A Method for Template-based Architecture Modeling and its Application to Digital Twins

A method to efficiently manage variability in DT architectures.
Presented at ECMFA 2024.

1. Daniel Lehner, Jerome Pfeiffer, Stefan Klikovits,
Andreas Wortmann, Manuel Wimmer
A Methodfor
Template-based Architecture Modeling
and its Applicationto Digital Twins

2. Digital Twins
AnomalyDetection
AI-basedPlanning
PredictiveMaintenance
Digital
Twin
Actual System

…
HowtoengineerDigital Twins?
2
Picture sources: https://www.iberdrola.com/innovation/smart-farming-precision-agriculture
geraltfrompixabay.com, MR-Panda frompixabay.com
engelat storyblok.com: produkte-grossmaschine-spritzguss.jpg

3. Digital Twin (DT) Platform
Digital Twin (DT) Platforms
ActualSystem
Middleware
Database
DT
Model
DT
Interface
System
DT
Marketplace
Value-AddingServices
Azure Digital Twin Service
EclipseHono
Bosch IoT-Hub
EclipseVorto
VortolangDTDL
AWS TwinMaker
AWS TwinMaker
TimeSeriesInsights
Azure IoT-Hub
AWS Greengrass
EclipseDitto
IBM Digital Twin Exchange
Pfeiffer, Lehner, Wortmann, Wimmer: Modeling CapabilitiesofDigital Twin Platforms: Old Winein newBottles?
ECMFA, 2022.
3
Picture sources: https://www.iberdrola.com/innovation/smart-farming-precision-agriculture
geraltfrompixabay.com, MR-Panda frompixabay.com
engelat storyblok.com: produkte-grossmaschine-spritzguss.jpg

4. Different Technological Spaces
Howcanwebridge
thisgap?
Howcanwemanage the
resultingvariability?
PlatformsforRealizing
Value-AddingServicesDT Platforms
4
Mindthe
Gap!!

5. ExamplesfromtheSofDCarProject
Self-Adaptation for
E-Drive Reconfiguration
ReactivePlanningfor
Air Conditioning
Different
Reference
Architectures
Different DT
Architectures
DTPlatform DeviationChecker Planner
Azure
Digital Twins
AWS TM MOMoT
SC-based
Planner
State-based
Checker
Time-based
Checker
DTPlatform Planner
Azure
Digital Twins
AWS TM
MOMoT
SC-based
Planner
5

6. Evolution Scenarios
6
MOMoT
Event-based
Checker
AzureDT
MOMoT
Event-based
Checker
EclipseVortoDTPlatform
Planner
DTPlatform1
Planner
Integrator
DTPlatform2
Reference Architecture Evolution DT Architecture Evolution
Howcanwemanage thevariabilityresultingfromthe
•variousrequired(Reference) Architecturesand
•evolutionofthese(Reference) Architectures?
Version 1 Version 2 Version 1 Version 2

7. Approach 2: Fixed MAPE Architecture
•Pluggingmodulesintopredefinedreferencearchitecture
forself-adaptive systems
•Problem: Onefixedreferencearchitecturenot enough
forcoveringthemanypossible DT applications
Approach 1: DT Modules
•PairwiseintegrationofDT modulestobuild
concreteDT architectures
•Problem: Pairwiseintegrationdoesnot scale
ExistingDT Architecture Engineering Approaches
7
Module 1 Module 2
Module 3 Module 1
PlannerDTPlatform
Module 2
Weneeda scalableand flexible approach!
•Pfeiffer, Lehner, Wortmann, Wimmer. Towards a product line architecture for digital twins.ICSA-C, 2023.
•Dalibor et al. Generating customized low-code development platforms for digital twins.COLA, 2022

8. Phase 2: Reference Architecture Definition
Phase 3: Product Line Configuration
Phase 1: Service Definition
4.1
Template-based Architecture Modeling Method
8
Flexibility
Scalability
Automation

