Nur Zincir-Heywood, PhD, is Full Professor of Computer Science with Dalhousie University in Nova Scotia, Canada. She is an Associate Editor of the IEEE Transactions on Network and Service Management and Wiley International Journal of Network Management. Marco Mellia, PhD, is Full Professor with Politecnico di Torino, Italy. He is an Associate Editor of the IEEE Transactions on Network and Service Management, Elsevier Computer Networks and ACM Computer Communication Reviews. Yixin Diao, PhD, is Director of Data Science and Analytics at PebblePost in New York, NY, USA. He is an Associate Editor of the IEEE Transactions on Network and Service Management and the Journal of Network and Systems Management.

• List of Contributors xvPreface xxiAcknowledgments xxvAcronyms xxviiPart I Introduction 11 Overview of Network and Service Management 3Marco Mellia, Nur Zincir-Heywood, and Yixin Diao1.1 Network and Service Management at Large 31.2 Data Collection and Monitoring Protocols 51.2.1 SNMP Protocol Family 51.2.2 Syslog Protocol 51.2.3 IP Flow Information eXport (IPFIX) 61.2.4 IP Performance Metrics (IPPM) 71.2.5 Routing Protocols and Monitoring Platforms 81.3 Network Configuration Protocol 91.3.1 Standard Configuration Protocols and Approaches 91.3.2 Proprietary Configuration Protocols 101.3.3 Integrated Platforms for Network Monitoring 101.4 Novel Solutions and Scenarios 121.4.1 Software-Defined Networking – SDN 121.4.2 Network Functions Virtualization –NFV 14Bibliography 152 Overview of Artificial Intelligence and Machine Learning 19Nur Zincir-Heywood, Marco Mellia, and Yixin Diao2.1 Overview 192.2 Learning Algorithms 202.2.1 Supervised Learning 212.2.2 Unsupervised Learning 222.2.3 Reinforcement Learning 232.3 Learning for Network and Service Management 24Bibliography 26Part II Management Models and Frameworks 333 Managing Virtualized Networks and Services with Machine Learning 35Raouf Boutaba, Nashid Shahriar, Mohammad A. Salahuddin, and Noura Limam3.1 Introduction 353.2 Technology Overview 373.2.1 Virtualization of Network Functions 383.2.1.1 Resource Partitioning 383.2.1.2 Virtualized Network Functions 403.2.2 Link Virtualization 413.2.2.1 Physical Layer Partitioning 413.2.2.2 Virtualization at Higher Layers 423.2.3 Network Virtualization 423.2.4 Network Slicing 433.2.5 Management and Orchestration 443.3 State-of-the-Art 463.3.1 Network Virtualization 463.3.2 Network Functions Virtualization 493.3.2.1 Placement 493.3.2.2 Scaling 523.3.3 Network Slicing 553.3.3.1 Admission Control 553.3.3.2 Resource Allocation 563.4 Conclusion and Future Direction 593.4.1 Intelligent Monitoring 603.4.2 Seamless Operation and Maintenance 603.4.3 Dynamic Slice Orchestration 613.4.4 Automated Failure Management 613.4.5 Adaptation and Consolidation of Resources 613.4.6 Sensitivity to Heterogeneous Hardware 623.4.7 Securing Machine Learning 62Bibliography 634 Self-Managed 5G Networks 69Jorge Martín-Pérez, Lina Magoula, Kiril Antevski, Carlos Guimarães, Jorge Baranda, Carla Fabiana Chiasserini, Andrea Sgambelluri, Chrysa Papagianni, Andrés García-Saavedra, Ricardo Martínez, Francesco Paolucci, Sokratis Barmpounakis, Luca Valcarenghi, Claudio EttoreCasetti, Xi Li, Carlos J. Bernardos, Danny De Vleeschauwer, Koen De Schepper, Panagiotis Kontopoulos, Nikolaos Koursioumpas, Corrado Puligheddu, Josep Mangues-Bafalluy, and Engin Zeydan4.1 Introduction 694.2 Technology Overview 734.2.1 RAN Virtualization and Management 734.2.2 Network Function Virtualization 754.2.3 Data Plane Programmability 764.2.4 Programmable Optical Switches 774.2.5 Network Data Management 784.3 5G Management State-of-the-Art 804.3.1 RAN resource management 804.3.1.1 Context-Based Clustering and Profiling for User and Network Devices 804.3.1.2 Q-Learning Based RAN Resource Allocation 814.3.1.3 vrAIn: AI-Assisted Resource Orchestration for Virtualized Radio Access Networks 814.3.2 Service Orchestration 834.3.3 Data Plane Slicing and Programmable Traffic Management 854.3.4 Wavelength Allocation 864.3.5 Federation 884.4 Conclusions and Future Directions 