臺灣博碩士論文加值系統

近年來，FMCW雷達技術在我們生活中的重要性日益凸顯。原本應用於國防科技的毫米波通訊技術，如今已被廣泛應用於監控系統和先進駕駛輔助系統（ADAS）等領域。隨著科技的進步，雷達技術也在不斷更新。如何有效提升雷達系統的目標偵測與追蹤能力，使其在汽車自動駕駛、交通監控系統和無人機避障等領域實現更高的效能與安全，已成為現今研究的熱點之一。

本論文提出了一種FMCW雷達系統中的多目標偵測與追蹤技術，能在環境複雜且干擾強烈的情況下持續追蹤多個目標。透過處理由多個chirps組成的一組幀（Frame）後，可獲得當前時刻的目標資訊，長時間掃描獲得的多組幀，可通過卡爾曼濾波器（Kalman Filter）進行軌跡追蹤。一組幀包含數筆連續的雷達數據，稱為chirps，我們對其進行距離-快速傅立葉轉換（1D-FFT）後存儲下來。為解決環境和雷達本身的雜訊干擾，我們使用移動目標指示（MTI）濾波器來消除雜訊。隨後對濾波後的頻域數據進行都卜勒-快速傅立葉轉換（2D-FFT），形成涵蓋距離與速度資訊的二維頻譜圖。在目標偵測判斷中，我們針對多目標需求採用改進式二維SOCA-CFAR算法。在方位角量測技術中，選擇波束成形（Beamforming）作為主要方法。確認目標資訊後，利用最近鄰資料關聯（NNDA）區分多物件，並使用卡爾曼濾波實現多目標追蹤。

在實驗方面，我們使用德州儀器（TI）研發的AWR1443 BOOST雷達模組，採用1根發射天線（Tx）和4根接收天線（Rx）的SIMO（Single-Input Multiple-Output）系統進行實驗。其數據通過DCA1000 EVM擷取卡接收，並使用MATLAB驗證我們所提出的演算法。與TI開發的軟體mmWave Studio相比，我們的系統不僅有更精確的目標偵測效能，還實現了該軟體缺少的軌跡追蹤功能。

在硬體實做方面，本論文使用TSMC 90nm製程技術，針對計算複雜的ADC數據到2D-FFT部分進行硬體設計。最終晶片核心面積為1.95 mm2，Gate Count為164.4 k，最高工作頻率可達到163 MHz，系統平均功耗為12.5 mW，throughput為10.48 M Sample/s。經過等效換算，含有4根接收天線資料的一組幀能在1.56 ms完成計算。

In recent years, the importance of FMCW radar technology in our daily lives has become increasingly prominent. Originally applied in defense technology, millimeter-wave communication technology is now widely used in surveillance systems and advanced driver assistance systems (ADAS). As technology advances, radar technology is continuously evolving. How to effectively enhance target detection and tracking capabilities of radar systems to achieve higher performance and safety in fields such as automotive autonomous driving, traffic surveillance systems, and drone obstacle avoidance has become a hot research topic.

This thesis proposes a multi-target detection and tracking technology in FMCW radar systems capable of continuously tracking multiple targets in complex and heavily interfered environments. The current target information can be obtained by processing a frame consisting of multiple chirps, and multiple frames obtained through long-term scanning can be used for trajectory tracking via a Kalman filter. A frame contains several continuous radar data sets named chirps stored after performing a range-fast Fourier transform (1D-FFT). To address noise interference from the environment and the radar itself, we use a moving target indicator (MTI) filter to eliminate noise. Subsequently, the filtered frequency domain data undergo Doppler-fast Fourier transform (2D-FFT), forming a two-dimensional spectrum covering range and velocity information. For target detection and judgment, we adopt an improved two-dimensional SOCA-CFAR algorithm to meet the demands of multi-target scenarios. Beamforming was chosen as the primary method for azimuth measurement. After confirming target information, the nearest neighbor data association (NNDA) is used to distinguish multiple targets, and the Kalman filter is employed to achieve multi-targets tracking.

In the experimental aspect, we use the AWR1443 BOOST radar module developed by Texas Instruments (TI), employing a SIMO (Single-Input Multiple-Output) system with 1 transmit antenna (Tx) and 4 receive antennas (Rx). Data is obtained through the DCA1000 EVM capture card, and our proposed algorithms are verified using MATLAB. In comparison with the software mmWave Studio developed by TI, our system can detect targets more accurately and has the function of trajectory tracking that the software lacks.

