The detection of a human being behind obstacles using radar technology has many promising applications, such as post-disaster search and rescue missions. In the case of large observation areas, the use of an unmanned aerial vehicle as the moving platform for radar is an appealing approach. However, this task is challenging because human vital signs provide a relatively weak signal compared with background noise and clutter. The drone's instability also negatively affects the obtained radar signal. Thus, this paper presents a new signal processing method to extract and enhance respiration signals from drone-mounted radar. The method works by extracting the analytic signal representation and applying the subspace component segregation. Laboratory experiments in a controlled environment show that the proposed method can suppress a considerable level of vibration generated by unbalanced motor motion and enhance the respiration signal from radar data. A field experimental study using an octocopter in hovering mode confirms the method performs well in real-world conditions. Index Terms-vital sign detection, respiration detection, through-the-wall radar, ultra-wideband impulse radar, drone, unmanned aerial vehicle I. INTRODUCTION atural disasters occur frequently and can result in both material and non-material losses [1,2]. In catastrophic cases, large areas are affected with a huge number of victims. Post-disaster search and rescue (SAR) missions play a vital role in evacuating many live survivors quickly, thereby reducing the potential death toll. Advanced technology is necessary to support this mission [3,4].Notably, the use of unmanned aerial vehicles or drones for SAR missions in post-disaster environments is of interest to