EAGER: Protecting Data Access Pattern Privacy in Hybrid Cloud Storage Systems

Iowa State University, Ames IA

Investigators

Abstract

Cloud-based storage services are gaining increasing popularity for their attractive pay-as-you-go model and high availability, reliability and economic efficiency. Meanwhile, there is an increasing privacy consciousness among users regarding this storage paradigm. While encryption provides some protection for data privacy, it cannot protect data access patterns, which can reveal private information about cloud storage clients. In particular, a curious owner or employee of a cloud storage service or an attacker invading the system can observe a client's access patterns, develop a model relating the patterns to the client's activities and later on use the model and observed access patterns to infer or predict the client's activities. Although several schemes have been proposed to protect the access patterns, in particular Oblivious Random Access Memory (RAM), it is hard to put these into practice owing to the high communication, storage, or computational overheads they incur. Towards addressing this problem, the project aims to offer an efficient, scalable and practical solution using Oblivious RAM that can protect the privacy of access pattern and can seamlessly integrate with existing cloud storage infrastructures. Specifically, the project leverages the emerging hybrid cloud storage architecture that has a cloud storage gateway with a moderate level of resource at the client side, the well-known reference locality principle for data access, and the availability of multiple independent cloud storage servers in the market. The project would benefit the community by enhancing the users' awareness of security and privacy risks in using cloud services and providing them with user-friendly protection tools so that they benefit from using such services with confidence. The project designs a new hierarchical, Oblivious RAM storage system to include multiple layers of Oblivious RAM modules, each of which is optimized for different performance metrics depending on its niche on the hierarchy; hence, the hierarchy as whole can attain high efficiency in communication, storage and computation simultaneously. The project formalizes the problem of protecting the data access pattern for the whole hierarchy of Oblivious RAM modules, and develops novel algorithms to solve the problem. The project also aims to deliver a set of provably-secure Oblivious RAM algorithms optimized for short data access delay or low server-side storage overhead, and a set of provably-secure algorithms for planning an optimal architecture for a hierarchy of Oblivious RAM modules, configuring the hierarchy and coordinating the operations of all Oblivious RAM modules in the hierarchy. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.