IISP Cybersecurity Fellows

Three Ph.D. Fellows Receive Funding Support for Spring '17

The Cybersecurity Fellowship Program supports unfunded and under-funded, emerging research so that outstanding Ph.D. students may do what they do best – take exceptional ideas from concept to proof. The Fellowship serves to motivate students with an advanced understanding of information assurance and cyberthreats to pursue innovative research without undue concern for lack of funds.

Learn more about the program or apply here: http://www.iisp.gatech.edu/cybersecurity-fellowship-program


Marie Le Pichon

Advised by Annie Antón, School of Computer Science

"Privacy and Security Requirements in Data Science Studies"

My current research addresses the privacy, security, and compliance challenges faced by researchers working on data science projects. Due to the emergent properties of big data, researchers regularly re-evaluate and modify their goals. These changes must be reflected in the project’s governing documents. These documents must be consistent, must cater to diverse and sometimes conflicting stakeholder needs, must be compliant in a complex regulatory landscape, and must ensure the privacy and security of research participants.

The goal of my research is to explore whether requirements engineering can be leveraged as a potential solution to these challenges. To this end, I have been working on the CampusLife project, a Georgia Tech study which aims to leverage data science to improve student well-being. This project provides an important first step for my dissertation research in which I am seeking to develop methods and tools that ensure that privacy and security requirements are adequately addressed in various types of systems. The CampusLife project is foundational in its role as a formative case study that will allow me to develop methods and tools that I plan to validate and evolve throughout a series of subsequent case studies.


Stacey Truex

Advised by Ling Liu, School of Computer Science

"Privacy-Preserving Ensemble Learning"

This semester I am focusing on the problem of privacy-preserving ensemble learning with two main objectives: (1) articulate the current state of the art, how researchers have been approaching the privacy-preserving ensemble learning domain, and different approaches that can create effective, usable solutions and (2) create a concrete solution for the problem of privacy-preserving distributed random forest evaluation. When the semester began, I had begun discussions with my advisor, Dr. Ling Liu, on both objectives and had sketched a scope and perspective for (1) as well as a preliminary plan of attack for (2). By the end of the Spring ’17 semester, I plan to have a completed survey paper covering common machine learning practices for ensemble learning, frequently used approaches in the domain of privacy-preserving machine learning, as well as a perspective on the limitations of these approaches. I also plan to have, in collaboration with other researchers in the College of Computing, a completed proposal for privately evaluating a distributed random forest model. This will include machine learning results, implementation results regarding protocol efficiency, and thorough security proofs for the proposed protocol.


Taimour Wehbe

Advised by Vincent Mooney, School of Electrical & Computer Engineering

“Physiological Features-Assisted Architecture for Rapid Detection of Hardware Trojans in Medical Devices”

My current research focuses on rapidly detecting malicious alterations and hardware attacks that compromise medical devices. To the best of our knowledge, our project is the first in literature to utilize physiological relationships between Electrocardiographic (ECG) and Ballistocardiographic (BCG) heart measurements to detect hardware attacks or errors in medical devices and differentiate the attacks/errors from health problems.

In our early research, we studied different physiological relationships between ECG and BCG signals. We then designed the necessary hardware architecture to extract these features at run-time. The feature extraction circuitry is used to assist in detecting and distinguishing hardware attacks from health problems in medical chips. Our preliminary simulations show promising results, and we recently had a paper accepted in one of the top IEEE hardware security conferences (Hardware Oriented Security and Trust). I will be presenting this work at the end of the Spring ’17 semester. In the meantime, I will be looking for more robust techniques to incorporate physiological features with traditional security primitives. Specifically, we are studying different ways of integrating physiological relationships into elliptic curve cryptography.