Skip to content

Best Paper Award Won at Cybersecurity Conference

Leslie Lestinsky

Department of Electrical Engineering (NDEE) graduate student Joseph Loof, advised by NDEE professor and an affiliated member of the Wireless Institute, Thomas Pratt, recently traveled to Denver, Colorado, for Resilience Week 2018. There he won an award for the Best Cyber R&D Paper. In his paper, Loof addresses the problem of frequency hopped signal classification in a propagation environment with multi-path. He explains his approach towards signal classification–using unconventional methods that, instead of seeing multi-path as a hinderance, leverage it to enable signal discrimination.

Resilience Week is a four-day conference in which universities, departments of government and major industry players come together to share research and best practices that will better prepare communities to respond to and recover from attacks on critical infrastructure systems. Examples of this are, re-routing airplanes during mid-air attacks, mitigating flood damage as it is encroaching and in the case of Loof’s research, protecting signal networks and data transfers.

His paper, “Unsupervised Classification of Frequency Hopped Signals in Frequency-Selective Channels,” explains how signal features in the polarization-frequency plane are leveraged. “We take advantage of polarization properties, which are a function of frequency, specific to each source and environment,” explained Loof. Each source and environment provide a unique polarization-frequency signature. These signatures can be used to group received signals according to which transmitter sent the information.

Government and commercial entities might apply this practice for signal authentication and to protect encryption from man-in-the-middle attacks. “The method provides a physical layer of security by discriminating communications nodes that are part of a user’s fixed network versus communications nodes that are pretending to be part of the user’s network,” explained Pratt. Such attacks can be detected and prevented based on the measured radio frequency signal properties on a packet-by-packet basis. The approach is also tolerant to changes in waveforms, meaning that the system can still classify or authenticate signals even if the transceiver uses different waveforms. “We can put a receiver system in a field, and it will essentially monitor signals and ascertain whether or not those signals are coming from a common source(s),” explained Loof. The benefits of this approach may be contrasted with statistical methods that rely on the detection of anomalous behavior, well after an attack has begun. “It’s a novel approach,” explained Pratt. “I’m proud of the work he’s doing. It is very diverse and has multiple applications.”