9. Phase 1: Service
Definition
Template-based Architecture Modeling Method
Phase 1
9
Abstract modules
intotemplates
Integratemodules
withtemplates
Define
Modules [1]
«manual» «manual»«manual»
«Template»DTPlatform
DataAccess
Interface
«Module» AzureDT
Azure
DT Service
DT
Definition
Language
«Module» AWSTM
AWS TM
Service
AWS TM
Language
Structure
Language
[1] Dalibor et al. Generating customized low-code
development platforms for digital twins.COLA, 2022

10. Phase 2: Reference
Architecture Definition
Phase 1: Service
Definition
10
Definereference
architecture
«manual»
Template-based Architecture Modeling Method
Phase 2
«Template»DTPlatform
DataAccess
Interface
Structure
Language
«Template»Planner
Planner
Interface
Planner Config
Language
import DTPlatform;
import Planner;
refarch ReactivePlanning {
software {
DataAccessInterface dt;
PlannerInterface planner;
dt.currentState -> planner.initialState;
planner.nextAction -> dt.opToExecute;
}
language {
DTLang.TypeDef -> PlannerLang.ElemType;
DTLang.PropertyDef -> PlannerLang.ElemProperty;
DTLang.MethodDef -> PlannerLang.Capability;
DTLang.IMethodCall -> PlannerLang.Action;
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

11. Phase 2: Reference
Architecture Definition
Phase 1: Service
Definition
11
Definereference
architecture
«manual»
Generate
productline
«automated»
Template-based Architecture Modeling Method
Phase 2
<<Template>> DTPlatform
DataAccess
Interface
Structure
Language
<<Module>> AzureDT
Azure
Accessor
DT
Definition
Language
import DTPlatform;
import Planner;
refarch ReactivePlanning {
...
}
Phase 1Step 2.1
ReactivePlanning
PlannerDTPlatform
AzureDT
Timed
Statecharts
Basic
Statecharts
AWS IoT MOMoT

12. Phase 2: Reference
Architecture Definition
Phase 3: Product
Line Configuration
Phase 1: Service
Definition
12
«manual»
4.1
Select features
in productline
«manual»
Integrateselectedfeatures
intoDT architecture
«automated»
Template-based Architecture Modeling Method
Phase 3
ReactivePlanning
PlannerDT Platform
Azure
Timed
Statecharts
Basic
Statecharts
AWS IoT MOMoT
Digital Twin
Azure
Basic
Statecharts
DTDL
Model
uses
SC
Models
uses
Azure DT
SC Model
ExecutionEngine
uses
uses
DTPlatform Planner

13. Phase 2: Reference
Architecture Definition
Phase 3: Product
Line Configuration
Phase 1: Service
Definition
Template-based Architecture Modeling Method
13
Flexibility
Scalability
Automation
Abstract modules
intotemplates
Integratemodules
withtemplates
Define
Modules
«manual» «manual»«manual»
Definereference
architecture
«manual»
Generate
productline
«automated»
«manual»
4.1
Select features
in productline
«manual»
Integrateselectedfeatures
intoDT architecture
«automated»

14. Tool Support
14
Phase 2: Reference
Architecture Definition
Phase 3: Product
Line Configuration
Phase 1: Service
Definition
Tool support availableon Github: https://github.com/cdl-
mint/DT-Product-Line-Architecture

15. Research Goal: Weneeda flexible and scalableapproachforDT (reference) architectures
Research Question: Whataretheprosand consofourtemplate-basedmethodcompared
tostateoftheart?
ComparativeStudy
•Template-basedapproach
•Module-onlyapproach
•MAPE-onlyapproach
Scenario-basedEvaluation
15
Service 1 Service 2
Service 3
Service 1
PlannerDTPlatform
Service 2
•Pfeiffer, Lehner, Wortmann, Wimmer. Towards a product line architecture for digital twins.ICSA-C, 2023.
•Dalibor et al. Generating customized low-code development platforms for digital twins.COLA, 2022Phase 2: Reference
Architecture Definition
Phase 3: Product
Line Configuration
Phase 1: Service
Definition
Flexibility
Scalability
Automation
Abstract modules
intotemplates
Integratemodules
withtemplates
Define
Modules
«manual» «manual»«manual»
Definereference
architecture
«manual»
Generate
productline
«automated»
«manual»
4.1
Select features
in productline
«manual»
Integrateselectedfeatures
intoDT architecture
«automated»

