Research Project III: Data-efficient and adaptive machine learning methods for next-generation wireless spectrum systems

Part 1: One major challenge of deep learning-based RF fingerprinting is that wireless signals are highly sensitive to environmental conditions, causing the device fingerprints captured in one environment to not transfer well to another. Hence, deep learning models are found to perform well in the same condition but lose their ability to classify devices in the new condition. In this project, we examine transfer learning techniques to mitigate the domain shift problem in RF fingerprinting and compare them with two well-defined baselines. The three RF fingerprinting datasets under various scenarios are examined to explore how environmental factors impact RF fingerprinting, such as transmitter locations, transmitter distance, and device configurations.

Part 2: We study open-set domain adaptation for RF fingerprinting with anchor devices, a realistic deployment setting in which a small number of controlled transmitters provide strong but limited target-domain supervision. The major challenge is to leverage this partial supervision to correct domain shift while preserving discriminability for unlabeled known devices and maintaining robustness to unknown devices. This problem generalizes existing RFFI settings and reflects the operational constraints of real wireless systems.