[Busec] Omer Paneth PhD defense! (Wed 10am)

Sharon Goldberg goldbe at cs.bu.edu
Mon Sep 26 20:13:32 EDT 2016


PhD Thesis Defense

*Program Obfuscation: Outside the Black Box*

*Omer Paneth*

*Wednesday, September 28, 10am.*

*Hariri Institute for Computing Seminar room MCS 180, 111 Cummington Mall*

*Abstract:*

Code is said to be obfuscated if it is intentionally difficult for humans
to understand. Obfuscation is often used to conceal sensitive
implementation details such as proprietary algorithms or licensing
mechanisms.

A general-purpose obfuscator is a compiler that obfuscates arbitrary code
(in some particular language) without altering the code's functionality.
Ideally, the obfuscated code would hide any information about the original
code that cannot be obtained by simply executing it.

The potential applications of general-purpose obfuscators extend beyond
software protection. For example, in computational complexity theory,
obfuscation is used to establish the intractability of a range of
computational problems. Obfuscation is also a powerful tool in
cryptography, enabling a variety of advanced applications.

The possibility of general-purpose obfuscation was put into question by
Barak et al. [CRYPTO 01], who proved that such obfuscation cannot have
ideal security. Nevertheless, they leave open the possibility of
obfuscation with weaker security properties, which may be sufficient for
many applications. Recently, Garg et al. [FOCS 13] suggested a candidate
construction for general-purpose obfuscation conjectured to satisfy these
security properties.

In this thesis we study the feasibility and applicability of different
notions of secure obfuscation.

In terms of applicability, we prove that finding a Nash equilibrium of a
game is intractable, based on a weak notion of obfuscation known as
indistinguishability obfuscation.

In terms of feasibility, we focus on a variant of the Garg at el.
obfuscator that is based on a recent construction of cryptographic
multilinear maps [Garg et al. EUROCRYPT 13]. We reduce the security of the
obfuscator to that of the underlying multilinear maps. Our first reduction
considers obfuscation and multilinear maps with ideal security.

We then study a useful strengthening of indistinguishability obfuscation
known as virtual-grey-box obfuscation. We identify security properties of
multilinear maps that are necessary and sufficient for this notion.

Finally, we explore the possibility of basing obfuscation on weaker
primitives. We show that obfuscation is impossible even based on ideal
random oracles.

*Examining Committee:*

First reader: Prof. Ran Canetti
Second reader: Prof. Shafi Goldwasser (MIT)
Third reader: Prof. Leonid Reyzin

Committee chair: Prof. Steve Homer

Additional committee member: Prof. Vinod Vaikuntanathan (MIT)









-- 
Sharon Goldberg
Computer Science, Boston University
http://www.cs.bu.edu/~goldbe
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs-mailman.bu.edu/pipermail/busec/attachments/20160926/d6afebfd/attachment.html>


More information about the Busec mailing list