Front Cover
Shaping the Future of IoT with Edge Intelligence
Contents
Preface
List of Figures
List of Tables
List of Contributors
List of Abbreviations
I Edge Intelligence with 5G/6G Networks
	1 Edge Networking Technology Drivers for Next-generation Internet of Things in the TERMINET Project
		1.1 Introduction
		1.2 Technology Drivers
			1.2.1 Software defined networking and network function virtualization
			1.2.2 Beyond 5G mobile networks
			1.2.3 Digital twin
			1.2.4 Multiple-access edge computing
		1.3 Conclusion
	2 AI-driven Service and Slice Orchestration
		2.1 Introduction
		2.2 Related Work
			2.2.1 Management and orchestration of 5G networks
			2.2.2 5G network slices
			2.2.3 Information models for 5G network slices
			2.2.4 Management of 5G network slices
		2.3 Architectural Principles
		2.4 Functional Architecture
			2.4.1 Orchestration components
			2.4.2 AI/ML and monitoring platform
		2.5 Example of AI/ML-based Network Slice Optimization
	3 Tactile IoT Architecture for the IoT–Edge–Cloud Continuum: The ASSIST-IoT Approach
		3.1 Introduction
		3.2 Concepts and Approach
			3.2.1 Design principles
			3.2.2 Conceptual approach
		3.3 Architecture Views
			3.3.1 Functional view
				3.3.1.1 Device and edge plane
				3.3.1.2 Smart network and control plane
				3.3.1.3 Data management plane
				3.3.1.4 Applications and services plane
			3.3.2 Node view
			3.3.3 Development view
			3.3.4 Deployment view
				3.3.4.1 Infrastructure and Kubernetes considerations
				3.3.4.2 Enablers deployment
			3.3.5 Data view
		3.4 Verticals
			3.4.1 Self-*
			3.4.2 Interoperability
			3.4.3 Security, privacy, and trust
				3.4.3.1 Security
				3.4.3.2 Privacy
				3.4.3.3 Trust
			3.4.4 Scalability
			3.4.5 Manageability
		3.5 Conclusion
II Artificial Intelligence of Things (AIoT) and AI at the Edge
	4 Machine Learning (ML) as a Service (MLaaS): Enhancing IoT with Intelligence, Adaptive Online Deep and Reinforcement Learning, Model Sharing, and Zero-knowledge Model Verification
