Abstract
Counterfeits in commercial-off-the-shelf (COTS) devices pose significant security risks to supply chains, emphasizing the need for robust authentication methods. As the Internet of Things (IoT) continues to expand across various domains, such as autonomous vehicles (AVs), smart homes, and industrial automation, IoT devices have become ubiquitous. Many of these devices are equipped with wireless transceivers, enabling the use of radio frequency fingerprinting (RFF) for device authentication. One of the common RFF techniques is to extract unique features from in-phase and quadrature (I/Q) imbalance measurements. However, existing RFF techniques often rely on the use of a common test receiver, which limits their applicability in real world deployments. In this work, we propose a novel RFF technique independent of a common test receiver that leverages the crosstalk effect in the transceiver. An individual I/Q imbalance measurement is converted into an abstract data point that can be a vector value or, more generally, a probability density function (PDF). Our approach employs a statistical modeling framework that is lightweight and computationally efficient, eliminating the need for large labeled datasets or frequent model retraining. This makes the method well suited to real time implementation on platforms with limited computation and memory, since constellation gating, I/Q imbalance estimation, and PDF comparison can be implemented with simple streaming updates and a single distance check per recording. To examine the feasibility of the proposed method, we conducted extensive experiments using software-defined radios (SDRs) in different environments. Statistical analysis of the extracted fingerprints reveals that for a given device, the maximum measurement deviation (MMD) between a measured data point and the center of multiple measured points remains consistently smaller than the minimum inter-device distance, suggesting that it is feasible to generate stable and discriminative RFFs to authenticate COTS devices with transceivers in the supply chain.