89Bibliography 925 AI in 5G Networks: Challenges and Use Cases 101Stanislav Lange, Susanna Schwarzmann, Marija Gaji´c, Thomas Zinner, and Frank A. Kraemer5.1 Introduction 1015.2 Background 1035.2.1 ML in the Networking Context 1035.2.2 ML in Virtualized Networks 1045.2.3 ML for QoE Assessment and Management 1045.3 Case Studies 1055.3.1 QoE Estimation and Management 1065.3.1.1 Main Challenges 1075.3.1.2 Methodology 1085.3.1.3 Results and Guidelines 1095.3.2 Proactive VNF Deployment 1105.3.2.1 Problem Statement and Main Challenges 1115.3.2.2 Methodology 1125.3.2.3 Evaluation Results and Guidelines 1135.3.3 Multi-service, Multi-domain Interconnect 1155.4 Conclusions and Future Directions 117Bibliography 1186 Machine Learning for Resource Allocation in Mobile Broadband Networks 123Sadeq B. Melhem, Arjun Kaushik, Hina Tabassum, and Uyen T. Nguyen6.1 Introduction 1236.2 ML in Wireless Networks 1246.2.1 Supervised ML 1246.2.1.1 Classification Techniques 1256.2.1.2 Regression Techniques 1256.2.2 Unsupervised ML 1266.2.2.1 Clustering Techniques 1266.2.2.2 Soft Clustering Techniques 1276.2.3 Reinforcement Learning 1276.2.4 Deep Learning 1286.2.5 Summary 1296.3 ML-Enabled Resource Allocation 1296.3.1 Power Control 1316.3.1.1 Overview 1316.3.1.2 State-of-the-Art 1316.3.1.3 Lessons Learnt 1326.3.2 Scheduling 1326.3.2.1 Overview 1326.3.2.2 State-of-the-Art 1326.3.2.3 Lessons Learnt 1346.3.3 User Association 1346.3.3.1 Overview 1346.3.3.2 State-of-the-Art 1366.3.3.3 Lessons Learnt 1366.3.4 Spectrum Allocation 1366.3.4.1 Overview 1366.3.4.2 State-of-the-Art 1386.3.4.3 Lessons Learnt 1386.4 Conclusion and Future Directions 1406.4.1 Transfer Learning 1406.4.2 Imitation Learning 1406.4.3 Federated-Edge Learning 1416.4.4 Quantum Machine Learning 142Bibliography 1427 Reinforcement Learning for Service Function Chain Allocation in Fog Computing 147José Santos, Tim Wauters, Bruno Volckaert, and Filip De Turck7.1 Introduction 1477.2 Technology Overview 1487.2.1 Fog Computing (FC) 1497.2.2 Resource Provisioning 1497.2.3 Service Function Chaining (SFC) 1507.2.4 Micro-service Architecture 1507.2.5 Reinforcement Learning (RL) 1517.3 State-of-the-Art 1527.3.1 Resource Allocation for Fog Computing 1527.3.2 ML Techniques for Resource Allocation 1537.3.3 RL Methods for Resource Allocation 1547.4 A RL Approach for SFC Allocation in Fog Computing 1557.4.1 Problem Formulation 1557.4.2 Observation Space 1567.4.3 Action Space 1577.4.4 Reward Function 1587.4.5 Agent 1617.5 Evaluation Setup 1627.5.1 Fog–Cloud Infrastructure 1627.5.2 Environment Implementation 1627.5.3 Environment Configuration 1647.6 Results 1657.6.1 Static Scenario 1657.6.2 Dynamic Scenario 1677.7 Conclusion and Future Direction 169Bibliography 170Part III Management Functions and Applications 1758 Designing Algorithms for Data-Driven Network Management and Control: State-of-the-Art and Challenges 177Andreas Blenk, Patrick Kalmbach, Johannes Zerwas, and Stefan Schmid8.1 Introduction 1778.1.1 Contributions 1798.1.2 Exemplary Network Use Case Study 1798.2 Technology Overview 1818.2.1 Data-Driven Network Optimization 1818.2.2 Optimization Problems over Graphs 1828.2.3 From Graphs to ML/AI Input 1848.2.4 End-to-End Learning 1878.3 Data-Driven Algorithm Design: State-of-the Art 1888.3.1 Data-Driven Optimization in General 1888.3.2 Data-Driven Network Optimization 1908.3.3 Non-graph Related Problems 1928.4 Future Direction 1938.4.1 Data Production and Collection 1938.4.2 ML and AI Advanced Algorithms for Network Management with Performance Guarantees 1948.5 Summary 194Acknowledgments 195Bibliography 1959 AI-Driven Performance Management in Data-Intensive Applications 199Ahmad Alnafessah, Gabriele Russo Russo, Valeria Cardellini, Giuliano Casale, and Francesco Lo Presti9.1 Introduction 1999.2 Data-Processing Frameworks 2009.2.1 Apache Storm 2009.2.2 Hadoop MapReduce 2019.2.3 Apache Spark 2029.2.4 Apache Flink 2029.3 State-of-the-Art 2039.3.1 Optimal Configuration 2039.3.1.1 Traditional Approaches 2039.3.1.2 AI Approaches 2049.3.1.3 Example: AI-Based Optimal Configuration 