24GHz車用雷達多目標物偵測演算法設計

Abstract

本篇論文提出了一個應用於24GHz車用防撞警示雷達(1T2R架構)的多目標物偵測演算法。為了達到較低的取樣頻率，較低的運算時間，吾人運用多斜率雷達波與自製的多目標物偵測演算法可使24GHz車用防撞警示雷達(1T2R架構)同時偵測多個目標物。根據1T2R的天線架構，新提出的演算法可以同時判斷多目標信號的角度、距離與速度。由於此演算法對於DSP硬體規格需求較2D-FFT演算法更低，經過DSP效能計算後，能使昂貴的車用防撞警示雷達能用較低廉的MCU架構來完成車用防撞警示雷達訊號處理的部分。

最後，用模擬的方式驗證新演算法的可行性。模擬環境考慮許多不同的情形以貼近現實真正的情況。同時，目標物的特性：距離、速度還有角度的資訊會被結合在一起，進一步準確地掌握目標物的動態資訊。

In this thesis presents a multi-target detection algorithm for 24GHz vehicle collision warning radar (1T2R architecture). In order to achieve the lower sampling frequency and the lower computation time.We take use of multi slope radar waveform for multi-target detection algorithm. It archive to detection multiple target simultaneously for 24GHz vehicle collision warning radar (1T2R architecture).according to our proposed algorithm we could simultaneously determine the multi target's angle, distance and speed with the 1T2R antenna architecture. The requirements of the hardware specification of the proposed algorithm is lower than the 2D FFT algorithm, by the DSP processing performance calculation, the MCU (ARM-M3 MCU) architecture could achieve the digital signal processing parts of vehicle collision warning radar. Using the MCU chipset to achieve the expensive DSP chip, it cost down the hardware cost of the vehicle collision warning radar.

Finally, by the simulation results will be illustrated to verify the feasibility of the algorithm. Many situations are taken into consideration to ensure the simulation environment close to the real case. At the same time, targets’ distance, speed, angle information and targets’ characteristics will be combined together for further tracking and detection.