Description of original award (Fiscal Year 2018, $599,121)
This application was submitted as a New Investigator/Early Career proposal. The goal of this project is to investigate a novel passive system with low-cost and high-efficacy, which is capable of identifying, localizing, and tracking contraband wireless devices as long as they transmit radio signals.
The study proposes to investigate advanced machine learning methods and mathematical models with computer simulations to exploit radio frequency (RF) signal characteristics that are unique to devices, called RF device fingerprints, and locations, called RF location fingerprints.
The proposal will develop RF device fingerprinting techniques; including radio signal preamble detection, multidimensional analysis, a deep complex-valued neural network, and zero-shot learning algorithms, for unauthorized device identification. To localize the device, we propose to develop RF location fingerprinting techniques via RF channel modeling, computer simulation, and learning.
The applicant proposes to design a robust multi-device tracking algorithm to associate identification and detection results in a probabilistic manner, and visualize the results in a user-friendly graphical user interface (GUI). This proposed project presents a feasible solution, including both hardware and software, to combat illegal wireless device use by inmates in prisons. The hardware is based on commercial software-defined-radio (SDR) technologies, so that it has a low cost and high reliability, and the software contains implementations of all methods, algorithms, and the GUI as described. Extensive experiments are proposed to be conducted in a laboratory environment to validate the effectiveness of the developed system.
"Note: This project contains a research and/or development component, as defined in applicable law," and complies with Part 200 Uniform Requirements - 2 CFR 200.210(a)(14).