Date of Award

Fall 2022

Embargo Period


Access Type

Dissertation - Open Access

Degree Name

Doctor of Philosophy in Electrical Engineering & Computer Science


Electrical Engineering and Computer Science

Committee Chair

Eduardo Rojas


The design and application of novel physical-layer security techniques have been increasing in the last decades as means to enhance the security that more traditional techniques provide to wireless communications systems. Well-known hardware security techniques, such as radio frequency fingerprinting, use unintended manufacturing process variations and unique hardware structures in the semiconductors for applications such as identification and classification of the source of different transmitted signals, and detection of hardware modifications. The uniqueness of the features that two different modules present, even maintaining the same design, can be used for modules characterization at a lower cost and complexity than other more traditional techniques, such as reverse engineering, require to perform these analyses. The distinctiveness of the hardware is limited by the margin between considered features and the number of components of the system. Considering these limitations, there is an increasing interest in features that can be exploited to preserve the margins between them in large-scale applications and that can be easily updated. The advantages that antenna features and additive manufacturing present to enhance the aforementioned characteristics are promising assets to consider for the security of wireless communications systems.

This dissertation is focused on the use of antenna features to exploit the radio frequency fingerprints of wireless modules and systems in which they are included. It presents the use of directly measured antenna parameters and impacted parameters of their transmitted signals to exploit the difference in the fingerprints of the modules and use them for source classification and hardware modifications detection. The fundamentally novel concept of engineering unique features to impact the fingerprint for each antenna produced, by using additive manufacturing techniques as part of the process, is analyzed and presented. The engineered fingerprints are based on intentional features (designed differences) added to the geometry of the designed antennas, enhancing the performance of RF fingerprinting systems used in this case for physical-layer security classification applications. The variability and complexity of the designs is improved by considering the additive manufacturing component. Considering the margin between features from different units, machine learning-based algorithms are used to exploit the information provided for classification, identification, or detection applications, which enables detailed analyses of new modules in a timely manner.

The concepts presented in this dissertation are implemented and evaluated using simulations and hardware analyses by designing, manufacturing and testing the modules, antennas, and systems that can take advantage of the described features. The techniques present promising characteristics to improve the physical layer security performance limitations at a low cost and time impact.