		4.1 Introduction
		4.2 MLaaS
			4.2.1 MLaaS features
			4.2.2 MLaaS architecture, services, and delivery
		4.3 Adaptive Online Deep Learning
			4.3.1 Introduction
			4.3.2 Features
			4.3.3 Technical solution
			4.3.4 Evaluation
		4.4 Model Sharing, Model Translation, and Zero-knowledge Model Verification
			4.4.1 Introduction
			4.4.2 Features
			4.4.3 Technical implementation
			4.4.4 Evaluation
		4.5 Conclusion
	5 Federated Learning Models in Decentralized Critical Infrastructure
		5.1 Introduction
			5.1.1 Definition and motivation
			5.1.2 Federated learning domains
			5.1.3 Use cases and applications
		5.2 How Federated Learning Works
			5.2.1 Overview of the architecture and process
			5.2.2 Key components
				5.2.2.1 Orchestrator
				5.2.2.2 Aggregator
				5.2.2.3 Worker
			5.2.3 Types of federated learning
			5.2.4 Model fusion algorithms
		5.3 Federated Learning vs. Traditional Centralized Learning
			5.3.1 Advantages and disadvantages of federated learning
			5.3.2 Real-world examples of federated learning
				5.3.2.1 Smart farming
				5.3.2.2 Smart, sustainable, and efficient buildings
				5.3.2.3 Industrial supply chains
				5.3.2.4 Industrial infrastructures
				5.3.2.5 Medical sector
		5.4 Implementing Federated Learning
			5.4.1 Tools and frameworks available
			5.4.2 Challenges
		5.5 Conclusion
	6 Analysis of Privacy Preservation Enhancements in Federated Learning Frameworks
		6.1 Introduction
		6.2 Privacy-preserving Federated Learning
			6.2.1 Federated learning frameworks
			6.2.2 Privacy preservation in federated learning
			6.2.3 State-of-the-art approaches in privacy-preserving federated learning
			6.2.4 Comparison of federated learning frameworks considering privacy preservation
		6.3 Conclusion
	7 Intelligent Management at the Edge
		7.1 Introduction to Intelligence at 5G/6G Networks Edge
			7.1.1 Edge automation
				7.1.1.1 State of the art
				7.1.1.2 Key enablers
			7.1.2 Edge intelligence
				7.1.2.1 State of the art
				7.1.2.2 Key enablers
			7.1.3 Edge computing and 5G/6G: a cloud native architecture
		7.2 Distributed Telemetry
			7.2.1 Hierarchical and distributed monitoring framework
				7.2.1.1 Monitoring agents
				7.2.1.2 Aggregators – monitoring servers
				7.2.1.3 Centralized aggregator – monitoring server
		7.3 AI Pipelines for the Edge-to-cloud Continuum
			7.3.1 Native AI for distributed edge-to-cloud environments
				7.3.1.1 Energy saving in distributed edge computing
				7.3.1.2 Latency-aware AI processes in edge computing
	8 IoT Things to Service Matchmaking at the Edge
		8.1 Introduction
		8.2 Semantic Matchmaking and Current Approaches
		8.3 TSMatch, an Example of Semantic Matchmaking for IIoT
			8.3.1 Setup
			8.3.2 Runtime
		8.4 Semantic Matchmaking Challenges for IoT
		8.5 Evolving Semantic Matchmaking at the Edge
			8.5.1 Hybrid semantic matchmaking
			8.5.2 Categorization
			8.5.3 Tradeoff
			8.5.4 Feedback Loop
		8.6 Conclusion
	9 A Scalable, Heterogeneous Hardware Platform for Accelerated AIoT based on Microservers
		9.1 Introduction
		9.2 Heterogeneous Hardware Platform for the Cloud-edge-IoT Continuum
			9.2.1 Cloud computing platform RECS|Box
			9.2.2 Near-edge computing platform t.RECS
			9.2.3 Far-edge computing platform u.RECS
		9.3 Accelerator Overview
			9.3.1 Reconfigurable accelerators
		9.4 Benchmarking and Evaluation
			9.4.1 Methodology
			9.4.2 Evaluation results
		9.5 Conclusion
	10 Methods for Requirements Engineering, Verification, Security, Safety, and Robustness in AIoT Systems
		10.1 Introduction
		10.2 Architecture Framework for AIoT Systems
			10.2.1 State-of-the-art for AI systems architecture
			10.2.2 A compositional architecture framework for AIoT
			10.2.3 Clusters of concern
			10.2.4 Levels of abstraction
			10.2.5 Compositional architecture framework
			10.2.6 Applying a compositional architecture framework in practice
		10.3 WebAssembly as a Common Layer for the Cloud-edge Continuum
			10.3.1 Building blocks of a seamless continuum for AIoT
			10.3.2 WebAssembly as a unifying solution
			10.3.3 The case for a TEE-backed WebAssembly continuum
			10.3.4 WebAssembly performance
			10.3.5 WebAssembly limitations
			10.3.6 Closing remarks concerning the common layer
		10.4 TOCTOU-secure Remote Attestation and Certification for IoT
			10.4.1 AutoCert – proposed mechanism
				10.4.1.1 Pre-deployment
				10.4.1.2 Remote attestation
				10.4.1.3 TOCTOU and integrity_proof
				10.4.1.4 Verification for TOCTOU security
			10.4.2 Implementation and experimental evaluation
			10.4.3 AutoCert – conclusion
		10.5 Conclusion
III Blockchain Solutions for Trusted Edge Intelligence in IoT Systems
	11 Decentralized Strategy for Artificial Intelligence in Distributed IoT Ecosystems: Federation in ASSIST-IoT
		11.1 Introduction
			11.1.1 Decentralized AI
			11.1.2 Federated learning
		11.2 Federated Learning Principles
		11.3 Federated Learning System of ASSIST-IoT Project
