Answers are grounded in this paper and cite their sources.
The security of 5G networks is challenged by their highly heterogeneous architecture, ultra-low latency requirements, and the presence of multi-stage, cross-layer cyber threats. Conventional anomaly detection systems, which typically operate at a single temporal scale and remain isolated from operational management workflows, are insufficient to address these challenges. In this paper, we propose a hierarchical multiscale anomaly detection framework that integrates discrete wavelet transform (DWT)–based traffic decomposition with convolutional long short-term memory (ConvLSTM) networks to jointly capture fine-grained packet-level dynamics and long-term behavioral patterns across the 5G protocol stack. Network traffic is decomposed into multiple temporal subbands using DWT, and each subband is analyzed through dedicated ConvLSTM branches to model spatiotemporal dependencies. A hierarchical cross-layer attention mechanism fuses multiscale insights, prioritizing correlated anomalies while suppressing transient noise. To operate detection, the framework integrates with IT service management (ITSM) systems by translating anomaly scores into standardized STIX 2.1 threat intelligence alerts, enabling automated incident response workflows. Experimental evaluation using telemetry collected from a controlled multi-vendor 5G testbed with representative attack scenarios demonstrates improved detection accuracy for complex multi-stage attacks, reduced false alarm rates, and lower processing latency compared to single-scale baselines. The proposed framework provides a practical and scalable solution for adaptive, intelligent security management in next generation 5G networks.
No takes yet. Share an insight, caveat, or question.
Alam et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: