JTEA: A Joint Trajectory Tracking and Estimation Approach for Low-Observable Micro-UAV Monitoring With 4-D Radar

Continuous trajectory tracking and movement estimation of microunmanned aerial vehicles (UAVs) in sensitive areas intended for unauthorized intrusion prevention are challenging due to their low-observable characteristics in urban low-altitude. Consequently, through a 4-D radar with range-Doppler-azimuth-elevation information, this article proposes a joint trajectory tracking and estimation approach (JTEA) toward unauthorized micro-UAV monitoring. First, for continuously realizing the 3-D trajectory tracking of micro-UAVs from clutters and noises, JTEA introduces an integrated detection and tracking strategy to directly accumulate the nonthresholding observations with a sequence of radar frames. The advantage is that it removes the threshold-decision process before tracking and thus avoids information loss in comparison to conventional approaches considering target detection and tracking independently. On this basis, JTEA presents a projection-based long short-term memory-connectionist temporal classification (PLSTM-CTC) network to directly predict movement class labels of micro-UAVs from in-progress motion trajectory without any presegmentation process. Remarkably, instead of directly inputting the tracking trajectory into the recognition network, PLSTM-CTC employs a 3-D projection layer to transform the position vectors into three orthogonal sequences for obtaining superior movement recognition performance of micro-UAVs with complex motions. Finally, simulation and experimental results show unique advantages of JTEA in micro-UAV tracking and movement recognition in contrast to conventional methods, especially under low signal-to-noise ratio (SNR) conditions.