Next-Generation AI-Driven Threat Detection in Smart and Connected Public Safety Networks

Abstract

As the world of smart, connected public safety networks expands in scale and sophistication, it is increasingly susceptible to attacks combining cyber and physical capabilities. The chapter suggests a novel framework for threat detection based on AI, which involves the fusion of multimodal intelligence, graph neural reasoning, transformer-based contextual learning, reinforcement learning for adaptive threat mitigation, and zero-trust behavioral security mechanisms. The envisioned architecture is based on decentralized edge-assisted analytics, which provides real-time threat monitoring in low-latency environments in heterogeneous public safety applications. The large-scale simulated smart city infrastructures with multimodal surveillance, communication and IoT traffic datasets were used in the experimental evaluation under dynamic attack conditions. The framework successfully managed to detect the attacks with 98.74% accuracy while also maintaining low false alarm rates and tolerating adversarial perturbations and high-density network conditions.