Gudala Prince, Renjith Prabhavathi Neelakandan
Control-plane signaling is essential in mobile communication networks, enabling mobility management, session control, and resource allocation. However, this dependence increases vulnerability to signaling anomalies, including signaling storms and service-based API flooding, which can deplete radio and core-network resources without corresponding user-plane traffic. These challenges are exacerbated in 5G and emerging 6G systems due to cloud-native core architectures, service-based interfaces, edge computing, and large-scale Internet of Things deployments. This study introduces an integrated analytical and AI-assisted framework for analyzing, detecting, and mitigating control-plane signaling anomalies in next-generation mobile networks. Radio Resource Control (RRC) behavior is represented using a continuous-time Markov chain that models state transitions, virtualization overheads, and queueing effects across core network functions such as AMF, SMF, and UPF. A scalable, event-driven digital-twin simulator is constructed to emulate realistic RAN–core interactions under both mobility and attack scenarios, calibrated through telemetry-driven parameter estimation. Anomaly detection employs a hybrid methodology that integrates analytical likelihood scoring with machine-learning models, including autoencoders, sequence models, and graph-based learning, to minimize false alarms across control-plane layers. Mitigation is formulated as a constrained decision problem and addressed using a safe reinforcement-learning strategy that adaptively regulates signaling activity while maintaining service quality. Simulations involving 10,000 heterogeneous user devices demonstrate up to a 45% reduction in peak signaling load and a 50% decrease in AMF queueing delay, while preserving application-level quality of service. These results offer practical guidance for enhancing control-plane resilience in future 6G networks.