16. Scenario: Create MAPE Architectures
16
LessonsLearned:
•MAPE-only: plugDT modulesintoexistingMAPE components
•Module-only: moduleshavetobemanuallyintegratedforeachofthe12 DT architectures
•Template-based: all DT architecturescanbegeneratedfromtheMAPE DT productline
Deviation Checker PlannerDTPlatform
MAPE
Planner
DTPlatform
12 concreteDT architectures
Deviation
Checker
Reference
Architecture
Product
Line
AzureDT
Timed
Statecharts
Basic
Statecharts
AWS IoT MOMoT
Event-based
Checker
Time-based
Checker

17. Scenario: Create Extended MAPE Architectures
17
Deviation
Checker
Planner
DT
Platform
ReactivePlanning
PlannerDTPlatform
12 DT architectures
MAPE
Planner
DT
Platform
Deviation
Checker
PlannerDTPlatform
6 DT architectures
LessonsLearned:
•MAPE-only: onlyapplicableto12 MAPE-basedDT architectures
•Module-only: manualintegrationofeachofthe30 DT architectures
•Template-based: effortforcreatingreferencearchitectures, othertasksareautomated
Planner
DTPlatform1
DT
Platform1
Integrator
12 DT architectures
ReactivePlanningExt
DT
Platform2
Planner
DTPlatform2
Integrator

18. Scenario: ReplaceAzureDTwithEclipseVorto
18
MOMoT
Event-based
Checker
AzureDT
MOMoT
Event-based
Checker
EclipseVorto
LessonsLearned:
•MAPE-only: asbefore, onlyapplicabletoMAPE-basedDT architectures
•Module-only: manuallyadapteachofthe15 DT architecturesusingAzureDTmodule
•Template-based: onechange, connect EclipseVortomoduletoDTPlatformtemplate
DT Architecture Evolution
Version 1 Version 2

19. Scenario: IntroduceDT Integrator forMAPE
19
DTPlatform
Planner
DTPlatform1
Planner
Integrator
DTPlatform2
MOMoT
AzureDT
MOMoT
AzureDT-based
Integrator
Eclipse
Vorto
AWS TM
LessonsLearned:
•MAPE-only: not possible
•Module-only: refinementforeachDT architecture
•Template-based: refinementonlyon referencearchitecturelevel
Reference Architecture Evolution DT Architecture Evolution
Version 1 Version 2 Version 1 Version 2

20. MAPE-onlyapproach
+OptimizedforsubsetofDT architectures(MAPE Reference Architecture)
−Not applicablebeyondMAPE
Module-onlyapproach
+Ad-hoc adaptationspossible
−Variabilityleadstohigh integrationeffort
Template-basedapproach
+Providesscalabilityforhigh variabilityofDT architectures
−UpfrontcosttocreateDT templates/referencearchitectures
Discussion
20

21. •Template-basedapproachforreferencearchitecturemodeling
•Tool support availablebasedon MontiCore/Java on github
•Evaluationfor4 referencearchitecturesand 57 DT architectures
Future Work
•Extendscenarios
•Quantitative evaluationwithpractitioners
•Extendreferencearchitectureswithbehavioral interfaces
Conclusionand Future Work
21

22. ThankYou!
Questions?
Comments?
Feedback?
22
Daniel Lehner
[email protected]
https://se.jku.at/daniel-lehner/
https://github.com/derlehner