Authors
Tianyi Zhao and Danijela Cabric
Abstract
Sensing and understanding the RF activities in a spectrum is important to ensure wireless security, enforce RF compliance and facilitate dynamic spectrum sharing. To enable such intelligent spectrum awareness, we present a complex-valued deep learning framework for simultaneous signal detection and transmitter fingerprinting. We evaluate the framework using an OTA dataset and show that compared to a real-valued network, the complex-valued framework is able to achieve better performance in both detection and fingerprinting tasks and also gain better generalization capability.