A Drone-Assisted Deep Learning-Based IoT System for Monitoring Ship Emissions in Ports Considering Adversarial Attacks
Mahmoud Elsisi, J. C. Yu, Chun-Che Lai, Chun‐Lien Su
Abstract
Monitoring ship emissions poses an intricate interdisciplinary challenge, encompassing atmospheric science, transportation, information communication, and computer technology. Key stakeholders in this effort include maritime institutions and shipping companies. However, effectively organizing information from ship emission monitoring, especially for policy implementation in the expansion of emission control areas (ECAs), faces substantial obstacles in data acquisition, transmission, analysis, and information services. To address these challenges, this study suggests an Internet of Things (IoT) paradigm using drones and deep learning for ship emission monitoring in the port areas, integrating vital technologies like industrial contact elements for IoT, assisted processing of ship emission, real-time data monitoring and online processing, and tailored information services for port control center. Additionally, the suggested IoT paradigm considers adversarial and false data injection (FDI) assaults, which impair the effectiveness of data processing and identification techniques. By actively supporting green port trends and maritime law enforcement, this new IoT paradigm is developing into the framework of the berth area inside the physical infrastructure. Furthermore, this model highlights the possibility for a more widespread adoption of ship pollution monitoring while simultaneously demonstrating the model’s efficacy in the current setting. A more expansive version of the suggested IoT paradigm might function as a strong cyber-physical infrastructure, supporting the capacity for thorough ship emission monitoring and enabling in-depth study on ship emissions.