EdgeShield: Attack resistant secure and privacy-aware remote sensing image retrieval system for military and geological applications using edge computing

• Published in: Earth science informatics Vol. 17; no. 3; pp. 2275 - 2302
• Main Authors: M, Ajitesh, M, Deekshith, Amaithi Rajan, Arun, V, Vetriselvi, D, Hemanth
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.06.2024; Springer Nature B.V
• Subjects: Availability; Earth and Environmental Science; Earth Sciences; Earth System Sciences; Edge computing; Feature extraction; Harnesses; Image enhancement; Image retrieval; Information Systems Applications (incl.Internet); Integrity; Military applications; Ontology; Privacy; Remote sensing; Retrieval; Simulation and Modeling; Space Exploration and Astronautics; Space Sciences (including Extraterrestrial Physics; Trusted third parties; Privacy preservation; Remote sensing images; Third party auditing; CBIR; Distributed cloud; Edge computing
• ISSN: 1865-0473; 1865-0481
• DOI: 10.1007/s12145-024-01256-z

The increasing production and processing of image data, especially in remote sensing applications, has raised concerns regarding image security, privacy, and efficient retrieval as it is most widely used in sensitive applications. In this article, to address these challenges, a novel privacy-preserving content-based image retrieval (PPCBIR) system has been proposed that leverages a trusted edge computing layer that performs image encryption and feature extraction tasks, reducing the processing overload on user devices and bolstering system efficiency. Feature extraction harnesses the MobileNetV2 deep learning model, which enables the extraction of intricate visual features, enhancing image retrieval accuracy in the presence of high inter-class similarity in the dataset. Furthermore, the system has been deployed in a distributed storage environment, ensuring image availability even during server outages. The proposed system also incorporates trusted third-party auditing (TPA) as a means to verify the integrity of images during the storage and retrieval processes. The presence of TPA plays a crucial role in maintaining the reliability and trustworthiness of the stored images. The proposed system achieves a high mean Average Precision (mAP) of 0.889, surpassing existing PPCBIR systems. Overall, the system prioritizes image retrieval performance, privacy, availability, and integrity, making it suitable for processing remote sensing image data efficiently and securely.