A Survey and an Analysis of Conflict Management in O-RAN

The emergence of the Open Radio Access Network (O-RAN) introduces unprecedented flexibility, programmability, and multi-vendor interoperability into cellular systems. However, the concurrent operation of multiple xApps and rApps within the RAN Intelligent Controllers (RICs) creates new risks of control conflicts that may degrade network performance, stability, and quality of service. Despite its growing relevance in the field, existing research on conflict management remains fragmented and lacks a unified analytical perspective. This survey consolidates and analyzes conflict detection and mitigation mechanisms proposed so far for O-RAN. We first revisit the architectural foundations of conflict management in O-RAN and introduce a taxonomy that classifies existing solutions according to conflict type, control layer, detection mechanism, mitigation strategy, validation methodology, considered xApps/rApps, and performance metrics. Using this taxonomy, we systematically examine both architectural Conflict Mitigation Frameworks (CMFs) and algorithmic conflict-resolution mechanisms, further distinguishing the latter into pre- and post-action approaches. Our analysis reveals that the literature remains at an early stage of maturity, with a limited number of studies, simplified evaluation scenarios, and a strong concentration on a narrow set of xApp types and use cases. We further observe an evolution from rule-based and optimization-driven solutions toward graph-learning, explainable artificial intelligence (XAI), and Large Language Model (LLM)-assisted conflict-management mechanisms, whilst most proposals still rely on limited observability and reactive Key Performance Indicator (KPI)-driven operation. Based on the lessons learned from the surveyed works, we derive a set of evidence-based design guidelines that relate observability levels, deployment characteristics, and operational requirements to suitable conflict-management mechanisms. Finally, we identify key open challenges related to scalability, xApp diversity, multi-vendor observability, cross-layer coordination, explainability, temporal control-plane awareness, and the practical feasibility of AI-driven solutions, outlining future research directions toward scalable, proactive, explainable, and trustworthy conflict-aware O-RAN ecosystems.