Lightweight Neuromorphic‐Temporal Graph Framework for Proactive Defense Against Evolving Cyber Attacks
Year: 2025
Type: article
Abstract: ABSTRACT The growing complexity of cyber threats, driven by intelligent malware and adaptive intrusion techniques, has exposed the limitations of conventional, reactive cybersecurity systems. To address this challenge, this paper presents a Lightweight Neuromorphic‐Temporal Graph Framework (LNTGF) that combines neuromorphic intelligence with temporal graph reasoning to achieve proactive and self‐adaptive defense in IoT and 5G/6G environments. The proposed framework employs spiking neural networks (SNNs) to perform real‐time, low‐power anomaly detection at the network edge, while contextually modulated temporal graph neural networks (CMTGNNs) model the evolving relationships between connected devices and threats. This dual‐layer design allows the system to anticipate attack patterns, adjust security policies dynamically, and mitigate risks before compromise occurs. Experiments using benchmark IoT datasets demonstrate 96.3% detection accuracy, 28% faster mitigation response, and a 65% reduction in energy consumption compared with existing models. The results confirm that integrating neuromorphic adaptability with graph‐based temporal intelligence offers an effective path toward scalable, energy‐efficient, and predictive cybersecurity. The proposed framework marks a step forward in designing intelligent defense architectures capable of learning, reasoning, and evolving alongside the threats they confront.
Source: Security and Privacy
DOI: 10.1002/spy2.70163
Authors: Haythem Hayouni, Farah Jbali
Institution: University of Jendouba
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4417310982