MMW Sensor for Hidden Targets Detection and Warning ... - IEEE Xplore

4890

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 9, SEPTEMBER 2014

MMW Sensor for Hidden Targets Detection and Warning Based on Reflection/Scattering Approach Ayman Elboushi and Abdel Sebak

Abstract—An antenna sensor for millimeter wave (MMW) hidden target detection and warning applications is introduced. The sensor consists of three adjacent high gain microstrip/horn hybrid antenna elements. The central antenna acts as bi-static radar, while the two side antennas are used to receive the scattered back signals from a hidden object. The proposed antenna sensor has been employed in a detection/imaging system prototype based on time domain reflectometry (TDR). A very short pulse generated by a vector network analyzer is used to illuminate a three layers body model made from cotton, natural leather and reinforced papers. The model construction is chosen to emulate the presence of human body. Experimental scanning of a hidden metallic target has been conducted for three different orientations of the target. Compared to a single antenna sensor, the triple sensor shows a great enhancement in the detection ability and the constructed image resolution of a hidden target. Index Terms—Bi-static radar, hidden weapons detection, high gain antenna, hybrid antenna, microwave Imaging, millimeter wave (MMW), reflection/scattering, scanning sensor.

I. INTRODUCTION Due to the increase of international terroristic threats, there is a need for a fast, reliable and inexpensive detection tool for concealed weapons with passengers at airports and border gates. Although metallic detectors, presented for the first time in 1920 by Fisher [1], can be considered as an efficient tool for detecting hidden metallic weapons, it fails to detect non-metallic weapons and modern plastic explosives. Microwave based detection/imaging systems can be considered as a better choice for such scenarios. Farhat and Tricoles [2] are pioneers in the field of microwave holography. Their work involved imaging of concealed targets using both mono-static and bi-static radar techniques. A key component in such systems is the antenna sensor. Many antenna sensors have been developed to be integrated within such systems. A flexible printed antenna sensor for crack detection is presented in [3]; however, it lacks a directive beam which is mandatory for detection and imaging applications. An active corrugated lens antenna sensor, as part of interferometer system, was used in [4], [5] at W-band for homeland security detection and imaging applications. Parabolic dish reflectors sensors are employed in the ABOSCA radiometer, LPAS-2 radiometer systems [6], [7] and in the École Polytechnique passive imaging system [8]. A planar array of switchable slot antennas is introduced in [9] as an improved imaging sensor for a real-time microwave portable microwave camera system. Moreover, 2D multistatic multi-modules sparse array is adopted as human body scanning sensor in Rohde and Schwarz imaging system introduced in [10]. In [11], a 32–element circular array sensor is used in a radar interferometer for topographic imaging in industrial environment for the steel industry. One reported limitation of the system is the crosstalk between the receiving antenna elements. A 2-element Manuscript received November 11, 2013; revised March 24, 2014; accepted June 27, 2014. Date of publication July 08, 2014; date of current version September 01, 2014. The authors are with the Electrical and Computer Engineering Department, Concordia University, Montreal, Canada (e-mail: [email protected]; [email protected]). Color versions of one or more of the figures in this communication are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TAP.2014.2336655

microstrip patch antenna array based sensor is adopted in [12] to remotely monitor structural displacements. A linear electronically scanned MMW imaging array capable of rapid image production is employed in the imaging system proposed in [13] for non-destructive test (NDE) applications. Despite the good performance of the former antenna sensors, most of them are characterized by the bulky size, big volume and heavy weight or by the lack of directive beam which reflect on the overall performance of a MMW imaging system. In this perspective, an efficient, small size and low cost antenna sensor for MMW detection and imaging applications is investigated and presented. The proposed design in this communication exhibits a small size compared with other commercially available antenna sensors including dish antennas, metallic horn and lens antennas. In addition, it is based on printed circuit board (PCB) technology which leads to a lighter overall sensor weight and lower cost compared with other antenna sensors. The presented antenna sensor at 30 consists of three adjacent antenna elements separated by GHz. The distance between the antenna elements are chosen to minimize the undesired mutual coupling between them. The antenna elements integrated in this sensor are based on the hybrid antenna design presented in [14]. The proposed sensor is employed in a MMW detection system utilizes a short pulse in the MMW frequency range around 30 GHz generated by Agilent PNA-E8364B vector network analyzer (VNA) after proper calibration over the operating frequency range. The detection/imaging system is based on the principles of time domain reflectometry (TDR). The time domain representation of the pulse can be obtained by performing Inverse fast Fourier transform (IFFT) on both transmitted and received signals (the reflected back signal from the target). This principle of operation is used in [15] for building a complete ultra wide-band imaging system for breast cancer detection. In order to verify the antenna sensor performance, experimental measurements have been carried out on a simple body phantom made of three layers. The first layer is made of cotton tissue with a thickness of 0.3 mm followed by a 1 mm layer of natural animal leather and finally a supporting layer of reinforced paper with 2 mm thick. As a hidden target example, a 4 cm length and 1 cm width copper strip has been placed under the cotton layer. Three different 2D target positions have been studied. The image reconstruction process starts with successive data recording operations from each antenna element in the triple sensor for both the reflected back and scattered signals by the target. Then, the collected data are processed through a Matlab code in order to generate an image for the hidden target beneath the cotton layer. II. HYBRID MICROSTRIP/HORN ANTENNA The proper choice for the antenna element design will reflect on the overall performance of the scanning sensor. In order to reduce the size and weight of the antenna sensor, the antenna shown in Fig. 1(a) [14], is chosen to be used in the proposed antenna sensor. This design exhibits high gain, light weight and compact size compared with standard horn antenna working in the same frequency range. This antenna design consists of a printed microstrip circular patch ” 0.7874-mm-thick grounded subon a low permittivity “ microstrip line on the backside of strate. The patch is fed by a a second 0.635-mm-thick dielectric layer with a higher permittivity ” through a rectangular coupling slot etched in the middle “ ground plane. The coupling slot has an approximate length of a quarter wave length around the antenna resonance frequency of 31 GHz. This mode of the circular patch that feeding mechanism excites feeds a metallic conical horn with wall thickness of 0.254 mm. The optimized antenna parameters are tabulated in [14].

