- This event has passed.
Ph.D. Proposal: Fanghui Liu
December 11, 2020 @ 9:00 am - 11:00 am EST
Doctoral Dissertation Proposal
Title: Tolerant Algebraic Side-Channel Attack on Modern Ciphers using Constraint Programming
Ph.D. Candidate: Fanghui Liu
Major Advisor: Prof. Laurent Michel
Associate Advisors: Prof. Alexander Russell, Prof. Benjamin Fuller
Date/Time: Friday Dec. 11, 2020 9:00 am – 11:00 am
Location:
Meeting link: https://uconn-cmr.webex.com/uconn-cmr/j.php?MTID=m9498f4d13b7e89b02207cd48af44dbd5
Meeting number: 120 756 5533
Password: vA2dDPsfA66
Join by phone: +1-415-655-0002 US Toll
Access code: 120 756 5533
Abstract:
Low-complexity side channel attacks pose difficulties when measurement errors are present. In a single trace scenario, algebraic methods can be used to recover secret information (i.e. Cryptographic Keys) by reducing the target cipher to a satisfiability problem while constraining the intermediate states to the corresponding side channel information. This approach is not able to handle any noise as it requires that the information extracted from the target device be exact. Tolerant Algebraic Side Channel Attack (TASCA), introduced by Oren et. al., improves the resiliency of algebraic side channel attacks by modeling the errors present in the measured side channel information and reformulating the attack as a pseudo boolean optimization problem. Yet, its scalability is still quite limited. In this thesis, we propose several TASCA-CP models that are far more scalable, improving upon the state of the art by orders of magnitude in both space and time.
Specifically, the thesis considers multiple ciphers (AES-128, AES-256 and Keeloq) and offers models that are capable of recovering secret keys under the assumption of a noisy side-channel signal based on power consumption of the circuit implementing the cipher. The novel models take advantage of constraint programming over bit-vector domains and extend the collection of consistency algorithms to filter constructions arising in the cipher implementation. It also exploits custom search procedures that take advantage of the semantics of the cipher implementation. Empirical evidence demonstrates the effectiveness of the approach. Finally, we demonstrate that TASCA-CP can be extended to other modern ciphers such as Keeloq with minimal efforts.
