| 摘要: | 隨著無人機（Unmanned Aerial Vehicle, UAV）技術日益普及，
其應用範疇已從傳統軍事偵察拓展至交通監控、農業巡檢與災害
搜救等民用場景。然而，面對日益嚴重的非法無人機入侵事件，
需要發展具即時性與高準確性的多目標 UAV 追蹤與定位系統，以
確保空域安全與公共管理之需求。
本文提出一套針對多目標 UAV 動態追蹤所設計的定位系統，
架構基於 RF Switch 分時接收之 L 型天線陣列，透過單路 RF/ADC
分別進行仰角與方位角之 MUSIC 角度估計，以同時兼顧推估精準
度與硬體資源節省。為解決分時量測所導致的角度配對不確定性，
本文引入最大概似配對法（Maximum Likelihood Pairing, MLP），建
立仰角與方位角的正確對應；此外，進一步整合聯合機率資料關
聯濾波器（Joint Probabilistic Data Association Filter, JPDA），提升跨
時間多目標角度關聯的穩定性，有效避免因軌跡交錯或角度接近
所致之識別錯誤。
在模擬評估部分，本文針對不同天線數與訊雜比條件下，進
i
行二至五架 UAV 的多目標追蹤實驗，並分析角度估測誤差、DOA
偵測率與配對成功率等性能指標。實作方面，本文以軟體定義無
線電（Software Defined Radio, SDR）平台建構完整之訊號處理流
程，並於真實場域中完成雙 UAV 實驗。實驗同時結合雙基站間之
三角觀測法，成功還原 UAV 在三維空間中的座標位置。
整體而言，本文成功設計並實現一套整合 RF-Switch 分時接
收、MUSIC 估計、MLP 配對與 JPDA 多目標追蹤之三維定位系
統，具備良好之擴充性、即時性與多目標辨識能力，展現其於智
慧監控與非法 UAV 偵測場域中之實用潛力。
;With the rapid proliferation of unmanned aerial vehicle (UAV) tech nology, its applications have extended beyond traditional military recon naissance to include civil domains such as traffic monitoring, agricultural
inspection, and disaster search and rescue. However, the increasing in cidence of unauthorized UAV intrusions poses a significant challenge to
airspace safety and public security, necessitating the development of real time and high-accuracy multiple UAV tracking and localization systems.
This paper proposes a multiple UAV tracking and localization sys tem based on an L-shaped antenna array architecture with time-division
reception via an RF switch. Using a single RF/ADC hardware path, high resolution direction-of-arrival (DOA) estimation for both elevation and
azimuth angles is performed independently via the MUSIC algorithm,
enabling precise angular estimation with reduced hardware complexity.
To resolve the ambiguity introduced by time-division measurements, a
maximum likelihood pairing (MLP) method is adopted to associate ele vation and azimuth angles accurately. Furthermore, a Joint Probabilistic
Data Association Filter (JPDA) is integrated to enhance data association
across time frames, improving robustness in scenarios involving trajec tory crossover or closely spaced targets.
In simulation, the system is evaluated under varying antenna array
sizes (M) and signal-to-noise ratio (SNR) conditions, with experiments
involving two to five UAVs. Key performance metrics, including angular
estimation error, DOA detection rate, and pairing success rate, are ana lyzed. In practical implementation, a complete tracking system is built on
a Software Defined Radio (SDR) platform, incorporating signal acquisi tion, angle estimation, pairing, and tracking. Field experiments involving
dual UAVs are conducted, and 3D localization is achieved by applying
triangulation across dual base stations.
Overall, this work presents a fully integrated 3D UAV localization
system combining RF-switch-based time-division reception, MUSIC es timation, MLP pairing, and JPDA multiple tracking. The proposed sys tem demonstrates strong scalability, real-time capability, and robustness
in complex multiple scenarios, showcasing its potential for applications
in intelligent surveillance and unauthorized UAV detection. |
