Dissertation Defense - William Unger

Security Analysis of Lightweight Cryptographic Primitives

• William Unger – Cybersecurity
• Zoom 

Abstract:

In this era of the Internet of Things (IoT), the massive connectivity of devices and enormous amounts of data in the cloud and on communication devices and channels have made information susceptible to different types of attacks. Achieving uncompromised security on devices and keeping sensitive data private requires cryptographic techniques. Cryptanalysis is the process of examining cryptographic systems for weaknesses by exploring the weaknesses of the underlying mathematics of cryptographic systems and weaknesses in their implementation, including side channel attacks. When conducting cryptanalysis, it is important to model the assumed amount of knowledge the adversary has. The attacks for these models are classified into two categories, white-box and black-box attack models. In the black box model the adversary has no initial knowledge of the cryptographic scheme or its implementation, but only can observe external information and behavior. In the white box model the adversary has full control over the implementation and its execution environment. In such a model, it is much more difficult to protect cryptographic implementations than in the classical black-box model.

In this dissertation we consider both types of attacks, the white-box model and the black-box model, and perform cryptanalysis on lightweight symmetric cryptographic schemes. Lightweight cryptographic primitives are designed for use in resource-constrained devices.

The first contribution of this dissertation is evaluation, for several lightweight cryptographic schemes, of the vulnerability of their S-boxes against Correlation Power Analysis (CPA). Three well-known theoretical metrics are evaluated: transparency order, non-linearity, and signal-to-noise ratio. These metrics aim to characterize the resistance of these S-boxes against adversaries that exploit physical leakages.

Furthermore, the work presented in this dissertation develops generalized quantitative metrics to measure the rate of information leakage and the accuracy of information learned through side-channel attacks. In doing so, we propose a novel implementation of ternary cryptographic schemes in firmware on a binary micro-controller and provide a comparative study of these against correlation power analysis.

Lastly, we perform a black-box security analysis of a proprietary protocol between two embedded systems that allow an encrypted tunnel between a client, and a server. Our goal is to evaluate the feasibility of reverse-engineering of the proprietary protocol by adversaries who have limited resources and only temporary access to the embedded systems. We propose a method that predicts several instances of the encryption process from the data collection/observations. In this work we also provide a structure for comparing how close our cryptanalysis results are to the true known values.

Committee:

Liljana Babinkostova, Mathematics  (Chair)

Robert Erbes, Idaho National Laboratory

Sin Ming Loo, Electrical Engineering

Marion Scheepers, Mathematics