Sustainable Commercial Fishery Control Using Multimedia Forensics Data from Non-trusted, Mobile Edge Nodes

Abstract

Uncontrolled over-fishing has been exemplified by the UN as a serious ecological challenge and a major threat to sustainable food supplies. Emerging trends within governing bodies point towards digital solutions by deploying CCTV-based video monitoring systems on a large scale. We conjecture that such systems are not feasible when reliant on satellite broadband in remote areas, and expose workers aboard fishing vessels to unneeded manual surveillance. To facilitate this, we propose Dorvu, a AI-based multimedia distributed storage system designed for edge environments, with a specific focus on commercial fishery monitoring. Dorvu addresses the challenges of secure data storage, fault tolerance, availability, and remote access in hostile edge environments. The system employs a novel data distribution scheme involving sensor readings and AI video content extraction to ensure the preservation of forensic evidence even in unstable conditions. Experimental evaluations demonstrate the feasibility of real-time multimedia data collection, analysis, and distribution in networks of edge devices on-board active fishing vessels. Dorvu offers a practical alternative to current governmental surveillance trends that compromise data security and privacy, and we propose it as a solution for edge-based forensic data management in commercial fisheries and similar applications.