RF-Based UAV Classification and Identification Using Machine Learning

In today's market, the position of Unmanned Aerial Vehicles (UAVs) are considered ubiquitous. They are used for hobby recreation and are also customized to meet clients' unique industrial needs, including in fields like the construction and agriculture sectors. Nonetheless, UAVs' involvement in criminal and illegal activities is also increasing. Hence, effective detection and identification of UAVs have become an essential field in research. In order to effectively recognize and identify UAVs, a new machine learning-based technique is proposed in this study. To be exact, radio frequency (RF) data pretreatment and feature extraction phases are described together with an enhanced approach for UAV identification and detection utilizing ensemble learning that utilizes hierarchical notions. The RF signals are filtered to increase the efficiency of the output. Raw signals from numerous UAVs in various operating states, spanning the modes of on, off, video taking, flying, connected, and hovering, were comprehensively researched using publicly accessible datasets. Additionally, we developed an algorithmic solution to find and identify intrusive drones using RF datasets. The algorithm has been trained to detect if a UAV is in range, immediately identify the UAV type, and accurately determine the mode of flight of said discovered UAV.