No open-access location is known for this record yet.
No open-access location is known for this record yet.
Natural and human-induced disasters such as floods, earthquakes, wildfires, and cyclones present significant threats to communities by disrupting lives, infrastructure, and ecosystems. The unpredictability of these events makes disaster management a critical discipline, requiring systems that can anticipate hazards, provide timely alerts, and support rapid decision making. Conventional disaster management approaches, including manual monitoring, centralized data collection, and statistical prediction models, have shown limited scalability, slower response, and reduced adaptability when challenged with complex and dynamic environments. To address these challenges, this study introduces a hierarchical Cloud-Internet of Things (IoT) framework supported by fifth generation (5G) connectivity, designed to integrate real-time sensing, edge analytics, and cloud-based forecasting into a unified disaster response system. At the edge, an Autoencoder enables low-latency anomaly detection, while the cloud employs a Long Short-Term Memory (LSTM) model to forecast disaster evolution and support proactive planning. The proposed approach was evaluated using IoT sensor-based dataset, demonstrating anomaly detection accuracy consistently exceeding 98.5%, forecasting accuracy with a determination coefficient (R2) of up to 0.997, and end-to-end latency below 2 seconds. Robustness testing under varying network conditions further demonstrated the stability of the framework, with timely response rates of 99.4% under normal 5G conditions, 98.1% under degraded performance, and 96.8% even under severe congestion. These findings establish the proposed framework as a reliable and scalable solution for real-time disaster management, combining rapid local responsiveness with accurate predictive intelligence to support effective emergency interventions.
Loading…
Stored only when both papers are in the corpus — 4 citations exist upstream.
Papers that cite the same references — the relational fallback while embeddings are still backfilling.
Imported from OpenAlex (CC0) — nothing here is generated.