Explainable Artificial Intelligence for Conflict Management: XAI4C for Conflict Detection and Mitigation in O-RAN Near-RT RIC

Artikel i vetenskaplig tidskrift, 2025

The open radio access network (O-RAN) architecture introduces a modular approach to network design, enabling flexibility by decoupling hardware and software while fostering innovation in a multivendor ecosystem. However, its distributed framework creates challenges in managing network control conflicts across various components. In this work, we provide a comprehensive overview of conflict management in O-RAN, focusing on near real-time RAN intelligent controller (Near-RT RIC) conflicts, and state-of-the-art strategies for addressing these challenges. We analyze the limitations of current approaches through the lens of explainable artificial intelligence (XAI). Thereafter, we propose XAI4C, a novel framework for conflict detection and mitigation among xApps in Near-RT RIC in O-RAN that leverages XAI to provide more transparent and explainable actions. Our solution improves detection accuracy by 30% and reduces the detection latency by 41% compared to a state-of-the-art benchmark, while also effectively mitigating conflicts to enhance network performance. Finally, we discuss future research directions and highlight key challenges in XAI-driven O-RAN conflict management, critical for adapting to the increasing complexity of next-generation network systems.