BROWSE
Detail
Conference Paper
Resource Allocation Leveraging Dual Traffic and Channel Predictions in Open-RAN Architecture
Cited - time in 
Share 
Share
- Authors
- Yeonghun Jeong, Dongho Ham, Minhyun Kim, Jungmo Moon, Jeongho Kwak
- Issue Date
- 2026-02
- Citation
- International Conference on Artificial Intelligence in Information and Communication (ICAIIC) 2026, pp.578-583
- ISSN
- 2831-6983
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- In this paper, we first envision an agentic-AI framework for open-RAN architecture that enables autonomous and context-aware network control by integrating predictive intelligence into key resource-management loops. Building upon this architectural vision, we develop a dynamic resource-allocation mechanism that leverages dual traffic and channel predictions to guide power-budget decisions across cells. By jointly forecasting cell-level traffic demand and channel conditions using transformer-based AI models, the proposed mechanism acquires predictive knowledge of spatial-temporal cell environments, allowing power resources to be proactively aligned with expected demand. Within this framework, we design an adaptive power budget allocation algorithm that dynamically distributes the total transmit-power budget across cells based on anticipated traffic load and channel quality. Simulation results demonstrate that this predictive strategy achieves a 27.1 % improvement in Geometric Average Throughput (GAT) compared to a baseline scheme without predictive information, confirming the effectiveness of embedding predictive and agentic intelligence into open RAN resource control.
- Keyword
- Open RAN, resource allocation, power budget, transformer
- KSP Keywords
- Adaptive power, Allocation algorithm, Allocation mechanism, Average Throughput, Budget allocation, Channel quality, Context aware, Dynamic resource, Geometric average, Network Control, Predictive strategy