| 摘要: | 近年來無人機(UnmannedAerial Vehicle, UAV) 的應用方式非常多元，然而隨著無人機應用的增加，對空域未經許可或未遵守相關法規，進入他人特定區域的空域，可能對安全、隱私和公共秩序遭成威脅。為了防治非法無人機入侵，對於無人機的偵測是一個重要的研究議題。我們採用了電子訊號監測到達角度(Direction of Arrival, DOA)的方法來估測無人機訊號源，我們考慮了單脈衝雷達演算法(Monopulse)一種廣為人知且技術成熟的雷達技術，該演算法能夠快速提供精準的目標物估測角度，非常適合即時追蹤的應用。而現今多天線波束成型技術(Beamforming)成熟，只需控制不同天線的相位和振幅大小也可以實現傳統單脈衝的能力。
對於到達角的估測，除了直接估測角度之外，自適應演算法也是不可或缺的一環，我們採用無跡卡爾曼濾波器(Unscented Kalman Filter, UKF)，對於移動的無人機能有效的降低估測誤差，使追蹤路線更平滑。考慮於多天線波束成型技術的天線不需轉動且估測更快速，但還是有缺點，例如雷達視野(Fieldofview,FOV)有限制，我們將比對固定天線追蹤和天線轉動的版本，分析其不同場景追蹤的優劣好壞，如無人機移動速、不同的運動軌跡，且考慮硬體成本的限制，如分時使用兩個線性陣列估測方位角和仰角，與使用均勻平面陣列一次估測兩個角度，分析其對於無人機的追蹤效能。;In recent years,the applications of Unmanned AerialVehicles(UAVs) have become highly diverse. However,with the increasing use of UAVs,unauthorized or non-compliant entry into specific airspaces poses threats to safety, privacy, and public order. To prevent illegal drone intrusions,drone detection has become a critical research topic. We employed the Direction of Arrival (DOA) method using electronic signal monitoring to estimate drone signals. We considered the monopulse radar algorithm, a well-known and mature radar technology that quickly provides accurate target angle estimations, making it highly suitable for real-time tracking.With the advancement of multi-antenna beamforming technology,traditional monopulse radar capabilities can now be achieved by controlling the phase and amplitude of different antennas. For DOA estimation, adaptive algorithms are also indispensable. We adopted the Unscented Kalman Filter (UKF), which effectively reduces estimation errors and smooths tracking paths for moving drones.
While multi-antenna beamforming allows for faster and more accurate estimations without moving antennas, it has
limitations such as a restricted field of view (FOV). We compared fixed-antenna tracking and rotating-antenna tracking versions, analyzing their advantages and disad
vantages in different scenarios, such as varying drone speeds and movement trajectories. Additionally, we considered hardware cost constraints, such as using two linear arrays in time-division mode to estimate azimuth and elevation angles versus using a uniform planar array to estimate both angles simultaneously, and analyzed their tracking performance for drones. |
