Saudi Arabian university investigates use of AI, blockchain to detect floods and save lives

Saudi Imam Abdulrahman Bin Faisal University’s Department of Informatics and Saudi Aramco Cybersecurity Chair recently published a study in MDPI exploring a solution for Smart Flood Detection, aiming to save lives through the integration of AI, Blockchain and drones.

Flooding is a serious hazard that requires immediate handling and strategic response measures. Despite its location on the Persian Gulf, the Saudi city of Mecca has experienced an increase in flooding due to climate change over the past decade. Surrounded by mountains, Mecca faced torrential rain on 23 December 2022, resulting in the unfortunate loss of many vehicles.

To combat these challenges, the researchers propose a Flood Detection Secure System (FDSS) as a secure means of flood detection in Saudi Arabia. FDSS leverages deep active learning (DeepAL)-based classification models within a federated learning framework, which effectively minimizes communication costs and maximizes global learning accuracy.

The abstract of the study explains that the proposed system includes blockchain-based federated learning and partially homomorphic encryption (PHE) to ensure privacy. Using imagery and IoT data, FDSS can train local models capable of detecting and monitoring floods. This innovative approach enables the estimation of inundated areas and facilitates the tracking of rapid changes in dam water levels, providing crucial insights into the flood threat. The study concludes by discussing the proposed method and its potential challenges in managing floods in remote regions using artificial intelligence and blockchain technology.

Furthermore, the study introduces a drone application that uses blockchain technology to safely and effectively manage flooding in remote areas in real time. The system’s architecture ensures data integrity and enhances the data analysis functions of the management system, thereby preventing information manipulation. By recording text data as part of transaction information in the blockchain network, operators can access transaction data through a visualization platform, thereby facilitating oversight of operations.

The researchers also present a scheme that improves the performance of the federated learning (FL) system. This scheme leverages DeepAL to select optimal edge nodes and integrates the learned model parameters into a blockchain-based FL approach to improve reliability and security. To achieve a high level of privacy and security features, modern cryptography techniques are used, including homomorphic encryption.

In the context of natural disasters, real-time data collection by unmanned aerial vehicles (UAVs) can effectively prevent damage by enabling efficient operations. UAVs can take aerial photos, measure water levels, wind speeds and water velocities, and predict weather events, thereby helping to prevent disasters and assist in rescue efforts. These complex interactions are made possible by the use of AI, which is a computer-based system capable of performing tasks that require intelligence.

By incorporating AI and machine learning, systems can effectively defend against new and sophisticated attacks with evolving capabilities. Drones must be equipped with a collective machine learning model that integrates data from IoT devices and webcams. This data can be transferred to Mobile Edge Computing (MEC) to create a robust algorithm with strong predictive capabilities.

The proposed framework assumes that UAVs collect data, which is then stored in the blockchain by MEC servers. This data includes important information such as device name, MAC address, type, and geographic data such as latitude and longitude, which facilitates data collection by MEC servers. Before adding data to the blockchain, MEC servers ensure the validity of UAVs.

To improve rescue efforts during floods and other catastrophic weather events in hard-to-reach areas, the study utilizes the Internet of Drones (IoD). IoT devices are used to collect data on the location and condition of individuals in affected areas, including vital signs, enabling the prioritization of rescue operations.

Collected data is sent to a central server, where deep learning algorithms analyze the information and formulate a comprehensive rescue plan. This plan is then shared with relevant organizations involved in rescue efforts, enabling quick and effective assistance to those in need.

In conclusion, the study underlines the significant potential of their system to improve the effectiveness and efficiency of rescue operations during disaster situations. By harnessing the power of AI, blockchain and IoT technologies, the system can quickly analyze massive amounts of data and generate comprehensive rescue plans, ultimately saving more lives.