Energy-aware detection of threat information propagation speed in social network of things using XGBoost and sequential pattern mining

In the rapidly expanding landscape of the Social Network of Things (SNoT), ensuring real-time cyber threat detection while maintaining device energy efficiency poses a significant challenge. Conventional intrusion detection approaches often fail to balance accuracy with energy constraints, limiting their scalability in resource-constrained IoT ecosystems. To address this gap, we present an Energy-Aware Threat Propagation Detection Framework (EAPDF) that combines temporal sequence analysis with energy consumption profiling to identify and characterize malicious propagation behaviors. The framework employs PrefixSpan-based sequential pattern mining and XGBoost classification, enhanced with two novel evaluation metrics: Threat Propagation Delay (TPD) and Threat Propagation Energy Footprint (TPEF), which quantify both the timing and energy cost of attacks. Performance was validated on the SNoT-IDS2025 dataset, containing 25,000 labeled IoT attack instances, where EAPDF achieved 99.21% accuracy, 0.994 AUC-PR, an average energy usage of 0.042 Joules, and an average detection delay of 1.83 ms. Compared with benchmark datasets such as CICIDS2017 and IoT-23, the proposed model demonstrated superior scalability, early-stage detection capability, and suitability for edge-deployable security systems. Beyond performance gains, this work provides a reproducible and data-driven framework with applicability to diverse domains including smart cities, healthcare IoT, and industrial control systems, thereby advancing energy-aware cyber threat intelligence in next-generation IoT networks.