Through-the-Wall Target Detection Using GPR A-Scan Data: Effects of Different Wall Structures on Detection Performance



Abstract—Ground penetrating radar (GPR) is an

electromagnetic sensor based on the ultra-wideband radar technology that can also be used for through-the-wall (TTW) target recognition. Search for the presence of designated targets hidden behind the walls, such as stationary or moving human bodies or certain types of weapons, is addressed in various critical applications; in rescue missions after earthquakes or in military operations, etc. In such inverse problems, type of the wall is as important as the properties and location of the hidden target. Interpretation of the basic A-Scan GPR signals is a challenging task in the TTW target detection problem especially when the wall is constructed by bricks containing air-filled holes.

In this paper, a simplified TTW target detection scenario is defined using cylindrical targets made of conductor or plastic materials. The target is placed behind the brick wall at different distances where the wall is made of either homogeneous solid bricks or inhomogeneous bricks that contain periodically located air-filled compartments. GPR signals are simulated for such target detection scenarios using a numerical computation tool that is based on the finite difference time domain (FDTD) technique. Then, simulated signals are analyzed in time domain for preprocessing and target detection. Energy based signal features are used to eliminate strong early-time reflections from the front face of the wall to enhance the signal components scattered by the target for better detection performance.

Index Terms— Ground penetrating radar (GPR),

through-the-wall target detection (TTW), ultra-wideband sensing, feature extraction, preprocessing, cumulative energy curves.

I. INTRODUCTION

PR is a powerful ultra-wide band (UWB) tool that enables sensing the targets which are either buried in ground or hidden behind obstacles such as walls [1-4]. Through-the-wall (TTW) target detection is an important research area that can be used for a variety of civilian or military applications [5-8].

In this study, we have investigated the effects of two different brick wall structures on the performance of TTW target detection problem. In the simpler case, the wall is This work was supported by the METU-BAP-1 Scientific Research Projects Funds under the Project No. BAP-03-01-2016-005.

Mesut Dogan is with the Electrical and Electronics Engineering Department, Middle East Technical University and Ardahan University, Ankara, Turkey, (e-mail: mesut.dogan@metu.edu.tr).

Gonul Turhan-Sayan is with the Electrical and Electronics Engineering Department, Middle East Technical University, Ankara, Turkey, (e-mail: gtsayan@metu.edu.tr).

composed of homogeneous solid bricks. In the case of an inhomogeneous brick wall with periodic air-filled compartments, however, the target detection problem may become much more complicated. In this study, we will simulate GPR signals in the presence of both types of brick walls using the open source electromagnetic software GPRMax [9]. A-scan and B-scan type GPR signals are computed for different TTW target detection scenarios using cylindrical targets made of perfect electric conductor (PEC) or dielectric materials. In these simulations, cylinders are located at different distances behind the wall such that their axes are perpendicular to the wall.

Energy based features extracted from the A-scan type GPR signals are used for preprocessing and target localization simultaneously. The cumulative energy curve corresponding to a given A-scan GPR signal can be used to identify major clutter effects such as the strong reflections coming from the front face of the wall. Then, these reflections are eliminated by a simple time gating to relatively enhance the weaker signal components, which are scattered by the target itself, for better detection performance. It is important to note that cumulative energy curves help to eliminate the clutter effects and to locate a hidden target at one easy step. Use of the total energy value of a GPR signal together with its cumulative energy curve enhances the target detection capability of this approach [10].

II. ENERGY AND POWER FEATURES

When x(t) is a real valued, casual and continuous time signal, integral of the instantaneous power signal | ( )| is defined as the total energy (ET).

³

The cumulative energy curve E(t) of the time domain signal

x(t) is defined as: E t

³

W

W

As can be seen in (2), the cumulative energy curve E(t) is a non-decreasing time function [11,12], that shows sharp distinctive jumps in the presence of strong reflections caused by material discontinuities along the line of sight of the radar.

Through-the-Wall Target Detection Using GPR

A-Scan Data: Effects of Different Wall

Structures on Detection Performance

Mesut Dogan and Gonul Turhan-Sayan

III. SIMULATIONS AND RESULTS

In this study, time domain A-scan and B-scan GPR signals are computed by GPRMax for the simulation scenarios where a cylindrical target is located 5 cm, 10 cm, 15 cm or 20 cm away from the wall. As shown in Fig. 1, two different brick wall structures are chosen for this problem scenario where the thickness for both type of walls is 8 cm and the distance between the GPR (transmitter and receiver antennas) and the wall is set to be 5cm. The axis of the cylindrical target (either a perfect electric conductor (PEC) or plastic) is located perpendicular to the wall. All the test targets are chosen to have the same size with a radius of 6 cm and a height of 12 cm.

Fig. 1. Problem geometry used in the TTW simulations for (a) homogeneous solid brick wall, (b) inhomogeneous brick wall with periodically located, air-filled inner compartments.

