Professor, John E. Kasch Endowed Faculty Fellow
Department of Electrical and Computer Engineering
The University of Texas at Austin
2501 Speedway, EER 4.814, Austin, TX, 78712
(512) 471-1082, orshansky@utexas.edu
Research
My recent research centers on hardware security (roots of trust based on physical unclonable functions, side-channel attacks and countermeasures, realizations of post-quantum cryptography), ML algorithm and hardware codesign, and approximate computing.
Teaching
Advising
News
- A post-doctoral position starting in Fall 2024 is available for a project involving applied ML, modeling, and optimization. Please contact me if interested.
Select publications
-
Variability-aware training and self-tuning of highly quantized DNNs for analog PIM
Z. Deng and M. Orshansky. DATE 2022.
-
Power-based Attacks on Spatial DNN Accelerators
G. Li, M. Tiwari, and M. Orshansky. JETC 2022.
-
Horizontal side-channel vulnerabilities of post-quantum key exchange and encapsulation protocols
F. Aydin, A. Aysu, M. Tiwari, A Gerstlauer, M Orshansky. ACM Transactions on Embedded Computing Systems 2021.
Top Pick in Hardware and Embedded Security 2021
-
Lattice PUF: A strong physical unclonable function provably secure against machine learning attacks
Y. Wang, X Xi, M Orshansky. HOST 2020.
-
Training with multi-layer embeddings for model reduction
B. Ghaemmaghami, Z. Deng, B. Cho, L. Orshansky, A. Singh, M. Erez, M. Orshansky. arXiv 2020.
-
A strong subthreshold current array PUF resilient to machine learning attacks
H. Zhuang, X. Xi, N. Sun, M. Orshansky. TCAS 2019.
-
Securing AES against localized EM attacks through spatial randomization of dataflow
G. Li, V. Iyer, M Orshansky. HOST 2019.
-
Using power-anomalies to counter evasive micro-architectural attacks in embedded systems
S. Wei, A. Aysu, M. Orshansky, A. Gerstlauer, M. Tiwari. HOST 2019. Best Paper Award Nomination
All publications
Book
My book "Design for Manufacturability and Statistical Design: A Constructive Approach" is available on Amazon. The book provides a thorough treatment of the causes of variability, methods for statistical data characterization, and techniques for modeling, analysis, and optimization of integrated circuits to improve yield.