BROWSE
Detail
Conference Paper
Supervised Learning of RLC Mode Switching Policy under QoE Constraints in 5G
Cited 0 time in 
Share 
Share
- Authors
- Nam-I. Kim, Jee-Hyeon Na
- Issue Date
- 2025-10
- Citation
- International Conference on Information and Communication Technology Convergence (ICTC) 2025, pp.504-505
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- In 5G networks, the choice between Acknowledged Mode (AM) and Unacknowledged Mode (UM) at the RLC layer directly affects service quality depending on traffic characteristics and network conditions. This paper proposes a supervised learning approach to dynamically select the optimal RLC mode based on two key QoE-related indicators: Round-Trip Time (RTT) and Packet Loss Ratio. Using a Random Forest classifier trained on simple rule-based labels, the model accurately predicts the appropriate mode across a range of conditions. Decision boundary visualization demonstrates clear policy learning, and results confirm the feasibility of lightweight AI for adaptive RLC control. The approach is interpretable, easy to implement, and suitable for real-time QoE-aware optimization in 5G systems.
- Keyword
- 5G New Radio (5G NR), QoE-aware Mode Selection, Radio Link Control (RLC), Supervised Learning
- KSP Keywords
- 5G networks, 5G system, Boundary visualization, Learning approach, Mode Selection, Packet Loss Ratio, Policy learning, Random Forest Classifier, Random Forests(RF), Round Trip Time(RTT), Rule-Based