Material properties (relative permittivity, relative permeability and conductivity) of the brick wall and of the plastic cylinder are shown in Table I.

TABLE I

MATERIAL PROPERTIES OF THEWALL AND THE PLASTIC TARGET

Object Material Ɛr μr σ (S/m)

Wall Brick 5 1 0.002

Target Plastic 2.1 1 0.000042

.

In the simulations, a single channel down-looking GPR is used with Hertzian dipole type transmitter and receiver antennas which are aligned along the x-axis. GPR moves in the z-direction (the down track direction) taking samples in every 2 cm. A broadband frequency scanning is simulated from 0.2 GHz to 6.7 GHz. Gaussian derivative type excitation pulse is used in the simulations.

(a)

(b)

Fig. 2. Unprocessed B-scan type GPR signals when a PEC cylinder is located 15 cm away from the (a) solid brick wall and (b) brick wall with air gaps.

B-scan type GPR signals are computed (combining A-scan signals which are simulated at 25 successive locations along the down-track direction) for a scenario where a PEC cylinder is placed at 15 cm distance from the back face of a brick wall. The contour plots for the B-scan signals are plotted in Fig. 2 (a) and (b) for the cases of solid (homogeneous) brick wall and for the inhomogeneous brick wall with air-filled compartments, respectively. In TTW problems, reflections from the front face of the wall are usually so strong that they might mask signals reflected and scattered by the target especially if target has a small radar cross section and/or placed far away from the wall. This situation is also observed in Fig. 2 for both types of the wall. Although the PEC cylinder is expected to create strong reflections when its flat cap surface is perpendicularly illuminated, these reflections are highly masked especially in the presence of the inhomogeneous brick wall due to multiple reflections caused by the air filled inner cavities of the wall.

A-scan type GPR signals of the PEC cylinder are shown in Fig. 3 for the placement distances of 5 cm, 10 cm, 15 cm and 20 cm. These raw signals are called “unprocessed” as they are not preprocessed yet to remove the strong reflections from the front surface of the wall. The strength of these reflections becomes weaker as the distance between the target and the wall increases, as expected.

(b)

Fig. 3. Unprocessed A-scan type GPR signals when a PEC cylinder is located at a distance of 5 cm, 10 cm, 15 cm and 20 cm from the back face of (a) the solid brick wall, (b) the inhomogeneous brick wall. These A-scan simulations correspond to the center location of the down-track range of the B-scan data.

Similar A-scan GPR simulations are also repeated for the plastic cylinder instead of using the PEC cylinder in the presence of inhomogeneous brick wall. The resulting unprocessed A- scan signals are shown in Fig. 4 revealing that the reflections from the plastic cylinder are almost completely masked by the strong response created by the inhomogeneous brick wall.

Fig. 4. Unprocessed A-scan type GPR signals of plastic cylinders located 5 cm, 10 cm, 15 cm and 20 cm away from the hollow brick wall (corresponding to the A-scan simulation at 25th_{cm of B-scan data).}

As mentioned earlier, a major aim in preprocessing is to remove most of the major and strong early-time clutter effects such as the antenna couplings and reflections from the wall surfaces. In this work, we remove the reflections from the front wall surface in the preprocessing step. Removal of the first back wall reflection is also straightforward in the homogeneous wall case but removal of multiple wall reflections is a difficult problem especially in the case of inhomogeneous brick wall (due to its inner cavity walls). These multiple wall reflections are major sources of errors in TTW detection problem as they overlap with signatures coming from the targets behind the walls. In this study, we used cumulative energy curves to detect all kinds of discontinuities due to wall structures or due to targets. The cumulative energy curves corresponding to the unprocessed A-scan signals (given in Fig. 3) are computed and plotted in Fig. 5.

The reflections caused by material discontinuities (along the GPR’s line of sight) are enhanced and localized in cumulative energy curves; sharp increases are clearly observed in these cumulative energy curves at major reflection locations as seen in Fig. 5 (a) and (b). Front face reflections of the wall can be removed from unprocessed A-scan signals by tracing the rapid variations in the local slope values of the corresponding cumulative energy curves.

(a)

(b)

Fig. 5. Cumulative energy curves of the unprocessed A-scan type GPR signals when the PEC cylinder is located at a distance of 5 cm, 10 cm, 15 cm and 20 cm from the (a) solid brick wall, (b) inhomogeneous brick wall.

Preprocessed A-scan type GPR signals (in the case of PEC cylinders) are shown in Fig. 6. Comparison of the unprocessed (shown in Fig. 3) and preprocessed (shown in Fig. 6) A-scan signals shows that the PEC cylinder is more easily detectable after preprocessing.

(b)

Fig. 6. Preprocessed A-scan type GPR signals when the PEC cylinder is located at a distance of 5 cm, 10 cm, 15 cm and 20 cm from the (a) solid brick wall and (b) inhomogeneous brick wall.