			11.3.1 FL enablers
				11.3.1.1 FL Orchestrator
				11.3.1.2 FL Repository
				11.3.1.3 FL Training Collector
				11.3.1.4 FL Local Operations
			11.3.2 Secure reputation mechanism for the FL system via blockchain and distributed ledger
		11.4 ASSIST-IoT FL Application in an Automotive Defect Detection Use Case
			11.4.1 Business overview and context of the scenario
			11.4.2 Proposed solution and benefits of decentralized learning strategy
			11.4.3 Proposed validation
		11.5 Conclusions
	12 Achieving Security and Privacy in NG-IoT using Blockchain Techniques
		12.1 Introduction – What Is Blockchain?
		12.2 Permission-less and Permissioned Blockchain
		12.3 Consensus Mechanisms
		12.4 Smart Contracts
		12.5 Blockchain Applications for Security and Privacy
		12.6 Conclusion
IV Novel IoT Applications at the Cloud, Edge, and ``Far-edge''
	13 Enabling Remote-controlled Factory Robots via Smart IoT Application Programming Interface
		13.1 Introduction
		13.2 IoT Application for Supply chain
			13.2.1 IoT applications in smart factories and warehouses
		13.3 Tactile IoT Applications
			13.3.1 Tactile Internet applications encountered in supply chain stages
				13.3.1.1 Teleoperation
				13.3.1.2 Autonomous driving
				13.3.1.3 Industrial automation
		13.4 Industrial and Tactile Application Programming Interface (API)
			13.4.1 Proof-of-concept within the iNGENIOUS project
		13.5 Conclusion
	14 A Practical Deployment of Tactile IoT: 3D Models and Mixed Reality to Increase Safety at Construction Sites
		14.1 Introduction
		14.2 Solution Architecture
			14.2.1 Tactile internet aspect
			14.2.2 Data integration
			14.2.3 Mixed reality interface
		14.3 Evaluation
		14.4 Conclusions and Future Work
	15 Haptic and Mixed Reality Enabled Immersive Cockpits for Tele-operated Driving
		15.1 Introduction
		15.2 Tele-operated Driving challenges
			15.2.1 Real-time issues
			15.2.2 Immersive devices
		15.3 Immersive Cockpit Architecture and Components
			15.3.1 Overall architecture
			15.3.2 Components
				15.3.2.1 Head mounted displays
				15.3.2.2 Haptic gloves
				15.3.2.3 Wheels and pedals
				15.3.2.4 5G mmWave modems
		15.4 Proof of Concept
			15.4.1 End-to-end use case description
			15.4.2 Remote cockpit
			15.4.3 On-site cockpit
			15.4.4 KPIs collection
		15.5 Conclusion
	16 The EFPF Approach to Manufacturing Applications Across Edge-cloud Architectures
		16.1 Introduction
		16.2 Related Work
		16.3 The EFPF Architecture
			16.3.1 The EFPF ecosystem as a federation of digital manufacturing platforms
		16.4 EFPF SDK
		16.5 EFPF Selected Pilots
			16.5.1 Aerospace manufacturing pilot: environmental monitoring
			16.5.2 Furniture manufacturing pilot: factory connectivity
			16.5.3 Circular economy pilot - a waste to energy scenario
				16.5.3.1 Predictive maintenance services
				16.5.3.2 Fill level sensors – IoT-based monitoring system
				16.5.3.3 Online bidding process
				16.5.3.4 Blockchain Track and Trace App
				16.5.3.5 Tonnage and price forecasting services
		16.6 Summary
V IoT Skills and Business Models
	17 The EU-IoT Skills Framework for IoT Training and Career Development Processes
		17.1 Introduction
		17.2 Related Work
		17.3 The Eu-IoT Skills Framework
			17.3.1 Main principles
			17.3.2 Top-level categorization of IoT skills
			17.3.3 The four categories of IoT skills
				17.3.3.1 IoT technical and technological skills
				17.3.3.2 Business, marketing, management, and regulatory skills
				17.3.3.3 IoT end-user and operator 4.0 skills
				17.3.3.4 Social, management, and other soft skills
			17.3.4 Using the EU-IoT skills framework
				17.3.4.1 End-user groups
				17.3.4.2 Supporting training, hiring, and skills development processes
		17.4 The Eu-IoT Skills Survey
			17.4.1 Survey identity and methodological overview
			17.4.2 Analysis of results and main findings
				17.4.2.1 Popularity of broadly applicable skills
				17.4.2.2 Importance of specialized skills for sector-specific audiences
				17.4.2.3 The importance of soft skills
				17.4.2.4 Skills clustering into skills profiles
		17.5 From IoT Skills to Profiles and Learning Paths
			17.5.1 Skills profiles and learning path construction methodology
			17.5.2 Examples of IoT learning paths
		17.6 Conclusions
	18 Digital Business IoT Maturity Patterns from EU-IoT Ecosystem
		18.1 Introduction
		18.2 Statement of Purpose
		18.3 Methodology and Relevant Tools
			18.3.1 Data collection and analysis methodologies
			18.3.2 Interviews
			18.3.3 Digital maturity assessment
			18.3.4 Business model patterns survey
			18.3.5 Business model evaluation – innovation and configuration
		18.4 Results and Analysis
			18.4.1 Use case companies overview
			18.4.2 Digital maturity
			18.4.3 Business model patterns
			18.4.4 Business model innovation and configuration
			18.4.5 Technology trends
			18.4.6 Relevant skill areas and patterns
		18.5 Conclusion
	Index
	About the Editors
	Back Cover