In this thesis, the hardware was designed for the more complex and computationally intensive parts of the overall system using TSMC 90nm process technology. Finally, the core area of the chip is 1.95 mm² with a gate count of 164.4 k, a maximum operating frequency of up to 163 MHz, an average system power consumption of 12.5 mW, and a throughput of 10.48 M sample/s. That is equivalent to a frame containing data from the 4 receiving antennas to be processed in 1.56 ms.

摘要 i
Abstract ii
目錄 iv
圖目錄 vii
表目錄 x
第一章 序論 1
1.1 研究背景 1
1.2 研究動機 2
1.3 論文架構 3
第二章 雷達系統理論 4
2.1 雷達原理 4
2.2 脈衝雷達(PULSED RADAR) 4
2.3 連續波雷達(CONTINUOUS WAVE RADAR) 5
2.3.1 都普勒雷達(Doppler Radar) 5
2.3.2 調頻連續波(Frequency Modulated Continuous Wave) 6
2.4 雷達陣列天線架構 11
2.4.1 雷達傳送端理論 12
2.4.2 雷達接收端理論 15
第三章 移動物件偵測方法與設計 17
3.1 物件偵測方法介紹 17
3.1.1 漢寧窗(Hanning Window) 18
3.1.2 距離-快速傅立葉轉換(Range-FFT) 19
3.1.3 都卜勒-快速傅立葉轉換(Doppler-FFT) 21
3.1.3.1 都卜勒效應分析 21
3.1.3.2 點波源速度估計 22
3.1.3.3 處理概述 23
3.2 移動目標指示濾波器 25
3.2.1 MTI背景減法 25
3.2.2 MTI脈衝響應濾波器 26
3.2.2.1 單延遲濾波器 26
3.2.2.2 有限脈衝響應濾波器 28
3.2.2.3 無限脈衝響應濾波器 29
3.2.3 MTI濾波器改良 30
3.3 恆定虛警率目標偵測 34
3.3.1 傳統一維CFAR算法 34
3.3.2 改進式二維CFAR算法 36
3.3.3 多物件模擬結果 38
3.4 方位角量測技術 40
3.4.1 MUISC演算法 40
3.4.2 角度-快速傅立葉轉換(Angle-FFT) 42
3.4.3 接收端波束成型(Beamforming) 44
3.4.4 方法比較 48
第四章 多目標卡爾曼濾波演算法 49
4.1 卡爾曼濾波器介紹 49
4.1.1 公式與基本原理 50
4.1.2 單物件追蹤模擬結果 55
4.2 多物件追蹤技術 57
4.2.1 資料關聯演算法 57
4.2.2 多物件追蹤系統 59
4.3 實驗 61
4.3.1 實驗系統 61
4.3.2 實驗結果 63
4.4 雷達物件追蹤文獻比較 69
第五章 FMCW雷達訊號前處理硬體設計 71
5.1 晶片設計流程 72
5.2 硬體架構設計與模擬 74
5.2.1 功能方塊圖 74
5.2.2 定點數量化分析 75
5.2.3 硬體架構 78
5.2.3.1 漢寧窗 78
5.2.3.2 RADIX-2 快速傅立葉轉換 79
5.2.3.3 資料存取與記憶體 83
5.2.3.4 MTI濾波器 84
5.2.4 硬體模擬結果 86
5.3 合成結果 89
5.3.1 面積 89
5.3.2 功耗 90
5.4 晶片設計結果 91
5.4.1 IO介面 91
5.4.2 佈局平面圖 92
5.4.3 Cell面積佔比 93
5.4.4 功能驗證 93
5.4.5 晶片規格 95
第六章 結論與未來展望 96
6.1 結論 96
6.2 未來研究方向 97
參考文獻 98

