利用超寬頻陣列雷達實現三維之呼吸運動偵測

Abstract

本研究的主旨是利用電子掃描控制之二維超寬頻雷達陣列組成二維合成孔徑，做三維影像並偵測與評估三維的人體呼吸運動。而本研究的主要貢獻為利用電子控制掃描，配合印刷天線陣列，組成二維交叉型陣列天線，並以此系統擷取三維影像。三維影像的較二維影像多出一個維度的資訊，其定位能力更佳，更符合真實的三維空間，也因此應用層面更廣，是以二維呼吸偵測有改進成為三維呼吸偵測之必要。在先前的研究中，合成孔徑是利用步徑馬達驅動喇叭天線，沿水平方向移動組合而成。但此方法在應用時有兩項問題:一者機械式掃描速度掃描速度慢，二者在系統架構上需外接馬達、軌道等。因此，我們提出以寬頻且具吸收性的射頻多工器做電子式掃描以解決上述之兩項問題。另外，我們目標將系統信號中心頻率由1.5 GHz提升到3 GHz，以符合聯邦通訊傳播委員會對於超寬頻信號免執照頻段的使用。另外，在發射信號相同部分頻寬前提下，提高中心頻率可使整體頻寬變寬，藉以得到較好的影像解析度。然而這部分的電路設計雖然達成頻率提高和增加頻寬，但產生信號的能量不足，因此後續系統中仍使用中心頻率1.5 GHz之發射電路。最後，我們利用三維超寬頻雷達成像系統，搭配呼吸偵測演算法: 廣義非同調性因子(generalized incoherence factor, GICF)以及濾波器組廣義同調性因子(filter bank assisted generalized coherence factor, FBGCF)，進行三維人體呼吸運動之偵測與評估，但偵測結果無法準確找出受試者位置，推測是天線發射能量不足、天線之間的串音，以及系統信號延遲時間校正不夠精確的緣故。

The purpose of this study is to using electronic scanning 2D PCB array antenna to achieve 3D imaging as well as 3D respiratory motion detection and estimation. The proposal for 3D imaging is because the positioning ability for 3D imaging is better than 2D imaging with providing additional dimension in real case and applications. The 2D array antenna applied in this study is a X-shaped array antenna. The imaging and the good positioning will be based on the use of UWB signal. The 1D synthetic aperture in previous study is contributed to mechanical scanning by step motor. However, the scanning speed and the setup time for mechanical scanning are concerns when it comes to applications. Therefore, using electronic scanning to scan over our array antenna is proposed. The center frequency of transmitted signal should follow the FCC defined unlicensed band for UWB use. Consequently, the transmission circuit is redesigned to generate signal with center frequency at 3GHz. Under the same fractional bandwidth, the bandwidth would increase and the resolution would be better with the raise of center frequency. However, in this study, the circuits meeting our proposal could not generate signals with enough power. Hence the original 1.5 GHz is still applied in the following system. Finally, 3D UWB imaging system and the respiration detection algorithm, generalized incoherence factor (GICF) and filter bank assisted generalized coherence factor (FBGCF) are applied to achieve both 3D imaging and human respiratory motion detection and estimation. However, the detection outcome is not ideal due to poor system performance such as antenna power gain, crosstalk, and the inaccurate system delay calibration.