Title: PINED-RQ: A Differentially Private Index on Encrypted Databases for Supporting Range Queries
By Cetin Sahin, PhD student at UCSB (http://www.cs.ucsb.edu/~dsl/?q=content/cetin-sahin)
Date : 9/5/2016 16h
Salle : 1/124
Abstract: Despite the benefits of Database-as-a-Service cloud services (DBaaS), legitimate privacy concerns continue hindering their adoption when data is personal and sensitive. To tackle this problem, we propose PINED-RQ, a privacy-preserving auxiliary data structure allowing the DBaaS provider to perform range queries efficiently over an encrypted database. The data owner, who wants to outsource her database to a public cloud provider, initially builds a differentially private index over a database, and then encrypts the entire database using semantically secure encryption scheme before shipping it to the public cloud. After outsourcing to the cloud, the cloud server processes queries with the help of the index and return corresponding results to end-users. The system provides strong security by employing differential privacy and semantically-secure encryption scheme jointly. Moreover, this system is capable of supporting update operations. Supporting updates with a differentially private index is very challenging. However, PINED-RQ overcomes the challenges by managing privacy budget wisely and performing update operations with the help of the data owner. PINED-RQ also introduce a probabilistic query execution strategy to enable end-users to ask analytical queries with some statistical guarantees like recall, precision and confidence interval.