[1] D. Deslandes, Ke Wu, "Integrated microstrip and rectangular waveguide in planar form," IEEE Microwave and Wireless Components Lett., 2001, vol.11, pp.68-70.
[2] D. Jung, S. Cheon, D. Kim, J. Yoon and B. Kim, "Short-Time Remote Heart Rate Measurement Based on mmWave FMCW Radar Frame Structure," IEEE Antennas and Wireless Propagation Letters, 2023, vol. 22, no. 6, pp. 1301-1305.
[3] D. G. Bresnahan and Y. Li, "Classification of Driver Head Motions Using a mm-Wave FMCW Radar and Deep Convolutional Neural Network," IEEE Access, 2021, vol. 9, pp. 100472-100479.
[4] M. S. R. Akash, M. N. Shahria, M. A. Morshed, S. S. Rodsee, N. Hannan and M. H. Imam, "Elderly Patient Monitoring and Fall Detection Using mmWave FMCW Radar System," International Conference on Computer and Information Technology (ICCIT), Cox's Bazar, Bangladesh, 2023, pp. 1-6.
[5] V. C. Chen, F. Li, S. S. Ho and H. Wechsler, "Micro-Doppler effect in radar: phenomenon model and simulation study", IEEE Transactions on Aerospace and Electronic Systems, 2006, vol. 42, no. 1, pp. 2-21.
[6] A. Anghel, G. Vasile, R. Cacoveanu, C. Ioana and S. Ciochina, "Short-Range Wideband FMCW Radar for Millimetric Displacement Measurements," IEEE Transactions on Geoscience and Remote Sensing, 2014, vol. 52, pp. 5633-5642.
[7] Q. Chen and T. Hong, "A novel mechanism for lane detection," International Congress on Image and Signal Processing, Chongqing, 2012, pp. 1750-1753.
[8] Z. Xu, Y. Chen and P. Zhang, "A Sparse Uniform Linear Array DOA Estimation Algorithm for FMCW Radar," IEEE Signal Processing Letters, 2023, vol. 30, pp. 823-827.
[9] Y. -Z. Ma, C. Cui, B. -S. Kim, J. -M. Joo, S. H. Jeon and S. Nam, "Road clutter spectrum of BSD FMCW automotive radar," European Radar Conference (EuRAD), Paris, France, 2015, pp. 109-112.
[10]R. Wan, Y. Song, T. Mu and Z. Wang, "Moving Target Detection Using the 2D-FFT Algorithm for Automotive FMCW Radars," International Conference on Communications, Information System and Computer Engineering (CISCE), Haikou, China, 2019, pp. 239-243.
[11]A. F. Martone, K. Ranney and C. Le, "Noncoherent Approach for Through-the-Wall Moving Target Indication," IEEE Transactions on Aerospace and Electronic Systems, 2014, vol. 50, no. 1, pp. 193-206.
[12]A. G. Stove, “Linear FMCW radar techniques,” IEE Proc. F Radar Signal Process, 1992, vol. 139, no. 5, pp. 343–350.
[13]M. Ash, M. Ritchie and K. Chetty, "On the Application of Digital Moving Target Indication Techniques to Short-Range FMCW Radar Data," IEEE Sensors Journal, 2018, vol. 18, no. 10, pp. 4167-4175.
[14]O. Panetas-Felouris, K. P. Pagkalos and S. Vlassis, "Moving Average Filter in Time-Mode Signal Processing," 2024 Panhellenic Conference on Electronics & Telecommunications (PACET), Thessaloniki, Greece, 2024, pp. 1-4.
[15]M. S. Shbat, J. H. Yi and V. Tuzlukov, "CFAR detectors employed by radar sensor systems," International Conference on Control, Automation and Systems, Jeju, Korea (South), 2012, pp. 518-522.
[16]M. Weiss, "Analysis of Some Modified Cell-Averaging CFAR Processors in Multiple-Target Situations," IEEE Transactions on Aerospace and Electronic Systems, 1982, vol. AES-18, no. 1, pp. 102-114.
[17]G. V. Trunk, "Range Resolution of Targets Using Automatic Detectors," IEEE Transactions on Aerospace and Electronic Systems, 1978, vol. AES-14, no. 5, pp. 750-755.
[18]C. Xu, Y. Li, C. Ji, Y. Huang, H. Wang and Y. Xia, "An improved CFAR algorithm for target detection," International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS), Xiamen, China, 2017, pp. 883-888.
[19]M. Devendra and K. Manjunathachari, "DOA estimation of a system using MUSIC method," International Conference on Signal Processing and Communication Engineering Systems, Guntur, India, 2015, pp. 309-313.