0018-926X © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.


4891

Fig. 1. Geometry of (a) the proposed antenna element [14] (b) triple antenna sensor. Fig. 3. The radiation pattern comparison between the stand alone antenna element and the antenna element within the proposed triple antenna sensor (a) E-plane (b) H-plane.

Fig. 2. Reflection coefficient element sensor.

and the mutual coupling

of the triple

Fig. 4. A photo of the fabricated prototype of the triple antenna sensor.

III. TRIPLE ELEMENT ANTENNA SENSOR As shown in Fig. 1(b), the proposed triple antenna sensor consists of three adjacent hybrid antenna elements with three separated input ports. The design parameters of these antenna elements are identical to the dimensions stated in [14]. The separation distance between the at 30 GHz in order to miniantenna elements is chosen to be mize the mutual coupling between them without increasing the overall size of the triple sensor. The reflection coefficient and the mutual coupling between the antenna elements are illustrated in Fig. 2. There is a frequency shift of about 0.5 GHz between the experimental and the which can be accounted for CST simulated reflection coefficient some fabrication errors especially in the soldering process and alignment of mounted horns. It can be noticed that the measured mutual over the range from 30 GHz to 32 coupling does not exceed GHz, while the measured impedance bandwidth of the antenna element ranged from 30.2 GHz to 33.6 GHz. The proposed triple antenna sensor exhibits remarkable radiation characteristics where it has broad side di-

rective patterns in both E-plane and H-plane with good radiation pattern stability and high average gain of 10.5 dB over the operating band. There is a good agreement between the CST and HFSS simulated results, and the corresponding measured E-plane (XZ) and H-plane (YZ) radiation patterns at 31 GHz. The radiation patterns of the middle antenna in the triple sensor, when the two other antenna ports are terminated with matched loads, are shown in Fig. 3. It can be concluded that there is a good match in the performance of the stand-alone antenna element which demonstrates the low mutual coupling level between the sensor elements. A photograph of the fabricated triple antenna sensor is shown in Fig. 4. IV. MMW DETECTION/IMAGING SYSTEM SETUP The proposed MMW detection/imaging system setup for hidden weapons detection is shown in Fig. 5. The system includes a 3D

4892


Fig. 5. MMW detection/imaging setup using the triple antenna sensor.

Fig. 7. Scanning points over the body model and metallic target locations.

V. DETECTION/IMAGING ALGORITHM

Fig. 6. A photo of the MMW detection/imaging setup.

movable Lynxmotion robotic arm (AL5D) [16] to provide the required scanning motion for the proposed triple-antenna sensor. The middle antenna element input port of the sensor is connected to channel (1) of the VNA, while channel (2) is connected to the side antenna and . elements successively to record received signal, to get The VNA is used to generate a synthesized pulse after a standard full two-port calibration procedure from 30 GHz to 32 GHz. An absorber is mounted vertically behind the mounted antenna sensor to prevent any undesired back reflections. The proposed sensor is centered to face the three layers body model in the X-Y plane, whereas the separation distance between the antenna sensor and the first wall 10 cm is chosen carefully to ensure working in the far-field region of the antenna. A photograph for the MMW detection/imaging setup is shown in Fig. 6. The body model to be scanned is built using three different layers; the first one is made of cotton with a thickness of 0.3 mm followed by a 1 mm layer of natural animal leather and finally a 2 mm thick of supporting layer of reinforced paper. A small area of the body model is chosen as the scanning area, shown in Fig. 7, where it . is divided into 111 square-shaped regions each with area of The robotic arm scanning motion, controlled by RIOS software [16], is adjusted to make the middle antenna of the sensor facing the center of each square, i.e., the X and Y resolution are chosen to be 1 cm. For demonstration purposes, a rectangular metallic strip made of copper with dimensions of 4 cm length, 1 cm width and 0.25 mm thickness is selected to be the hidden target. The metallic target is placed under the cotton layer with three different orientations. As shown in Fig. 7, the first orientation is vertical (red dashed rectangle), the second is inclined by 45 degree (green dashed rectangle); while the third is horizontal (blue dashed rectangle).

