2544 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 8, AUGUST 2007 Multistatic Ground-Penetrating Radar Experiments Tegan Counts, Student Member, IEEE, Ali Cafer Gurbuz, Student Member, IEEE, Waymond R. Scott, Jr., Senior Member, IEEE, James H. McClellan, Fellow, IEEE, and Kangwook Kim, Member, IEEE Abstract—A multistatic ground-penetrating radar (GPR) sys- tem has been developed and used to measure the response of a number of targets to produce data for the investigation of multi- static inversion algorithms. The system consists of a linear array of resistive-vee antennas, microwave switches, a vector network analyzer, and a 3-D positioner, all under computer control. The array has two transmitters and four receivers which provide eight bistatic spacings from 12 to 96 cm in 12-cm increments. Buried targets are scanned with and without surface clutter, which is a layer of rocks whose spacing is empirically chosen to maximize the clutter effect. The measured responses are calibrated so that the direct coupling in the system is removed, and the signal reference point is located at the antenna drive point. Images are formed us- ing a frequency-domain beamforming algorithm that compensates for the phase response of the antennas. Images of targets in air validate the system calibration and the imaging algorithm. Bistatic and multistatic images for the buried targets are very good, and they show the effectiveness of the system and processing. Index Terms—Beamforming, ground-penetrating radar (GPR), measurement, multistatic radar, ultrawideband. I. I NTRODUCTION G ROUND-PENETRATING radars (GPRs) image subsur- face structures by transmitting short electromagnetic pulses into the ground and by receiving the reflections [1]. Considerable work has been performed on the development of the hardware and signal processing for these systems. Most of this work has been directed at monostatic systems or bistatic systems with closely spaced antennas. In this paper, we will investigate a multistatic system with eight different antenna spacings. We hypothesize that the multiple looks at a target from the variety of antenna spacings will make it easier to distinguish targets of interest from clutter. Other investigators have examined multistatic GPRs with varying degrees of suc- cess [2]–[4]. What makes this work different than the prior work is that measurements are made with the same system in three environments. This gives a range of environments to develop multistatic inversion algorithms. The targets are placed Manuscript received September 1, 2006; revised March 15, 2007. This work was supported in part by the U.S. Army Research Office under Contract DAAD19-02-1-0252. T. Counts was with the School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0250 USA. He is now with the Raytheon Company, Tucson, AZ 85706 USA. A. C. Gurbuz, W. R. Scott, Jr., and J. H. McClellan are with the School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0250 USA (e-mail: waymond.scott@ece.gatech.edu). K. Kim is with the School of Information and Mechatronics, Gwangju In- stitute of Science and Technology, Gwangju 500-712, Korea (e-mail: mkkim@ gist.ac.kr). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TGRS.2007.900677 in air for the first environment, which has a minimal uncertainty, and can be used to validate the performance of the algorithms in a known environment. The targets are buried in relatively homogeneous sand for the second environment; this scenario has more clutter and an unknown wave velocity that results in a more challenging inversion than the air environment. In the third environment, rocks are placed on the surface of the sand, which adds significant clutter, making the inversion much more difficult. Generally, clutter is the limiting factor for the performance of a GPR; therefore, the variety of clutter in the experiments should be useful in evaluating data processing algorithms. A large variety of signal processing algorithms have been in- vestigated. These can be divided into two categories: detection algorithms that attempt to detect specific sets of targets such as buried land mines or tunnels [5]–[8], and imaging algorithms that attempt to make an image of an underground region [2], [3], [9]–[13]. Back projection algorithms have proven to be rel- atively robust in imaging this type of data [2]. In this paper, we will present the results of our experiments using a frequency- domain imaging algorithm that is based on beamforming, which compensates for the phase response of the antennas as a function of both angle and frequency. The data are pub- licly available online at http://users.ece.gatech.edu/~wrscott/ in Matlab format files. We believe that the data set can be a good starting point to prove the efficiency of multistatic inversion algorithms. This paper is organized as follows. In Section II, the config- uration of the multistatic GPR system and its data collection procedure are described. In Section III, the imaging algorithm is detailed. In Section IV, the experiments performed on tar- gets suspended in air are described. Because air is a known and homogeneous medium, the inversion of the data obtained from the targets in air involves less uncertainties than the data obtained from the buried targets. Thus, the data described in Section IV can be used to calibrate the data obtained from the buried targets or to demonstrate the performance of multistatic inversion algorithms in a homogeneous medium. In Section V, the experiments performed on targets, which are buried in clean sand and in cluttered sand, are described. II. MULTISTATIC GPR A. System Description The multistatic GPR in this paper consists of a linear array of resistive-vee antennas, a microwave switch matrix, a vector network analyzer (Agilent 8720D), and a 3-D positioner, all under computer control (Fig. 1). The resistive-vee antenna is 0196-2892/$25.00 © 2007 IEEE