Integrating Fog Computing With AI for Real-Time Disaster Management in Smart Cities

This paper explores the integration of Fog computing and AI for real-time disaster management in smart cities. Traditional cloud-based systems face latency and scalability issues during crises. The proposed hybrid Fog-Cloud architecture decentralizes data processing using edge nodes with lightweight AI models, such as quantized LSTMs and CNNs, achieving 88% lower latency than cloud-only systems. It ensures sub-500 ms responsiveness for urgent situations like flood evacuations and earthquake alerts. Tests show a 94.1% F1-score for flood prediction and 89.2% accuracy for fire detection, outperforming threshold-based methods. Scalability tests confirm support for up to 10,000 IoT nodes, ensuring functionality in megacities. By activating safety protocols at the edge and feeding insights into the cloud, it balances low-latency processing with centralized scalability. This serves as a blueprint for adaptive smart city upgrades. Future work includes energy-harvesting Fog nodes and 6G-enabled network slicing for enhanced disaster response.