The detection algorithm starts by recording the frequency domain from the hidden target representation data for the reflected signal and and the body model as well as the scattering signals received from the side antennas. Using inverse fast Fourier transform (IFFT) embedded code in the VNA, a time domain representation of the reflection and scattering information can be obtained ( and ). In order to cancel the effect of antenna elements, cables and connectors, the received signals are normalized by subtracting the and sensor response in front of an absorber

(1) and contain only the reThe resulted signals flection and the scattering information from the body model and the target under study. Due to the path difference between the reflection and the scattering signals, time equalization process is needed for the scattering signals. To eliminate all the remaining unwanted reflections during the digital signal processing, the normalized signals are multiplied by a low shape factor Kaiser Window,

(2) where, is a small delay time for path difference equalization of the and , and is the time shift required scattered signals to center the Kaiser window at the required position, i.e., the first peak of the reflected/scattered signals, taken here to be 0.8 ns.

(3) otherwise where, is the zeroth order of the modified Bessel function of the first kind, is an arbitrary real number that controls the shape of the


Fig. 9. Normalized reflection signal away.

Fig. 8. Triple antenna sensor response in front of an absorber ( and ).

4893

over the metallic target and 2 cm

window (taken here to be 0.5) and m is an integer, and the length of the . sequence is Using a simple integration subroutine the area under each curve, which represents the reflection/scattering magnitude, can be estimated. Furthermore, the resulted signals are added together after multiplying them by weighting factors: (4) where is the square number of the scanning domain shown in Fig. 9 are the required adjustment weights. and The above algorithm steps are repeated for each scanned point within the scanned area. Finally, image reconstruction can be carried out after correlating the obtained results to the 2D points’ locations and interpolating them over the scanned area. The detection/imaging algorithm steps are implemented using a Matlab code.

Fig. 10. Normalized scattering signal cm away.

over the metallic target and 2

VI. RESULTS AND DISCUSSION After constructing and connecting the system shown in Fig. 5 and calibrating the VNA using standard calibration procedure over the frequency range of 30 GHz to 32 GHz, the reflection and the scattering data in frequency domain are measured. The measured results are then converted into time domain using the time domain conversion tool embedded in the VNA. Data collection and recording processes are done sequentially to cover all the testing points in the scanning area. Fig. 8 shows the sensor response in front of an absorber. There are some differences between the peaks of the scattered signals and because of some fabrication, alignment and soldering variations in the three mounted horns. However, these differences can be considered as static errors that can be added to every sampling point and easily eliminated through the normalization process. Figs. 9–11 show a comparison between the normalized reflected , the normalized scattered signals and back signal signal levels for the vertical target orientation in two different sensor positions. The first one when the sensor is centered over square No. 50, i.e., over the metallic target, while the second when the sensor position is shifted 2 cm in the negative y direction away from the target, i.e., centered over square No. 48. It can be noticed a level dif, ference of 10 mV, 15 mV and 2 mV in the first peak of and , respectively. These levels are fair enough to differentiate between the existence and no existence of the target. Almost same level differences with some minor variations can be detected for the other target orientations.

Fig. 11. Normalized scattering signal away.

over the metallic target and 2 cm

Reflection and scattered data can be processed using Matlab in order to plot a complete image of the scanning domain with the hidden target. The first investigated case is the vertical orientation of the target. As shown in Fig. 12(a), the reflection based image of the target is about 1 cm shorter than its actual size. The image based on combined reflection and scattering data, shown in Fig. 12(b), shows a better realistic size.Fig. 13(a) and (b) show the reconstructed images for the hidden

4894


sensor shows a remarkable imaging/detection performance and target size estimation in the experimental work.

REFERENCES

Fig. 12. Reconstructed image for the vertical target using (a) reflection data only (b) reflection plus scattering data.

Fig. 13. Reconstructed image for the 45 inclined target using (a) reflection data only (b) reflection plus scattering data.

