System Design for ML based Detection of Unauthorized UAV and Integration within the UTM Framework

Unauthorized UAVs present a major concern for public safety, and their detection poses challenges in terms of technical solution and integration with existing aviation regulatory systems. This paper presents an effective detection system setup in this regard and its integration with the UTM system for monitoring the airspace for UAVs. The detection mechanism relies on measuring received RF power at the detector units emanating from the UAV and using machine learning techniques on these measured powers to estimate the location of the unauthorized UAV. The proposed system has been tested for both location estimation and tracking of the flight trajectory of the UAV. With a three-detector system, the R-squared values for detection achieve 0.99 in both LOS and NLOS propagation environments while the accuracy performance in distance yields percentage error of $(4.13 \%, 4.62 \%)$ and $(1.14 \%$ and $1.59 \%)$ in (line-of-sight (LOS) and non-LOS (NLOS)) environments for three and four detector system, respectively. The detector system includes an auxiliary visual detection system based on 120° sectored camera system for supporting law enforcement steps after the successful detection stage. Lab measurements for estimating the distance of a test object for visual detection has been performed and reported in this paper. Additionally, the neutralization of such UAVs by the proposed detector units has been presented. The integration of the proposed detection system with an existing UTM system has been discussed in detail in this paper.