A fuzzy logic-based xApp for dynamic resource allocation in O-RAN network slicing
Computer Communications, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.comcom.2026.108656
- Dergi Adı: Computer Communications
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Compendex, INSPEC, Library, Information Science & Technology Abstracts (LISTA), Information Science & Technology Abstracts (LISTA), Business Source Ultimate (EBSCO), Engineering Source (EBSCO)
- Anahtar Kelimeler: 5G networks, Fuzzy logic, Network slicing, O-RAN, Resource allocation, xApp
- Bursa Uludağ Üniversitesi Adresli: Evet
Özet
One of the primary challenges in 5G and beyond networks is managing the coexistence of heterogeneous services (eMBB, uRLLC, mMTC). Due to fluctuating traffic conditions, satisfying diverse Service Level Agreement (SLA) requirements remains difficult using standard static allocation methods. This paper investigates this problem within the Open Radio Access Network (O-RAN) architecture by proposing a dynamic Physical Resource Block (PRB) allocation mechanism governed by a fuzzy logic-based xApp for the Near-Real-Time RAN Intelligent Controller (Near-RT RIC). Operating seamlessly within the O-RAN near-real-time control loop, the proposed model processes multi-dimensional inputs—such as channel quality, slice load, and latency—and employs a deterministic “Proportional Scaling” mechanism alongside a uRLLC-specific “Elastic Pool.” To evaluate its feasibility, the proposed Fuzzy xApp was compared against both a static heuristic baseline and a Reinforcement Learning (Q-Learning) agent. Simulation results indicated that while the Q-Learning agent achieved low average latency, it suffered from unpredictable latency spikes. In contrast, the Fuzzy model strictly protected uRLLC SLA constraints, bounding the 95th-percentile latency to 3.58 ms. Furthermore, it efficiently increased the proportion of users in the high Quality of Service (QoS) class to 52% compared to 28% in the static baseline. Ultimately, this study demonstrates that the proposed fuzzy logic system offers a transparent, zero-training, and computationally lightweight (∼33.7 ms) alternative to RL models, making it highly feasible for real-time intelligent network automation.