Query log attack on encrypted databases

Abstract

Encrypting data at rest has been one of the most common ways to protect the database data against honest but curious adversaries. In the literature there are more than a dozen mechanisms proposed on how to encrypt data to achieve different levels of confidentiality. However, a database system is more than just data. An inseparable aspect of a database system is its interaction with the users through queries. Yet, a query-enhanced adversary model that captures the security of user interactions with the encrypted database is missing. In this paper, we will first revisit a few well-known adversary models on the data encryption schemes. Also, to model the query-enhanced adversaries we additionally need new tools, which will be formally defined. Eventually, this paper introduces query-enhanced adversary models which additionally have access to the query logs or interact with the database in different ways. We will prove by reduction that breaking a cryptosystem by a query-enhanced adversary is at least as difficult as breaking the cryptosystem by a common adversary. © 2014 Springer International Publishing.