Enforcing Confidentiality Constraints on Sensitive Databases with Lightweight Trusted Clients

Abstract

Existing approaches for protecting sensitive information stored (outsourced) at external "honest-but-curious" servers are typically based on an overlying layer of encryption that is applied on the whole information, or use a combination of fragmentation and encryption. The computational load imposed by encryption makes such approaches not suitable for scenarios with lightweight clients. In this paper, we address this issue and propose a novel model for enforcing privacy requirements on the outsourced information which departs from encryption. The basic idea of our approach is to store a small portion of the data (just enough to break sensitive associations) on the client, which is trusted being under the data owner control, while storing the remaining information in clear form at the external (honest-butcurious) server. We model the problem and provide a solution for it aiming at minimizing the data stored at the client. We also illustrate the execution of queries on the fragmented information.

Trust Based Service Optimization Selection for Cloud Computing

“…Literature [8] summarized the privacy of research results in the field, and described the basic principles of privacy protection technologies, then pointed out the future direction of privacy protection technologies. Literature [9][10][11] had effective combination confidential information and data decomposition. They proposed to achieve information decomposition using the concept of privacy constraints, which describe the data attributes needing to be encrypted and simultaneously back to the privacy of the data leaked.…”

An Index Structure for Private Data Outsourcing

“…2. The final method proposed for outsourcing data as plaintext was introduced in [6,5]. This method, similar to the method in [1] divides the data into two fragments.…”

“…The solution in [6,5] describes a scheme that relies on that the data owner to maintain a significant portion of the data. The amount of storage needed to store a single fragment only differs from storing the entire database by a constant factor.…”

“…The scenario we consider is the same one addressed by [1,3,4,5,6], primarily that the data to be outsourced is represented as a relational schema R (a 1 , a 2 , . . .…”

“…, a n } where the domain of attribute a i is denoted by D i . Formally, we let A be the set of all attributes and D be the set of all domains in R. We use the privacy model of confidentiality constraints that was originally introduced in [12], and subsequently has been used in [1,3,4,5,6,7,11]. Confidentiality constraints are sets of attributes within R that cannot be revealed together.…”