Research

Our research agenda focuses on secure design automation methods to prevent information leakage in interconnected computing devices within the realm of:

  1. the Internet of Things,
  2. bio-medical devices, and
  3. our communication infrastructure.

Information surrounds us all – it is in everything from our DNA, to our health, travel, and spending habits, to our personal interactions with others. We transmit this information to other parties on a continuous basis, and that frequency will continue to increase with our interconnected computing infrastructure. The question becomes, how do we keep our information safe? Our research focuses on mitigating information leakage within hardware devices – specifically, developing automated methods to mask information-rich data during hardware processing. This work is already of interest to DOD agencies, government labs, DHS agencies as well as NSF, NIH and industry sponsors.

Our current research focuses on manipulation, anomaly detection, and security of data within hardware systems, forming two main tenets of our research agenda:

  1. Information Discovery: Enable the discovery of information in complex data sets (e.g. biological sequences, financial and travel records, social media, astronomical audio/visual spectrum, etc.) using manipulation and anomaly detection algorithms derived from the emergent patterns of the data sets creation; and
  2. Information Masking: Secure information by manipulating data sets (e.g. biological sequences, swarm interaction, hardware devices, etc.) by focusing on and eliminating the relationship between a data source and its creation and instantiation.

Research Philosophy

To have a sustained, active, and impactful research our agenda contains a balance of broad crosscutting ideas (e.g. information theory, structures & function, distributed coordination) as well as depth in several differentiating, areas of interest as defined by major science and health funding agencies (e.g medicine, cyber-security, big data, multi-agent systems). Our prior research has been funded through a variety of source including the National Science Foundation, the National Institutes of Health, the Intel Corporation, Cincinnati Children’s Hospital Medical Center, the National Radio Astronomy Observatory, the Wyoming Department of Education, the University of Cincinnati and the University of Wyoming. We have contributed to most aspects of the funding cycle, including: proposal writing, content development, R&D, evaluation, coordination and agency reporting. While our own personal identity as researchers is a critical component of our future success – we believe just as much in developing partnerships with others – partnerships within the department, college, university, K12 community, industry, and also with government agencies. These collaborations strengthen all partners and enable sustained, high-valued growth and meaningful societal impacts. In creating and developing partnership, it ultimately promotes the abilities of both the research and the department across the field of partners – essentially leading by example. The balance of funding and partnerships enables both personal growth and over time sustains a pipeline of K12 students into the ECE program who then become high-valued assets in industry and research.

Description of Current Research Interests

Our research efforts focus on three interrelated areas of manipulation, anomaly detection, and security of data in hardware systems. Below are three high-level examples of our research.

Secure System Design and Automation using Graph & Automata Theory

Today’s electronics are defined, designed and implemented using automated synthesis mechanisms. They are designed as black boxes but implemented in physical world – and thus bound by the rules of physics. For decades, hardware and software security was a function of cryptographic strength, limiting exposure of brute-force attacks, the end-user and the cost ratio between attack and protection. Protection has historically been a costly endeavor since generic automation techniques do not exist, and specific automation techniques that do exist perform expensive low-level circuit modification. Our previous research, using automata theory and understanding the relationship between hardware attacks and the circuit-level physics has enabled the development of high-level automation algorithms that can be applied at any level of the design process. Our funded research in this area has produced half a dozen publications and contributed directly to at least 7 masters thesis.

Organization & Anomaly Detection in Agent Swarms

The future research of drones and autonomous vehicles is not in the behavior of the individual – but rather the behavior of the collective as a function of the behavior of the individual. The emergent behavior of agent collectives (e.g. populations, organisms, cells) is a well studied phenomenon in biology – ”why do bird’s flock?” – ”why do fish school?” – ”why do zebras have stripes?” Biomorphic systems re-implement the ”rules of nature,” this research focuses on developing proof about system behavior from the rules of agents. For example, agents given two simple rules for approaching a target and moving away from their nearest neighbor a swarm will form a circle in two-dimensional space, or a sphere in three-dimensional space. Having a proof system for such problems is critical in guaranteeing the provable security and safety of autonomous swarms. Early stages of this work have been used in various K12 teacher outreach programs at the University of Wyoming.

Compression, Encryption and Parallelized Feature Detection

Traditional big data research generally ignores that data has some relationship to prior events, state (hysteresis) or structures – parallelization that occurs can not take advantage of this additional information. Understanding the history and creation of a data source enables the design of more efficient parsing, tokenization, and parallelized processing of a data source, while also enabling the correct recovery of higher-level information. Furthermore, some data streams are meant to be private but still accessible when needed – in their raw form they are an impediment to research and privacy. Genomic sequences highlight these issues – they are processed in raw, uncompressed, un-encrypted form, and any parallelizations occurs across multiple sequences. The foundations of our work in this area back a web-based bio-informatics tool from Cincinnati Children’s Hospital Medical Center that was recently published in Current Plant Biology