Foundation for Digital Twins

ROBERTO MINERVA is an Associate Professor with the Service Architecture Laboratory, Institut Mines Telecom—Telecom Sud Paris, Institute Polytechnique de Paris, France. From 2016 to 2018, he was the Technical Project Leader of SoftFIRE, a European Project devoted to the experimentation of NFV, SDN, and edge computing. He was Chairperson of the IEEE IoT Initiative from 2014 to 2016. NOEL CRESPI is a Professor and MSc Programme Director, leading the Data Intelligence and Communication Engineering laboratory (DICE) at the Institut Mines Telecom—Telecom Sud Paris, Institute Polytechnique de Paris, France, where he has been since 2002. He coordinates the standardisation activities for Institut Mines-Telecom at ETSI, 3GPP and ITU-T.

• ContentsContributors Foreword Preface Acknowledgments Acronyms Introduction 1 What is a Digital Twin ?1.1 Introductory Concepts1.1.1 Basic Definitions1.1.2 Formalized Definitions1.2 Digital Twin: a Rationale1.2.1 Models and Modeling1.2.2 DT as a Combination of Models1.3 Digital Twin Definition: a step further1.3.1 Digital Twin Properties1.4 Digital Twin Representation, Model, and more1.5 The Software Part of a Digital Twin1.6 Specification Methodologies1.7 Domain Knowledge and Operation in Digital Twins1.8 Clarifying Scope and Terminology for Digital Twin Architectures1.8.1 Digital Twin Architecture and Scope Definition1.8.2 Evolving Terminology in Digital Twin Research2 Models and Modeling Aspects of a Digital Twin2.1 The Representational Aspects of a Digital Twin2.2 Data Modeling2.2.1 Data and Behavior Modeling2.2.2 Data Models as Enablers of Passive Digital Twins2.3 Behavior Modeling2.4 Predictive Models2.5 Prognosis Model2.6 Prescriptive Model2.7 A flexible Approach to multi-facet Models2.7.1 Relationships between different Models2.7.2 Models and Design, the Complexity of DT Modeling2.7.3 Modeling the Environment2.7.4 Context and Situation2.7.5 From Top or Bottom ?3 Foundational Data for a Digital Twin3.1 Descriptive Models3.2 Data-driven Architecture3.3 Which Data to include in a Digital Twin3.4 Organizing Data into Data Models3.4.1 Minimal DT Structure and multiple types of Data3.4.2 Data Representations for the Digital Twin3.4.3 Data Models and Application Programming Interfaces3.5 Data Models for a Digital Twin3.5.1 Data Modeling Techniques and Semantic Enrichmentin Digital Twins3.5.2 Existing Data Models for Digital Twins3.6 Extending Data Models to Fit the Stakeholders’ Needs3.7 Relationships between Descriptive and Behavior Models3.7.1 Data Exploitation Chain: from Passive to PrescriptiveDT4 Digital Twin as a Behavior Model of a Physical System4.1 Behavior Modeling4.1.1 An Example: the Traffic Light System Behavior4.1.2 Behavior Modeling Challenges and Needs4.2 Foundational Behavior Models4.3 Developing a Behavior Model4.3.1 Implementing the Twin as a State ManagementComponent4.3.2 Environment and DT Structure4.4 AI and Digital Twins4.4.1 Iterative Refinement of Digital Twin Behavior Modelsusing Generative AI  4.4.2 Extraction of Behavior Rules from Data4.4.3 Enhancing AI Explainability in Digital Twins throughBehavior Models4.4.4 Autonomic Digital Twins and Agentic AI4.5 Simulation Models, Behavior Validation, and GenAI Integrationin Digital Twins4.5.1 Applicable Simulation Models4.5.2 MetaModel for Unified Execution and Simulation4.5.3 Evolving the Traffic Light Digital Twin with Situation-Aware Behavior Modeling4.6 Behavior Models for Prognosis and Prescriptive Digital Twins4.6.1 Behavior Models for Prognostic Digital Twins4.6.2 Prescriptive Model and Integration with PrognosisCapabilities4.6.3 Recommendation Systems in Digital Twins4.6.4 The Role of Autonomics in Prescriptive Digital Twins4.6.5 Benefits of Behavior Models in Prescriptive DigitalTwins4.7 Behavior Modeling Recap5 Architecting and Implementing Digital Twin Systems: Approaches,Guidelines and Best Practices5.1 Requirements, Structural Concepts and Terminology for a DigitalTwin Architecture5.1.1 Requirements and Properties of Digital Twins  5.1.2 State of the Art in Digital Twin Architectures5.1.3 Reimagining a Versatile