"Program Obfuscators are algorithms that take as input a program P and output another program that has the same input-output functionality, and at the same time is ""hard to understand"". Ideally, seeing the obfuscated program in full should not give any insight or ability other than the ability to simply run it ""as a black box"" on inputs of one's choice. Program obfuscation is a hard task. Indeed, practical experience shows that most, if not all ah-hoc attempts at obfuscating programs eventually fail. Furthermore, theoretical results rule out the existence of algorithms that can obfuscate any program in the above ideal sense. Still, recent results demonstrate that provably secure obfuscators do exist for some simple classes of programs. The proposed research aims to further our understanding of this intriguing concept and its usefulness in cryptography. Some of the proposed directions are: - Expanding the boundaries of what's provably obfuscatable. As first steps, we propose new methodologies for obfuscating richer classes of programs, along with some exciting applications. - Formulating and achieving stronger notions of obfuscation than the currently considered ones. The currently considered notions of obfuscation do not suffice for some prominent applications. The unaddressed concerns include malleability of obfuscated programs, composability, and security for low-entropy inputs. - Finding ways to circumvent the known impossibility results for general-purpose obfuscators. Two directions look promising here: One is to relax the current formulations in a way that preserves the essence of the notion. Alternatively, analyze obfuscators in an abstract models that represent limited adversaries. - Finding new applications of the program obfuscation in other areas in cryptography."
Field of science
- /natural sciences/computer and information sciences/computer security/cryptography
Call for proposal
See other projects for this call