| With the variety of radars and the maturity of signal processing technology,radars have played a significant role in all aspects of life.The main purpose of radar is to estimate the target's distance,speed,and angle of arrival relative to the radar based on the signal reflected from the target,and then to locate and identify the target.Among many radars,Frequency Modulated Continuous Wave(FMCW)radars have many advantages,such as relatively low hardware cost,simple implementation,high accuracy,and the ability to penetrate some common obstacles,and are now widely used for positioning and motion detection and other fields.This article researches and designs a FMCW radar working in the 2.7G-3.7GHZ frequency band,which has 4 transmitting antennas and 12 receiving antennas,and is used to detect the movement of the human body in real time.The radar system can penetrate obstacles such as screens,wooden doors,glass doors,etc.,and can perform single-target or even multiple-target real-time positioning and motion detection with high accuracy.The hardware of the radar system is mainly divided into five modules: signal genera-tor,SWITCH module,antenna array module,microcontroller control module,and USRP(Universal Software Radio Peripheral)module.The signal generator is used to contin-uously output a stable frequency-modulated continuous wave signal to the antenna for signal transmission and reception.Since a multi-antenna combination of 4 transmissions and 12 receptions is used,the SWITCH module and the single-chip microcomputer mod-ule are required to jointly control the switching antennas to ensure that only one antenna is transmitting and one antenna is receiving at each moment.All the data of the 48 antenna pairs finally passed to the USRP module,and further uploaded to the PC for subsequent processing through the RJ45 interface.Based on the characteristics of the FMCW radar signal,this paper designs a set of real-time motion detection algorithms: for the data received on the PC,first perform phase synchronization to obtain data frames arranged in order of the antenna? then,through 2D-FFT transformation,we obtain a power spectrum image in the range-angle domain? then background removal and spatial filtering are performed to suppress the multipath effect of imaging? finally,the power spectrum image is binarized according to the background threshold,and then the weighted centroid and pixel intensity of each connected region are calculated.In this way,a heat map image with coordinates of the centroid of the detection target is obtained.Because the antennas of the hardware platform are exposed to the air for a long time during actual use,they are easily damaged,which will cause the hardware system to fail to work.In addition,due to the large number of antennas and the short signal transmission interval,the computing and storage capabilities of the PC have higher requirements.In order to solve these two problems,this paper introduces the Compressed Sensing(CS)algorithm in the framework of the original motion detection algorithm to recover the orig-inal complete 48 antenna pair data with high accuracy from a small amount of received antenna data.As a result,the life cycle of hardware equipment can be extended,and the demand for computing and storage capacity on the PC side can be reduced? even when a new radar of the same type needs to be built,the hardware cost and equipment size can be reduced.Finally,through experimental verification,when the radar system locates the tar-get,the error in the horizontal direction is 0.148 m,and the error in the vertical direction is 0.076 m.When the number of antenna transmitting / receiving pairs that are actually working is greater than or equal to 16,the original 48 antenna pairs can be completely recovered.In other words,after introducing the compressed sensing algorithm,only the original 33% data is needed to achieve the initial detection effect. |