[20]M. Yang, W. Li and W. Zhang, "FFT Angle Measurement Method Based on Sparse Line Array," Chinese Control Conference (CCC), Shenyang, China, 2020, pp. 3083-3086.
[21]B. D. Van Veen and K. M. Buckley, "Beamforming: a versatile approach to spatial filtering," IEEE ASSP Magazine, 1988, vol. 5, no. 2, pp. 4-24.
[22]U. Hamid, R. A. Qamar and K. Waqas, "Performance comparison of time-domain and frequency-domain beamforming techniques for sensor array processing," International Bhurban Conference on Applied Sciences & Technology (IBCAST) , 2014, Islamabad, Pakistan, pp. 379-385.
[23]胡庭睿, "距離優先多物件偵測之束波成型FMCW雷達研究", 碩士論文,國立中興大學電機工程學系所, 2021.
[24]C. T. Ginalih, A. S. Jatmiko and R. Darmakusuma, "Simple Application of Kalman Filter On a Moving Object in Unity3D," International Conference on Interactive Digital Media (ICIDM), Bandung, Indonesia, 2020, pp. 1-3.
[25]H. Liu, Z. Hu, D. Zhou and H. Tian, "Cumulative Probability Distribution Model for Evaluating User Behavior Prediction Algorithms," International Conference on Social Computing, Alexandria, VA, USA, 2013, pp. 385-390.
[26]Q. Li and D. Xu, "Application of an Improved Nearest Neighbor Method," International Applied Computational Electromagnetics Society (ACES-China) Symposium, Chengdu, China, 2021, pp. 1-2.
[27]F. P. Witte and D. Lucas, "Probabilistic tracking in a multitarget environment," IEEE Conference on Decision and Contro, San Diego, CA, USA, 1978, pp. 1212-1216.
[28]D. Reid, "An algorithm for tracking multiple targets," IEEE Transactions on Automatic Contro, 1979, vol. 24, no. 6, pp. 843-854.
[29]Kuo-Chu Chang and Y. Bar-Shalom, "Joint probabilistic data association for multitarget tracking with possibly unresolved measurements and maneuvers," IEEE Transactions on Automatic Control, 1984, vol. 29, no. 7, pp. 585-594.
[30]T. Huang, J. Hao, Z. D. Chen, G. Wen, J. Li and H. Zhao, "An indoor FMCW localization system with the application of the extended Kalman filter," IEEE MTT-S International Wireless Symposium (IWS), Chengdu, China, 2018, pp. 1-4.
[31]R. Xiong, X. Feng, H. Zheng and Z. D. Chen, "Linear FMCW Radar System for Accurate Indoor Localization and Trajectory Detection," International Conference on Computing, Networking and Communications (ICNC), Big Island, HI, USA, 2020, pp. 741-745.
[32]R. Liu, B. Jiang, T. Fan, L. Yang, Y. Lin and H. Xu, "An Indoor Human Tracking System Using a Compact 24GHz FMCW Radar with 250MHz Bandwidth," IEEE MTT-S International Wireless Symposium (IWS), Harbin, China, 2022, pp. 1-3.
[33]Z. Zhang, X. Wang, D. Huang, X. Fang, M. Zhou and Y. Zhang, "MRPT: Millimeter-Wave Radar-Based Pedestrian Trajectory Tracking for Autonomous Urban Driving," IEEE Transactions on Instrumentation and Measurement, 2022, vol. 71, pp. 1-17.
[34]S. H. Javadi, R. Feng, A. Bourdoux and H. Sahli, "Multi-target tracking pipeline for MIMO-FMCW radars based on modified GM-PHD," European Signal Processing Conference (EUSIPCO), Helsinki, Finland, 2023, pp. 1818-1822.
[35]B.-N. Vo and W.-K. Ma, “The Gaussian Mixture Probability Hypothesis Density Filter,” IEEE Transactions on Signal Processing, 2006, vol. 54, no. 11, pp. 4091–4104.
[36]L. Charaabi, E. Monmasson and I. Slama-Belkhodja, "FPGA-based real-time emulation of induction motor using fixed point representation," Annual Conference of IEEE Industrial Electronics, Orlando, FL, USA, 2008, pp. 2393-2398.
[37]D. Krishnegowda and P. Nagaraju, "Design and Synthesis of a 256-Point Radix-2 DIT FFT Core with Design Ware Library using Fixed-Point Number Representation," IEEE Delhi Section Conference (DELCON), New Delhi, India, 2022, pp. 1-5.