActuatebrakesActuateThrottleDigitalMap
ComputationHandWheel
PositionTraditionalScheduleInTraditionSchedule:IncrementalchangesimpossiblewithoutfullreschedulingEvaluationOptimizedScheduleInTraditionSchedule:IncrementalchangesimpossiblewithoutfullreschedulingInOptimizedSchedule:AlotmoreporositytoaccommodatenewtasksandmessagesEvaluationOptimizedScheduleInTraditionSchedule:IncrementalchangesimpossiblewithoutfullreschedulingInOptimizedSchedule:AlotmoreporositytoaccommodatenewtasksandmessagesNewfunctionsadded:
WithoutdisturbinglegacyschedulesEvaluationAgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkConclusionSuccessfullycapturedextensibilityandscalabilitymetricsRecognizedimplicationsinacceleratingtime-to-marketofembeddedsystemdevelopmentReducere-verificationburdeninincrementaldesignflowNoincreaseinresourcerequirementsFormulatedtheschedulingproblemasamathematicalprogrammingproblemConstructedmulti-objectcostfunctionsabstractedfromthemetricsThecostfunctionisshowntobeeffectiveforthemetricsThemetricsisshowntobeeffectiveinindustrycasestudyAgendaProjectMotivationProblemStatementPreviousWorkInvestigativeApproachMetricDefinitionMathematicalFormulationMulti-ObjectiveCostFunctionCaseStudySystemDescriptionCostFunctionEvaluationMetricsEvaluationConclusionFutureWorkFutureWorkProtocolComparisonFlexRayVs.TTPSlotSizeOptimization...14012345681079111213......6012435...COMMUNICATIONCYCLECOMMUNICATIONCYCLESlotSizeExplorationRead/Write
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 颅内压力监测芯片行业技术发展策略研究
- 纺织企业生产设备安全操作规程
- 绿色环保猫砂行业深度解析
- 2025至2030中国自行车部件行业发展趋势分析与未来投资战略咨询研究报告
- 文化遗产保护工程质量保证措施
- 健康养生:中草药产品开发策略
- 2025至2030中国膳食纤维糖果行业市场发展分析及商业模式与投融资报告
- 2025至2030中国脑膜炎奈瑟菌感染药物行业产业运行态势及投资规划深度研究报告
- 2025至2030中国能源和公用事业分析行业产业运行态势及投资规划深度研究报告
- 2025至2030中国胃癌药物行业发展趋势分析与未来投资战略咨询研究报告
- 婚前协议(保护女方利益)
- 奉贤区教育系统师德师风建设学习测试附有答案
- 西方经济学(第二版)完整整套课件(马工程)
- 扶贫农产品购销合同协议(农产品购销合同模板)
- 汽车维修高级工考试试题及参考答案
- 检验科安全管理制度汇总
- GB/T 5782-2016六角头螺栓
- GB/T 23445-2009聚合物水泥防水涂料
- GB/T 13451.2-1992着色颜料相对着色力和白色颜料相对散射力的测定光度计法
- GB/T 11264-2012热轧轻轨
- 山东省中小学校档案管理暂行办法
评论
0/150
提交评论