Dissertations, Theses, and Capstone Projects

Date of Degree

5-2019

Document Type

Dissertation

Degree Name

Ph.D.

Program

Computer Science

Advisor

Spiridon Bakiras

Committee Members

Abdullah Uz Tansel

Sven Dietrich

Efstathios Zachos

Subject Categories

Digital Communications and Networking

Keywords

Private, Information, Retrieval

Abstract

A vast amount of today's Internet users' on line activities consists of queries to various types of databases. From traditional search engines to modern cloud based services, a person's everyday queries over a period of time on various data sources, will leave a trail visible to the query processor, which can reveal significant and possibly sensitive information about her. Private Information Retrieval (PIR) algorithms can be leveraged for providing perfect privacy to users' queries, though at a restrictive computational cost. In this work, we consider today's highly distributed computing environments, as well as certain secure-hardware devices, for optimizing existing PIR solutions. In particular, we initially employ available secure-hardware in a novel approach with the goal of providing faster and constant private query responses, by sacrificing some degree of privacy. Further on, we utilize the widely used Message Passing Interface (MPI) protocol for designing a library which can be used in third party software for performing private queries.

Share

COinS