Behavior-Driven Slice-Aware Traffic and KPI Prediction for 6G Radio Access Networks
IEEE Access, vol.14, pp.116156-116169, 2026 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Volume: 14
- Publication Date: 2026
- Doi Number: 10.1109/access.2026.3713023
- Journal Name: IEEE Access
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, Directory of Open Access Journals
- Page Numbers: pp.116156-116169
- Keywords: behavior-driven modeling, CNN, deep learning, digital twin, GRU, LSTM, Network slicing, ns-3, radio resource management, slice-level KPIs, time-series prediction
- Ataturk University Affiliated: Yes
Abstract
Accurate prediction of traffic demand and slice-level key performance indicators (KPIs) is essential for enabling proactive resource management in next-generation radio access networks. However, most existing studies focus on aggregate traffic modeling and provide limited insight into slice-level dynamics under varying mobility and traffic conditions. This paper proposes a slice-aware deep learning framework for the joint prediction of traffic demand and multiple KPIs within a simulation-driven environment. A realistic multivariate dataset is generated using ns-3 by modeling user mobility variations, slice-specific traffic patterns, and adaptive radio configurations under varying traffic conditions. This enables controllable and reproducible data generation while capturing diverse slice behaviors across eMBB, URLLC, and mMTC scenarios. LSTM, GRU, CNN, and a traditional linear regression baseline are comparatively evaluated under a unified preprocessing and time-series validation framework. Experimental results demonstrate that CNN consistently achieves lower prediction errors across most KPIs, while recurrent models provide competitive performance for smoother traffic patterns. The results further show that deep learning models more effectively capture nonlinear slice-level dynamics compared with traditional forecasting approaches. Latency prediction remains the most challenging task due to its sensitivity to rapid traffic fluctuations and varying network conditions. Beyond prediction, the proposed framework provides a foundation for learning slice-level network dynamics and supporting future digital twin–oriented network representations and closed-loop optimization in 6G-ready wireless systems.