Files in this item

FilesDescriptionFormat

application/pdf

application/pdfAGRAWAL-DISSERTATION-2015.pdf (857kB)Restricted to U of Illinois
(no description provided)PDF

Description

Title:Cryptographic agents
Author(s):Agrawal, Shashank
Director of Research:Prabhakaran, Manoj
Doctoral Committee Chair(s):Prabhakaran, Manoj
Doctoral Committee Member(s):Gunter, Carl; Borisov, Nikita; Vaidya, Nitin; Vaikuntanathan, Vinod
Department / Program:Computer Science
Discipline:Computer Science
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):cryptography
security framework
functional encryption
obfuscation
Abstract:Over the last decade or so, thanks to remarkable breakthroughs in cryptographic techniques, a wave of ''cryptographic objects'' -- identity-based encryption, fully-homomorphic encryption, functional encryption, and most recently, various forms of obfuscation -- have opened up exciting new possibilities for computing on encrypted data. Initial foundational results on this front consisted of strong impossibility results. Breakthrough constructions, as they emerged, often used specialized security definitions which avoided such impossibility results. However, as these objects and their constructions have become numerous and complex, often building on each other, the connections among these disparate cryptographic objects, and among their various security definitions, have become increasingly confusing. The goal of this work is to provide a clean and unifying framework for diverse cryptographic objects and their various security definitions, equipped with powerful 'reduction' and 'composition' theorems. We model the functionality desired from a cryptographic object via a 'schema' in an ideal world. Our new security definition, indistinguishability preservation, is parametrized by a family of 'test' functions. We say that a scheme securely implements a schema against a test family in the real world if for every test in the family, if test is able to hide some bit of information from all adversaries in the ideal world, then this bit should be hidden in the real world too. By choosing test families appropriately, we are able to place known security definitions (along with new ones) for a given object on the same canvas, enabling comparative analysis. Next, we explore the implications of a meaningful relaxation of our security definition, the one obtained by considering all-powerful adversaries in the ideal world. Thanks to our framework, we are not only able to substantially generalize known results connecting two important flavors of security definitions (simulation and indistinguishability) in cryptography under this relaxation, but significantly simplify them too. We also initiate a systematic study of the security of fundamental cryptographic primitives like public-key encryption under a new class of attacks that had not been considered so far in the literature. Once again, owing to the flexibility of our framework, we are able to model such attacks, along with existing ones, in a clean and satisfactory way.
Issue Date:2015-12-01
Type:Thesis
URI:http://hdl.handle.net/2142/89129
Rights Information:Copyright 2015 Shashank Agrawal
Date Available in IDEALS:2016-03-02
Date Deposited:2015-12


This item appears in the following Collection(s)

Item Statistics