Digital Twin Architecture5.2 Software Interaction Paradigms for Digital Twin Implementation5.2.1 Advantages for DT Implementation5.2.2 Home Automation Example5.3 Componentization of DT Architecture5.3.1 System Engine5.3.2 Data Management5.3.3 DT Engine5.3.4 DT Life-cycle Management5.4 Model First5.4.1 Top-Down Digital Twins5.4.2 Bottom-Up Digital Twins5.4.3 Hybrid Approaches5.4.4 Interacting DTs5.5 From Modules to Components and Microservices: A DigitalTwin Perspective5.5.1 Rationale for further Decomposition: Digital TwinSpecifics5.5.2 Microservice Design Representation5.5.3 Rationale for Microservice Decomposition: DigitalTwin Advantages5.5.4 Preparing the Components for Deployment5.5.5 Comparison of the Architecture with ISO 23247 Standard5.6 Testing and Validation of Digital Twins5.6.1 Importance of Testing and Validation5.6.2 Continuous Impact within the Enterprise5.6.3 Techniques and Approaches for Testing and Validation5.6.4 Value of a Validated Digital Twin6 Deploying and Operating a Digital Twin6.1 Distribution of DT Components and Functions6.1.1 Centralized or Distributed DT6.1.2 A Deployment Scenario6.2 Operating the Digital Twin6.2.1 Product and Digital Twin Life-cycle Managementand Phase Transitions6.2.2 Management of Digital Twin Functionalities6.2.3 Artificial Intelligence for DT Management and Operation6.2.4 Data Management for the Digital Twin6.2.5 Management of the System Infrastructure6.2.6 Additional relevant Topics6.3 Example: Traffic Light Service Digital Twin6.4 Digital Twin Impact on Organization Processes6.5 Life-cycle Insights from Industrial Experiences6.5.1 Life-cycle Phases6.5.2 Exemplary Industrial Digital Twin Projects6.5.3 Enablers and Barriers to Industrial Exploitation6.5.4 Advantages and Enterprise Effort7 Some Examples of Applicability of the Digital Twin Architecture7.1 Introduction7.2 Developing Digital Twins for Smart Cities: A Bottom-Up Approach7.2.1 From Simple Digital Twins to Specialized Behavior2.2 Microservices and Component Flexibility7.2.3 Integrating Heterogeneous Data Streams for HolisticUrban Insights7.3 Network Digital Twin: the Edge-Cloud Continuum Representation7.3.1 A Top-Down Approach for NDT Design7.3.2 NDT Template: Monitoring and Optimization7.3.3 Stakeholder Views and Insights7.3.4 NDT Architecture Overview7.3.5 Optimization and Prognosis: Example Workflows7.4 The Challenge of DTs for Cultural Heritage7.4.1 A Hybrid Approach: Bottom-Up and Top-Down7.4.2 Multi-View Digital Twins for Artifacts7.4.3 Web of related Digital Twins7.4.4 Personalized and Adaptive Experiences7.4.5 Case Study: Egyptian Scarabs and the Power of DigitalTwins7.4.6 Context DT Architecture Support for Cultural Heritage7.5 Digital Twin as an Integral Part of the Metaverse7.5.1 State of the Art of Metaverse Platforms and Mappingto the Context Digital Architecture7.5.2 Integration Points and Mutual Enhancement7.5.3 Use Cases: Education, Tourism, and Factory Management7.6 Implementing Services with the Envisaged Architecture8 The Digital Twin of the Future8.1 The Evolution of DT8.2 Evaluating Digital Twins as a General Solution8.3 Promising Application Domains8.4 Anticipated Evolution of Digital Twin Technologies8.4.1 DT Platforms8.4.2 DT Creation and Development8.4.3 Decomposition of Modules8.4.4 Extensible Architecture8.4.5 Integrated Methodologies8.4.6 AI Integration8.4.7 Testing and Assessment8.5 Improved Operations8.5.1 Life-cycle Management8.5.2 Managing the Switch of States8.5.3 Operations Tools8.6 Interoperability and Standards8.6.1 Standardization Efforts8.6.2 Open APIs and Data Models8.7 Ethical, Privacy, and Trust Considerations in Digital Twin Systems8.8 Human-in-the-Loop and User Experience8.8.1 User-Centric Design8.8.2 Visualization and Immersive Interfaces8.9 Sustainability and Societal Impact8.9.1 DTs for Sustainable Development8.9.2 Societal Impact and Digital Inclusion8.10 Future StepsA Listings and DetailsA.1 Descriptive ModelingA.1.1 Data Model and Application Programming InterfacesA.1.2 Extending the Data Model to Fit Stakeholder NeedsA.2 Behavior ModelingA.2.1 NGSI-LD Data Model for Traffic Light Behavior