Study and Implementation of Homomorphic Encryption Based Protocols
Loading...
Files
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Unuversity Setif 1 Ferhat Abbas . Faculty of Sciences
Abstract
Homomorphic encryption is a promising solution for ensuring the confidentiality of sensitive data while allowing its processing without prior decryption. This technology is particularly suitable for cloud environments, where medical and forensic data require a high level of protection.
This thesis presents a study of the main homomorphic encryption schemes, namely Paillier (PHE), DGHV (SHE), and CKKS (FHE), along with their characteristics, advan-tages, and limitations. A case study on secure post-mortem interval (PMI) estimation was conducted based on Henssge’s cooling law. Computations were performed on encrypted data to preserve the confidentiality of information extracted from autopsy reports.
Experimental results show that the Paillier scheme ensures good accuracy but has a high computational cost. The DGHV scheme enables homomorphic operations on integers but requires mathematical approximations and remains limited by noise accumulation. The CKKS scheme appears to be the most suitable for medical applications due to its ability to handle real numbers directly with a good trade-off between accuracy, security, and performance.
Finally, a contribution based on a bootstrapping-inspired approach was proposed to mitigate noise accumulation in CKKS. The results show improved computational stability and reduced error in PMI estimation, confirming the relevance of this approach for secure computation over encrypted medical data.
