Privacy-preserving cloud-based secure digital locker with differential privacy-based deep learning technique

• Published in: Multimedia tools and applications 2024-03, Vol.83 (34), p.81299-81324
• Main Authors: Shanthi, P., Vidivelli, S., Padmakumari, P.
• Format: Article
• Language: English
• Subjects: Algorithms; Authentication; Cloud computing; Computer Communication Networks; Computer Science; Curves; Data Structures and Information Theory; Deep learning; Digital imaging; Encryption; Multimedia Information Systems; Privacy; Special Purpose and Application-Based Systems; Track 2: Medical Applications of Multimedia
• ISSN: 1573-7721; 1380-7501; 1573-7721
• DOI: 10.1007/s11042-024-18566-5

Practically in this modern era, data privacy preservation is somewhat a complex technique due to the development of advanced technologies. During user interaction, the confidentiality of data is ensured by privacy preservation. Many researchers have conducted on privacy preservation in cloud-based services to prevent privacy loss attacks. In this paper, a novel cloud-based secure digital locker with a differential privacy-based Deep Learning (DL) technique is proposed. This privacy preservation technique is implemented for a cloud-based digital image locker system. The operations followed in this privacy preservation technique are initialization, user registration, user authentication, image upload and retrieval, image encryption and differential privacy-based DL technique. In user authentication operation an improved Elliptic Curve Cryptography technique is proposed to do encryption on user-entered passwords. The SISSO algorithm is proposed for tuning the encryption key. Consequently, the image encryption process deals with improved image encryption by making changes in Arnold’s cat map and Chebyshav map. The proposed approach is finished off with a differential privacy-based DL technique which is introduced to maintain the privacy of user data from intrusion attacks. Thus, the efficiency of the proposed cloud-based secure digital locker with differential privacy-based DL technique is proved by several analyses over other existing privacy preservation techniques. Finally, the efficiency of the proposed SISSO algorithm is evaluated over traditional algorithms with respect to computation time, decryption time, encryption time, key sensitivity and latency. The computation time of the SISSO is minimal (11.294) over other models.