Collusion resistant secret sharing scheme for secure data storage and processing over cloud

Abstract

Shamir secret sharing (SSS) is considered as a promising method for outsourcing the data securely due to its ability to support privacy preserving data processing while ensuring data availability. Major drawbacks of original SSS scheme are its susceptibility to collusion attack and high storage overhead. Hence in this paper, we first propose a modified SSS scheme (MSSS) which can resist collusion attack and provide adequate security even with two shares. However, the storage overhead of this scheme is high when it is extended to ensure data availability and integrity in cloud storage systems. Therefore, a modified ramp secret sharing (MRSS) with reduced storage overhead compared to MSSS scheme is also proposed in this paper. The proposed schemes can be employed for any privacy preserving data processing application which involve linear operations on the data. In this paper, in order to demonstrate the capability of proposed schemes to support privacy preserving data processing, Haar discrete wavelet transform (DWT) computation on medical images is considered as an example as DWT is widely used in feature extraction for disease diagnosis from pathological images. We present an algorithm for computing Haar DWT from medical image shares. The security of the proposed scheme is evaluated through mathematical cryptanalysis and resistance against various statistical attacks. The performance analysis shows that shared domain DWT offers same accuracy levels as that of plaintext domain.