Game-Theoretic and Learning-Based Optimization Frameworks for Adaptive Resource Management in Next-generation Wireless Networks

다국어 초록 (Multilingual Abstract)

Next-generation wireless networks (NGWNs) aim to deliver seamless connectivity through unified architectures spanning cellular, Wi-Fi, and non-terrestrial systems. This integration worsens resource-optimization challenges, as dense, heterogeneous deployments and dynamic traffic require rapid and scalable decision-making. Centralized or model-driven methods struggle under these conditions, leading to inefficiencies and QoS degradation. Distributed approaches that can reason about interdependent network decisions and adapt online are therefore essential. Game-theoretic models capture strategic interactions and promote stable, efficient operating points, while learning-based methods provide the agility needed under uncertainty—together forming a strong basis for autonomous resource optimization in NGWNs.

In Part I, we develop a distributed load balancing algorithm for multi-radio access technology (multi-RAT) heterogeneous networks (HetNets). The random deployment of small cells and the user mobility make the network load distribution uneven, which degrades overall network capacity and the QoS for users. Furthermore, the disparate capabilities of multiple RATs such as the different propagation delays in terrestrial and non-terrestrial RATs affect the latency requirements. To balance the network load in multi-RAT HetNets, we propose a distributed game-theoretic algorithm considering QoS for users. To that end, a cost function is defined to reflect the cell load status, required data rates, and delay constraints. In the proposed algorithm, overloaded cells iteratively minimize user cost in a distributed manner by associating a user with a less-loaded cell based on its cost. We provide analysis to show that minimizing each user cost in the game theory setting achieves a balanced load distribution among cells. We show that the proposed algorithm brings each cell load to a balanced state in a finite number of iterations. Using simulation, we show that the proposed algorithm achieves superior performance in terms of even load distribution, network throughput, and the number of users with satisfactory QoS.

In Part II, we develop a load-aware, load-balancing procedure for fifth-generation (5G) New Radio-Unlicensed (NR-U) networks in order to address performance degradation and resource inefficiencies caused by load imbalance. Load imbalances frequently occur in NR-U networks due to factors such as the dynamic spectrum, user mobility, and varying traffic demand. To tackle these challenges, a load-aware, load-balancing procedure utilizing game theoretic reinforcement learning (GT-RL) is introduced. For load awareness, an extended System Information Block (SIB) is incorporated within the framework of 5G wireless networks. The load-balancing problem is addressed as a game theoretic cost-minimization task combining conditional offloading with reinforcement learning traffic-steering to dynamically distribute loads. Reinforcement learning applies a game theoretic policy to move users from overloaded cells to less congested cells that best serve their needs. Analytically, the proposed method is proven to spread the network load toward equilibrium. The proposed method is validated through simulations that show the effectiveness of its load balancing. The proposed method achieved better performance than previous work by attaining lower load variances while achieving higher throughput and greater quality of service satisfaction. Especially under high-load dynamics, the proposed method achieved an 8% gain in UE satisfaction with QoS and a 7.61% gain in network throughput compared to existing RL-based approach, whereas compared to the non-AI approaches, UE QoS satisfaction and the network throughput were enhanced by more than 15%.

목차 (Table of Contents)

• Vita i
• Acknowledgements ii
• Abstract iii
• Table of Contents vi
• 1 Introduction 1
• 1.1 Background and Motivation 1
• 1.2 Game-Theoretic and Learning-Based Approaches in Wireless Networks 3
• 1.2.1 Game-Theoretic Frameworks 4
• 1.2.2 Learning-Based Optimization 4
• 1.3 Research Necessities and Scope 6
• 1.4 Publications Contributing to the Dissertation 7
• 1.5 Contributions and Significance 8
• 1.6 Organization of the Dissertation 9
• 1.7 Conclusion 9
• 2 Game Theory and Reinforcement Learning Preliminaries 10
• 2.1 Game Theory and Resource Allocation Games 10
• 2.1.1 Congestion Games 11
• 2.1.2 Atomic Splittable Congestion Games 11
• 2.2 Reinforcement Learning in Wireless Networks 12
• 2.2.1 Single-Agent RL 12
• 2.2.2 Multi-Agent Reinforcement Learning (MARL) 13
• 2.3 Hybrid Game-Theoretic and Learning-Based Methods 13
• 3 Distributed load balancing algorithm considering QoS for next generation multi-RAT HetNets 14
• 3.1 Introduction 14
• 3.2 System Model 18
• 3.2.1 Network Architecture 18
• 3.2.2 Load and QoS Measure Metric for the Multi-RAT 20
• 3.2.3 Problem Formulation 23
• 3.3 Wardrop Equilibrium for Load Balancing in Multi-RAT HetNets 24
• 3.3.1 Wardrop Model for 5G HetNets 25
• 3.3.2 Equilibrium Analysis for Load Balancing 26
• 3.4 The Proposed Distributed Load Balancing Algorithm 30
• 3.4.1 Discussion 33
• 3.5 Performance Evaluation 36
• 3.5.1 Impact on Load Distribution 40
• 3.5.2 Impact of Network Load Variations 42
• 3.5.3 Impact of Threshold Settings 43
• 3.5.4 Impact of User Mobility 44
• 3.5.5 Impact of Large-Scale Networks 45
• 3.6 Closing Remarks 46
• 4 NR-U Network Load Balancing: A Game Theoretic Reinforcement Learning Approach 48
• 4.1 Introduction 48
• 4.2 System Model and Problem Formulation 53
• 4.2.1 The Network Model 53
• 4.2.2 NR-U Channel Access Procedure 55
• 4.2.3 Channel Model 56
• 4.2.4 Load Measurement 56
• 4.2.5 Problem Formulation 57
• 4.3 Game Theoretic Reinforcement Learning Model of the Load Balancing Problem 60
• 4.3.1 Game Theory 60
• 4.3.2 Reinforcement Learning 61
• 4.4 Proposed Load-Balancing Procedure and Algorithm 62
• 4.4.1 Operational Framework 62
• 4.4.2 Load Balancing 65
• 4.4.3 Discussion of Convergence, Complexity, and Scalability 68
• 4.5 Performance Evaluation 70
• 4.5.1 Convergence Rate Comparison 73
• 4.5.2 Impact of Network Load 75
• 4.5.3 Impact of Load Dynamics 78
• 4.5.4 Impact of Network Size 81
• 4.6 Closing Remarks 83
• Publications 85
• References 86