Fig. 14. Reconstructed image for the horizontal target using (a) reflection data only (b) reflection plus scattering data.

target with 45 degree orientation. The reconstructed image in Fig. 13(a) based on reflection data only shows strong reflections that may be due to the edge diffraction by testing points located on the edges of the target. However, considering both reflection and scattering data plays an important role in decreasing the edge diffraction effect as shown in Fig. 13(b). The horizontal orientation target case is presented in Fig. 14. Considering both reflection and scattering data in this case enhances the reconstructed target image dimensions, especially the target width and its resolution.

VII. CONCLUSION A triple antenna sensor for MMW detection/imaging applications is presented. The proposed sensor characterized by its small size, low cost and light weight which makes it a good candidate for detection and warning applications. The developed imaging/detection algorithm is implemented using a Matlab code to process, interpolate and plot the measured data over the scanning domain. Sample detection/imaging experiments over the MMW range have been carried out to detect a rectangular metallic target hidden under a cotton layer of a three-layer target emulating the human body. Three different target orientations were considered (inclined, horizontal and vertical). The triple antenna

[1] Wikipedia, (2013, Nov. 10), Metal Detector [Online]. Available: http://en.wikipedia.org/wiki/Metal_detector [2] G. Tricoles and N. H. Farhat, “Microwave holography: Applications and techniques,” Proc. IEEE, vol. 65, no. 1, pp. 108–121, Jan. 1977. [3] H. Huang, “Flexible wireless antenna sensor: A review,” IEEE Sensors J., vol. 13, no. 10, pp. 3865–3872, Oct. 2013. [4] A. Dallinger, S. Schelkshorn, and J. Detlefsen, “Short distance related security millimeter-wave imaging systems,” presented at the German Microwave Conf., Apr. 2005. [5] A. Dallinger, S. Schelkshorn, and J. Detlefsen, “Millimeter-wave imaging of humans—basic experiments,” in Dig. of the Joint 29th Int. Conf. on Infrared and Millimeter Waves Conf. and 12th Int. Conf. on Terahertz Electron., Sep.-Oct. 1-27, 2004, pp. 521–522. [6] M. Peichl, S. Dill, M. Jirousek, and H. Suess, “Passive microwave remote sensing for security applications,” in Proc. Eur. Radar Conf., Oct. 10–12, 2007, pp. 32–35. [7] D. Stephan, P. Markus, S. Eric, and S. Helmut, “Passive MMW imaging systems for security applications,” in Proc. Int. Radar Symp., Sep. 2009, pp. 9–11. [8] D. Ali, A. Ghiotto, T. Djerafi, and K. Wu, “Early demonstration of a passive millimeter-wave imaging system using substrate Integrated Waveguide Technology,” in Proc. 11th Mediterranean Microwave Symp., Sep. 8–10, 2011, pp. 215–218. [9] M. T. Ghasr, M. A. Abou-Khousa, S. Kharkovsky, R. Zoughi, and D. Pommerenke, “Portable real-time microwave camera at 24 GHz,” IEEE Trans. Instrum. Meas., vol. 60, no. 2, pp. 1114–1125, Feb. 2012. [10] S. S. Ahmed, A. Schiessl, and L. Schmidt, “A novel fully electronic active real-time imager based on a planar multistatic sparse array,” IEEE Microw. Theory Tech., vol. 59, no. 12, pp. 3567–3576, Dec. 2011. [11] E. Nilsson and L. Baath, “Radar interferometric measurements with a planar patch antenna array,” IEEE Sensors J., vol. 7, no. 7, pp. 1025–1031, Jul. 2007. [12] V. Rizzoli, A. Costanzo, E. Montanari, and A. Benedetti, “A new wireless displacement sensor based on reverse design of microwave and millimeter-wave antenna array,” IEEE Sensors J., vol. 9, no. 11, pp. 1557–1566, Nov. 2009. [13] M. T. Ghasr, S. Kharkovsky, R. Bohnert, B. Hirst, and R. Zoughi, “30 GHz linear high-resolution and rapid millimeter wave imaging system for NDE,” IEEE Trans. Instrum. Meas., vol. 61, no. 9, pp. 4733–4740, Jun. 2013. [14] A. Elboushi and A. Sebak, “High-gain hybrid microstrip/conical horn antenna for MMW applications,” IEEE Antennas Wireless Propag. Lett., vol. 11, pp. 129–132, 2012. [15] W. C. Khor, M. E. Bialkowski, and S. Crozier, “Microwave imaging using a planar scanning system with step-frequency synthesized pulse,” in Proc. Asia-Pacific Microwave Conf., Dec. 4–7, 2005, vol. 1, pp. 4–4. [16] Lynxmotion Company, (2013, Nov. 10), Al5d Robotic Arm [Online]. Available: http://www.lynxmotion.com/c-130-al5d.aspx