Detail View

Dynamic IM Offloading for Cost Minimization in Open RAN : From Distributed Optimization to Centralized Deep Reinforcement Learning
Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

Title
Dynamic IM Offloading for Cost Minimization in Open RAN : From Distributed Optimization to Centralized Deep Reinforcement Learning
Alternative Title
비용 최소화를 위한 Open RAN 동적 간섭 관리 오프로딩: 분산 최적화에서 중앙집중형 심층 강화학습으로
DGIST Authors
Sojung LeeYoungsik KimJeongho Kwak
Advisor
김영식
Co-Advisor(s)
Jeongho Kwak
Issued Date
2026
Awarded Date
2026-02-01
Type
Thesis
Description
Open-RAN, 6G, interference management, opportunistic control, Lyapunov optimization
Table Of Contents
1 Introduction 1

2 Related Work 4
2.1 Centralized Interference Management 4
2.2 Distributed Interference Management 5
2.3 Hybrid Coordination Approaches 5
2.4 O-RAN-based Interference Management 6

3 System Model 8
3.1 Network Model 8
3.2 Radio Resource Model 9
3.3 Link Model 10
3.4 Offloading Model 11
3.5 Cost Model 11

4 Problem Formulation 13
4.1 Long-Term Network Utility Maximization 13
4.2 From Time-Averaged Objective to a Slot-Wise Formulation 14
4.3 Per-Slot Drift-Plus-Penalty Problem 17
4.4 Stage-Wise Decomposition 18
4.5 Three-Stage Online Solution of the Per-Slot Problem 19
4.6 Stage 1: Auxiliary Update (Backlog Regulation) 20
4.7 Stage 2: LOCAL vs. CENTRAL Mode Selection 20
4.8 Stage 3: Scheduling and Power under the Chosen Mode 20
4.8.1 LOCAL Mode: Lightweight RT-Side Refinement 21
4.8.2 CENTRAL Mode: Near-RT Joint Optimization via Deep RL 22
4.9 Operational Realization of ORION 24

5 Performance Evaluation 26
5.1 Simulation Setup 26
5.1.1 Network Configuration 26
5.1.2 DQN Architecture 26
5.2 Baseline Algorithms 26
5.3 Performance Metrics 27
5.4 Results and Analysis 27
5.4.1 Performance Comparison 27
5.4.2 Cost–Performance Trade-off 28
5.4.3 Impact of Cost Weight Parameter 29
5.4.4 Adaptive Offloading Behavior under Varying Interference Levels 30
5.5 Discussion 30
5.5.1 Performance and Fairness Improvement 30
5.5.2 Operational Cost Efficiency 31
5.5.3 Context-Aware Adaptation 31
5.5.4 Implications for Hierarchical O-RAN Control 31

6 Conclusion 32

References 34
URI
https://scholar.dgist.ac.kr/handle/20.500.11750/59712
http://dgist.dcollection.net/common/orgView/200000951643
DOI
10.22677/THESIS.200000951643
Degree
Master
Department
Department of Electrical Engineering and Computer Science
Publisher
DGIST
Show Full Item Record

File Downloads

  • There are no files associated with this item.

공유

qrcode
공유하기

Total Views & Downloads