(a)

(b)

Fig. 7. Cumulative energy curves of the preprocessed A-scan type GPR signals (shown in Fig. 6 in the case of the PEC cylinder) located 5 cm, 10 cm, 15 cm and 20 cm away from the (a) solid brick wall and (b) inhomogeneous brick wall.

Cumulative energy curves of the preprocessed A-scan signals in the presence of the plastic cylindrical target are also computed and plotted in Fig. 8 (a) and (b) for homogeneous solid brick wall and for inhomogeneous brick wall with air gaps, respectively. As shown in all the results demonstrated so far, the cumulative energy curves and total energy values of the preprocessed A-scan signals (the final value reached by the cumulative energy curves at sufficiently late times) are computationally simple but very useful energy features to detect conducting and dielectric targets in the TTW problem.

(a)

(b)

Fig. 8. Cumulative energy curves of the preprocessed A-scan type GPR signals (in the case of the plastic cylinder) located 5 cm, 10 cm, 15 cm and 20 cm away from the (a) solid brick wall and (b) inhomogeneous brick wall. IV. CONCLUSION

In this paper, we have demonstrated an energy-based preprocessing and target detection method for the TTW target detection problem. Cumulative energy curves and total energy values corresponding to A-scan GPR signals are used for both preprocessing (removal of early-time strong clutter effects resulting from the front face of the wall for better detection performance) and target detection purposes, simultaneously.

TTW detection of conducting and dielectric targets of cylindrical shape is investigated for both homogeneous and inhomogeneous wall structures when the target is placed at different distances from the wall. The results have demonstrated that detection of a target may be very difficult in the presence of an inhomogeneous brick wall with multiple inner reflections, especially in the case of plastic targets.

REFERENCES

[1] L. P. Peters, J. J. Daniels and J. D. Young, "Ground Penetrating Radar as a Subsurface Environmental Sensing Tool," in Proceedings of the IEEE, vol. 82, no. 12, pp. 1802-1822, Dec 1994.

[2] H. M. Jol and C. S. Bristow, ”GPR in sediments: advice on data collection, basic processing and interpretation, a good practice guide,” Geological Society, London, Special Publications, 211, 9-27, January 2003, https://doi.org/10.1144/GSL.SP.2001.211.01.02

[3] M. C. Wicks, "RF Tomography with Application to Ground Penetrating Radar," 2007 Conference Record of the Forty-First Asilomar Conference on Signals, Systems and Computers, Pacific Grove, CA, 2007, pp. 2017-2022. doi: 10.1109/ACSSC.2007.4487591

[4] N. Maaref, P. Millot, C. Pichot and O. Picon, "A Study of UWB FM-CW Radar for the Detection of Human Beings in Motion Inside a Building," in IEEE Transactions on Geoscience and Remote Sensing,

[5] E. J. Baranoski, “Through-wall imaging: Historical perspective and future directions,” Journal of the Franklin Institute, vol. 345, issue 6, 2008, pp. 556-569, ISSN 0016-0032, http://dx.doi.org/10.1016/j.jfranklin.2008.01.005.

[6] P. K. Verma, A. N. Gaikwad, D. Singh, and M. J. Nigam, "Analysis of clutter reduction techniques for through wall imaging in UWB range," Progress In Electromagnetics Research B, Vol. 17, 29-48, 2009. doi:10.2528/PIERB09060903

[7] Chieh-Ping Lai, Ram M. Narayanan, "Through-wall imaging and characterization of human activity using ultrawideband (UWB) random noise radar," Proc. SPIE 5778, Sensors, and Command, Control, Communications, and Intelligence (C3I) Technologies for Homeland Security and Homeland Defense IV, May 2005

[8] S. Hantscher, B. Praher, A. Reisenzahn and C. G. Diskus, "Comparison of UWB Target Identification Algorithms for Through-Wall Imaging

Applications," 2006 European Radar Conference, Manchester, England, 2006, pp. 104-107. doi: 10.1109/EURAD.2006.280284

[9] C. Warren, A. Giannopoulos and I. Giannakis, “GPRMax: Open Source Software to Simulate Electromagnetic Wave Propagation for Ground Penetrating Radar,” Computer Physics Communications, vol. 209, pp. 163-170, 2016

[10] M. Dogan and G. Turhan-Sayan, "Preprocessing of A-Scan GPR Data Based on Energy Features," Proc. SPIE 9823, Detection and Sensing of Mines, Explosive Objects, and Obscured Targets XXI, Baltimore, Maryland, USA, 98231E, May 2016

[11] G. Turhan-Sayan, “K-Pulse Estimation and Target Identification, Ohio State University,” Ph.D. dissertation, Electrical Engineering Department, Columbus, Ohio, USA, December 1988.

[12] M. Ayar, “Design of an Electromagnetic Classifier for Spherical Targets,” M.S. thesis, Middle East Technical University, Department of Electrical and Electronics Engineering, Ankara, Turkey, April 2005