Privacy preservation of genome data analysis using homomorphic encryption

International audienceVery few organizations can afford the necessary infrastructure for genomic data analytic, which requires a very large amount of storage and computational resources. In this context, cloud computing platforms can be adopted as a practical solution for storage and computations. However, sharing such sensitive information with cloud providers can lead to violating privacy preservation and public regulations. To resolve this problem, homomorphic encryption (HE) can be used. Homomorphic encryption enables computation over encrypted data, which helps tackling the problem of privacy preservation. Despite this advantage, existing HE schemes suffer from high computational complexity and storage overhead; designing a practical HE scheme that provides simultaneously the efficiency and the required level of security still remains an open question. In this paper, we propose a secure cloud based scheme for storing and analyzing classified genetic data using homomorphic encryption (HE). In our scheme, we adopted an optimization technique based on decomposing a homomorphic cipher text into a number of independent cipher texts with lower storage overhead using the Chinese remainder theorem (CRT). Optimization is then accomplished by applying parallel processing over the independent cipher texts. The presented technique is applied to improve the efficiency of two well-known HE schemes: the Domingo Ferrer (DF) and the Paillier cryptosystems. After examining the correctness of the proposed optimization, it is used to design a secure cloud-based application dedicated for genome data analysis. A virtual cloud environment was created for this purpose. Different performance and security analyses have shown the efficiency of such solution and its compatibility with real-world applications. The execution time is reduced to more than half while maintaining a high security level