Network and service coordination: conventional and machine learning approaches / Stefan Schneider ; Supervisor: Prof. Dr. Holger Karl. Paderborn, 2022
Inhalt
- Abstract
- Introduction
- Background
- Problem Statement
- Conventional Coordination Approaches
- Centralized Coordination
- Hierarchical Coordination
- Introduction
- Related Work
- Problem Statement
- Hierarchical Coordination Approach
- Domains and Hierarchies
- Bottom-Up Information Advertisement (Phase 1)
- Top-Down Coordination Decisions (Phase 2)
- Mixed-Integer Linear Program (MILP)
- Evaluation
- Conclusion
- Fully Distributed Coordination
- Machine Learning Coordination Approaches
- Machine Learning for Dynamic Resource Allocation
- Introduction
- Related Work
- Problem Statement
- Machine Learning Approach
- Overview
- Performance Profiling
- Learning Resource Requirements
- Integration with Coordination Approaches
- Evaluation
- Conclusion
- Self-Learning Coordination
- Introduction
- Related Work
- Problem Statement
- DeepCoord DRL Approach
- Joint Scheduling, Scaling, and Placement
- Partially Observable Markov Decision Process (POMDP)
- DRL Coordination Framework
- Implementation and Deployment
- Evaluation
- Evaluation Setup
- Self-Adaptation to Varying Scenarios
- Generalization to Unseen Scenarios
- Optimizing Multiple Objectives
- Scalability
- Conclusion
- Self-Learning Distributed Coordination
- Introduction
- Related Work
- Problem Statement
- Distributed DRL Approach
- Joint Scaling, Placement, Scheduling, and Routing
- Partially Observable Markov Decision Process (POMDP)
- DRL Coordination Framework
- Evaluation
- Evaluation Setup
- Varying Traffic Patterns
- Varying Deadlines
- Generalization to Unseen Scenarios
- Scalability
- Conclusion
- Self-Learning Coordination for Mobile Networks
- Introduction
- Related Work
- Problem Statement
- DeepCoMP: Centralized DRL
- DD-CoMP & D3-CoMP: Distributed DRL
- Trade-Offs and Design Choices
- Partially Observable Markov Decision Process (POMDP)
- DD-CoMP Algorithm
- D3-CoMP Algorithm
- Architecture and Deployment Options
- Evaluation
- Evaluation Setup
- Self-Adaption to Varying Scenarios
- Generalization and Transfer Learning
- Scalability
- Conclusion
- Future Research and Conclusion
- Acknowledgments
- Acronyms
- List of Algorithms
- List of Figures
- List of Tables
- Bibliography: Own Work
- Bibliography: Others' Work