Enhancing Spectrum Busting: A Federated Learning Approach for Radio Fingerprinting for Defence Forces

Each transmitter has a unique transmission pattern due to semiconductor variations. Using Software Defined Radio (SDR) and Artificial Intelligence (AI), transmitters can be identified through their Radio Frequency patterns. With the growing number of transmitters, ensuring security, availability, and data privacy is crucial for military and civilian applications. This paper introduces a novel hybrid method for radio fingerprinting via RF classification and proposes a Federated Learning (FL) model, achieving 86% accuracy in the client model and 89% in the universal model. The solution addresses spectrum sensing, signal classification, spoofing countermeasures and ensures security, low latency, and resource optimisation.