Bringing Machine Learning to Software-Defined Networks

ISBN-13: 9789811948732 / Angielski / Miękka / 2022

Zehua Guo

Opis

Emerging machine learning techniques bring new opportunities to flexible network control and management. This book focuses on using state-of-the-art machine learning-based approaches to improve the performance of Software-Defined Networking (SDN). It will apply several innovative machine learning methods (e.g., Deep Reinforcement Learning, Multi-Agent Reinforcement Learning, and Graph Neural Network) to traffic engineering and controller load balancing in software-defined wide area networks, as well as flow scheduling, coflow scheduling, and flow migration for network function virtualization in software-defined data center networks. It helps readers reflect on several practical problems of deploying SDN and learn how to solve the problems by taking advantage of existing machine learning techniques. The book elaborates on the formulation of each problem, explains design details for each scheme, and provides solutions by running mathematical optimization processes, conducting simulated experiments, and analyzing the experimental results.

Spis treści

1        Machine Learning for Software-Defined Networking

1.1   Introduction of Software-Defined Networking

1.1.1         Software-Defined Wide Area Network

1.1.2         Software-Defined Data Center Networks

1.2   Introduction of Machine Learning Techniques

1.2.1         Deep Reinforcement Learning

1.2.2         Multi-Agent Reinforcement Learning

1.2.3         Graph Neural Network

 

2        Deep Reinforcement Learning-based Traffic Engineering in SD-WANs

2.1   Introduction of Traffic Engineering

2.2   Motivation

2.2.1         Problems of Existing Solutions

2.2.2         Opportunity

2.3   Overview of ScaleDRL

2.4   Design Details of ScaleDRL

2.4.1         Pinning Control in the Offline Phase

2.4.1.1     Pinning Control

2.4.1.2     Link Selection Algorithm

2.4.2         DRL Implementation of the Online Phase

2.4.2.1     DRL Framework

2.4.2.2     Customization of Neural Networks and Interfaces

2.5   Performance Evaluation

2.5.1         Simulation Setup

2.5.2         Comparison Scheme

2.5.3         Simulation Results

2.6   Conclusion

 

3        Multi-Agent Reinforcement Learning-based Controller Load Balancing in SD-WANs

3.1   Introduction of Controller Load Balancing

3.2   Motivation

3.2.1         Problems of Existing Solutions

3.2.2         Opportunity

3.3   Controller Load Balancing Problem Formulation

2.3.1 Control Plane Resource Utilization Modeling

2.3.2 Control Plane Load Balancing Problem Formulation

2.3.3 Problem Complexity Analysis

3.4   Overview of MARVEL

3.5   Design Details of MARVEL

3.5.1         Training Phase

3.5.2         Working Phase

3.5.3         MARVEL Model Implementation

3.6   Performance Evaluation

3.6.1         Simulation Setup

3.6.2         Comparison Scheme

3.6.3         Simulation Results

3.7   Conclusion

 

4        Deep Reinforcement Learning-based Flow Scheduling for Power Efficiency in Data Center Networks

4.1   Introduction of Data Center Networks

4.1.1         Traffic Classification

4.1.2         Traffic Dynamic Analysis

4.2   Motivation

4.2.1         Problems of Existing Solutions

4.2.2         Opportunity

4.3   Problem formulation

4.3.1         Design Considerations

4.3.2         Problem Formulation

4.4   Overview of SmartFCT

4.5   Design Details of SmartFCT

4.5.1         Flow Information Collection

4.5.2         DRL Algorithm Framework

4.5.3         DRL Implementation Details

4.6   Performance Evaluation

4.6.1         Simulation Setup

4.6.2         Comparison Scheme

4.6.3         Simulation Results

4.7   Conclusion

 

5        Graph Neural Network-based Coflow Scheduling in Data Center Networks

5.1   Introduction of Coflow

5.2   Motivation

5.2.1         Problems of Existing Solutions

5.2.2         Opportunity

5.3   Problem Formulation

5.4   Overview of DeepWeave

5.5   Design Details of DeepWeave

5.5.1         DRL Framework for Training

5.5.2         Neural Network Implementation

5.5.3         Policy Converter

5.6   Performance Evaluation

5.6.1         Simulation Setup

5.6.2         Comparison Scheme

5.6.3         Simulation Results

5.7   Conclusion

 

6        Graph Neural Network-based Flow Migration for Network Function Virtualization

6.1   Introduction of Network Function Virtualization

6.1.1         Network Function Virtualization

6.1.2         State Migration in NFV

6.2   Motivation

6.2.1         Problems of Existing Solutions

6.2.2         Opportunity

6.3   Flow Migration Problem Formulation

6.4   Overview of DeepMigration

6.5   Design details of DeepMigration

6.5.1         Training Framework

6.5.2         GNN-based Function Approximator

6.5.3         Training Process

6.6   Performance Evaluation

6.6.1         Simulation Setup

6.6.2         Comparison Scheme

6.6.3         Simulation Results

6.7   Conclusion

 

7  Conclusion and Future work

 

O autorze

Dr. Zehua Guo received B.S. degree from Northwestern Polytechnical University, Xi’an, China, M.S. degree from Xidian University, Xi’an, China, and Ph.D. degree from Northwestern Polytechnical University, Xi’an, China. He is an Associate Professor at Beijing Institute of Technology, Beijing, China. He was a Research Fellow at the Department of Electrical and Computer Engineering, New York University Tandon School of Engineering, New York, NY, USA, and a Postdoctoral Research Associate at the Department of Computer Science and Engineering, University of Minnesota Twin Cities, Minneapolis, MN, USA. His research interests include programmable networks (e.g., software-defined networking, network function virtualization), machine learning, and network security. He is an Associate Editor of the IEEE Systems Journal, and EURASIP Journal on Wireless Communications and Networking (Springer), an Editor of the KSII Transactions on Internet and Information Systems, and a Guest Editor of the Journal of Parallel and Distributed Computing. He was the Session Chair for the IJCAI 2021, IEEE ICC 2018,  and currently serves as the Technical Program Committee Member of Computer Communications, AAAI, IWQoS, ICC, ICCCN, and ICA3PP. He has published 58 papers in prestigious IEEE/ACM/Elsevier journals and conferences, including TON, JSAC, IJCAI, TNSM, Computer Networks, ICDCS, IWQoS, and applied/owned 14 patents. He is a Senior Member of IEEE, China Institute of Communications, and Chinese Institute of Electronics, and a Member of China Computer Federation, ACM, ACM SIGCOMM, and ACM SIGCOMM China.