Faculty of Technical Sciences

This paper proposes Yggdrasil, a protocol for privacy-aware dual data deduplication in multi-client settings. Yggdrasil is designed to reduce Cloud storage space while safeguarding the privacy of Clients' data. This is achieved by exploiting a dual setting, where both the Cloud and the Clients store a fraction of the data. Yggdrasil combines two innovative techniques to achieve this goal. First, generalized deduplication, an emerging technique to reduce data footprint; second, nondeterministic lightweight transformations that insures a high level of privacy while improving the degree of cross-user data compression in the Cloud. Our Client preprocessing guarantees that an honest-but-curious Cloud storage provider faces a high degree of uncertainty as to determine what the original Clients' data are. We introduce an uncertainty metric to measure the privacy of the Client's outsourced data and three compression metrics to investigate the compression potential of Yggdrasil. Our experiments with a dataset of DVI files show that Yggdrasil achieves an overall compression rate of 43%, which means that Yggdrasil can represent the same database using less than half of the original space. Moreover, for the same experiment Clients only store 17% of the original data, the Cloud hosts the remaining 26%, and the Client preprocessing ensures each outsourced fragment has 10^293 possible original strings. Higher uncertainty is possible, with the expanse of reducing the compression capability.