2069.3.2 Performance Anomaly Detection 2079.3.2.1 Traditional Approaches 2089.3.2.2 AI Approaches 2089.3.2.3 Example: ANNs-Based Anomaly Detection 2109.3.3 Load Prediction 2119.3.3.1 Traditional Approaches 2129.3.3.2 AI Approaches 2129.3.4 Scaling Techniques 2139.3.4.1 Traditional Approaches 2139.3.4.2 AI Approaches 2149.3.5 Example: RL-Based Auto-scaling Policies 2149.4 Conclusion and Future Direction 216Bibliography 21710 Datacenter Traffic Optimization with Deep Reinforcement Learning 223Li Chen, Justinas Lingys, Kai Chen, and Xudong Liao10.1 Introduction 22310.2 Technology Overview 22510.2.1 Deep Reinforcement Learning (DRL) 22610.2.2 Applying ML to Networks 22710.2.3 Traffic Optimization Approaches in Datacenter 22910.2.4 Example: DRL for Flow Scheduling 23010.2.4.1 Flow Scheduling Problem 23010.2.4.2 DRL Formulation 23010.2.4.3 DRL Algorithm 23110.3 State-of-the-Art: AuTO Design 23110.3.1 Problem Identified 23110.3.2 Overview 23210.3.3 Peripheral System 23310.3.3.1 Enforcement Module 23310.3.3.2 Monitoring Module 23410.3.4 Central System 23410.3.5 DRL Formulations and Solutions 23510.3.5.1 Optimizing MLFQ Thresholds 23510.3.5.2 Optimizing Long Flows 23910.4 Implementation 23910.4.1 Peripheral System 23910.4.1.1 Monitoring Module (MM): 24010.4.1.2 Enforcement Module (EM): 24010.4.2 Central System 24110.4.2.1 sRLA 24110.4.2.2 lRLA 24210.5 Experimental Results 24210.5.1 Setting 24310.5.2 Comparison Targets 24410.5.3 Experiments 24410.5.3.1 Homogeneous Traffic 24410.5.3.2 Spatially Heterogeneous Traffic 24510.5.3.3 Temporally and Spatially Heterogeneous Traffic 24610.5.4 Deep Dive 24710.5.4.1 Optimizing MLFQ Thresholds using DRL 24710.5.4.2 Optimizing Long Flows using DRL 24810.5.4.3 System Overhead 24910.6 Conclusion and Future Directions 251Bibliography 25311 The New Abnormal: Network Anomalies in the AI Era 261Francesca Soro, Thomas Favale, Danilo Giordano, Luca Vassio, Zied Ben Houidi, and Idilio Drago11.1 Introduction 26111.2 Definitions and Classic Approaches 26211.2.1 Definitions 26311.2.2 Anomaly Detection: A Taxonomy 26311.2.3 Problem Characteristics 26411.2.4 Classic Approaches 26611.3 AI and Anomaly Detection 26711.3.1 Methodology 26711.3.2 Deep Neural Networks 26811.3.3 Representation Learning 27011.3.4 Autoencoders 27111.3.5 Generative Adversarial Networks 27211.3.6 Reinforcement Learning 27411.3.7 Summary and Takeaways 27511.4 Technology Overview 27711.4.1 Production-Ready Tools 27711.4.2 Research Alternatives 27911.4.3 Summary and Takeaways 28011.5 Conclusions and Future Directions 282Bibliography 28312 Automated Orchestration of Security Chains Driven by Process Learning 289Nicolas Schnepf, Rémi Badonnel, Abdelkader Lahmadi, and Stephan Merz12.1 Introduction 28912.2 RelatedWork 29012.2.1 Chains of Security Functions 29112.2.2 Formal Verification of Networking Policies 29212.3 Background 29412.3.1 Flow-Based Detection of Attacks 29412.3.2 Programming SDN Controllers 29512.4 Orchestration of Security Chains 29612.5 Learning Network Interactions 29812.6 Synthesizing Security Chains 30112.7 Verifying Correctness of Chains 30612.7.1 Packet Routing 30612.7.2 Shadowing Freedom and Consistency 30612.8 Optimizing Security Chains 30812.9 Performance Evaluation 31112.9.1 Complexity of Security Chains 31212.9.2 Response Times 31312.9.3 Accuracy of Security Chains 31312.9.4 Overhead Incurred by Deploying Security Chains 31412.10 Conclusions 315Bibliography 31613 Architectures for Blockchain-IoT Integration 321Sina Rafati Niya, Eryk Schiller, and Burkhard Stiller13.1 Introduction 32113.1.1 Blockchain Basics 32313.1.2 Internet-of-Things (IoT) Basics 32413.2 Blockchain-IoT Integration (BIoT) 32513.2.1 BIoT Potentials 32613.2.2 BIoT Use Cases 32813.2.3 BIoT Challenges 32913.2.3.1 Scalability 33213.2.3.2 Security 33313.2.3.3 Energy Efficiency 33413.2.3.4 Manageability 33513.3 BIoT Architectures 33513.3.1 Cloud, Fog, and Edge-Based Architectures 33713.3.2 Software-Defined Architectures 33713.3.3 A Potential Standard BIoT Architecture 33813.4 Summary and Considerations 341Bibliography 342Index 345