Efficient privacy-preserving data distribution in outsourced environments: a fragmentation-based approach

With the increasing cost of maintaining IT centers, there is a trend among organizations to outsource data management functions to a third-party service provider to reduce storage and computational cost. However, this opens the door for privacy violations. Existing approaches for protecting data confidentiality are based on encryption or a combination of encryption and fragmentation. In this paper, we propose an approach based only on fragmentation. In particular, we discuss the issue of employing both vertical and horizontal fragmentation to a database relation so that a minimum amount of data is stored at the owner. We represent the privacy (confidentiality) constraints as a graph. The constraint graph may have some cycles. We employ the two-coloring technique for the acyclic portion of the graph. We propose some heuristic algorithms to eliminate cycles and complete the coloring of all the nodes in the graph. This leads to an effective fragmentation process. The algorithm assigns one set of nodes to owner and the other to the external server. The proposed scheme considers both data dependent and data independent confidentiality constraints. In addition, it considers dependency constraints. We illustrate the proposed scheme using an example and show its efficacy. We provide proof of correctness for the proposed scheme.