Frequency based signal processing technique for pulse modulation ground penetrating radar system

ABSTRACT


INTRODUCTION
Ground penetrating radar (GPR) is used in detection or locating an embedded object underground [1]- [3]. The system uses the application of electromagnetic waves propagation characteristics in locating and estimating underground objects [4]. Time domain, frequency domain and spatial domain are referring to the category of operating domain in the GPR system [3]. However, based on this operating domain, there are only two types of operation domain always applied in the GPR system, time and frequency. In general, the operating system of GPR in the time domain uses techniques of amplitude modulation such as pulse modulation (PM). Whereas, the GPR system operating in the frequency domain always uses the stepped frequency continuous wave (SFCW) technique [5], [6]. Moreover, these types of GPR system had gotten a lot of attention from researchers because the system involves only one-dimensional data processing which makes it easier to be processed compared to the operating system of GPR in the spatial domain [7]- [9].
Referring to the PM GPR system, there are two types of signals involved which are information signal and carrier signal. The difference between the information signal and the carrier signal is that the information signal has a low-frequency signal using the sinusoidal type while the carrier signal has a highfrequency signal using Gaussian pulse signal [9]- [11]. In contrast, the SFCW GPR system uses only one type of signal that is sinusoidal with variable frequency values. The signal processing technique of this system is usually done in the frequency domain where the receiving signal of the GPR system will be converted from the time domain using the fast fourier transform (FFT) algorithm [12]- [15]. Basically, a GPR system contains a transmitter/receiver system and antenna, as shown in Figure 1. Based on Figure 1, the pulse generator acts as a transmitter system while the Oscilloscope acts as the receiver system. As the PM GPR system typically operated in the time domain, this study will propose the development of the signal processing technique of the system in the frequency domain [16]- [18]. The proposed system simulation model of PM GPR is performed using CST microwave software studio while signal processing of the system will be done using MATLAB software. Analysis of the PM GPR system performance operated in the frequency domain will be done based on the GPR radargram produced.

PM GPR SYSTEM SIMULATION DEVELOPMENT
The design of the dipole antenna with operating frequency from 70 MHz until 80 MHz, having cooper material in developing PM GPR system simulation using call setup time (CST) software is as shown in Figure 2 [11]. Referring to the SFCW GPR system, receiving signal of the system needs to be transformed from the time domain into the frequency domain by using the discrete fourier transform (DFT) technique. The signal transformation of the discrete frequency domain S(k) is based on (1): Based on (1), n is sample value while N is sample number, thus the calculation of the transformation required N 2 multiplication. Computationally efficient algorithms to compute the DFT using only N log2 N multiplication can be done using fast fourier transform (FFT). The algorithm is considered a computational efficiency algorithm which has an ability to produce high accuracy in signal levels [19]- [21]. As the FFT analysis is a Fourier transform that converts signals from the time domain to the frequency domain [22]- [24], based on this algorithm, the important value of the signal in time will be stored, and exponent components will be calculated repeatedly to complete the history of time [25], [26]. The transformation signal calculated using this algorithm can be classified into two parameters which are magnitude and phase calculation.
The dipole antenna design and its parameters can be referred to Figure 2 and Table 1 respectively. Figure 3 shows an overview of the GPR system simulation is setup consists of a Dipole antenna on the top of the model size of the ground object was made as rectangular form using dry sandy soil material with a diameter of 3000 mm Length, 3000 mm Width and 2000 mm height. Meanwhile, an iron bar with a diameter of 800 mm Length, 800 mm width and 400 mm height was buried inside a ground object. The detailed information on the PM GPR system simulation model design in CST software and its scanning methods can be referred to the paper [11].

Signal processing of PM GPR system based on fast fourier transform (FFT)
The proposed signal processing technique applied in this study using the FFT algorithm consists of two calculation values which are magnitude and phase. Based on the value of this calculation, the GPR radargram is produced. The processing algorithm in the proposed technique will need a reference input signal. This signal is actually an input signal of the simulation system of the GPR developed in this study. As for the real application, this input signal as a reference signal for the algorithm can be achieved from the signal generator of the transmitter system which will be used to feed the transmitter antenna.  Figure 4 represents the flow chart of the signal processing algorithm of the PM GPR system based on the magnitude calculation in the frequency domain using the FFT technique. Referring to this flowchart, the algorithm starts by uploading input and output signals obtained from the PM GPR system simulation in CST software into MATLAB software. After that, the signal readings in the time domain will be converted into the frequency domain using the FFT method. Next, the ratio of the magnitude values from the FFT transformation between output and input signals is calculated. The construction of the PM GPR radargram is done by arranging this calculated value in the form of a matrix referring to the column of each scanning position in the simulation system.

Signal processing technique for PM GPR system based on phase of FFT calculation
The simplified flow chart for PM GPR radargram based on the phase of FFT calculation as depicted in Figure 5. Based on the flow chart, the phase difference of the FFT calculation between the output and the input signals is calculated to track the reflections in the presence of an embedded object in the radargram image of the proposed PM GPR system. Comparing this flow chart with the flow chart of Figure 4, the difference is as can be seen in the third block where this flow chart uses phase value while Figure 4 uses magnitude value.

Simulation result based on magnitude of FFT calculation
The analysis of the PM GPR system based on the magnitude of FFT calculation capabilities developed in this study is based on the image of the GPR radargram generated using MATLAB software. In this study, simulation of PM GPR system in scanning sandy soil area contains the buried object of iron at depths of 10 mm, 100 mm, 500 mm, 900 mm and 1000 mm was made. As for the reference to assist interpretation in this study, the simulation of the PM GPR system without the buried object of iron has also been created. Referring to Figure 6(a) to Figure 6(g), the image of the GPR radargram generated from the propose PM GPR system based on the magnitude value, the buried object of the iron was detected at positions between 1 and 11 of frequency samples in all simulations. Referring to the scanning position, the object can be estimated to embed at the scanning position of 8 until 10. However, referring to the radargram pattern, the correct depth position of the embedded object cannot be clearly estimated. Referring to the colour pattern of these GPR radargrams, the simulation without embedded iron object in Figure 6(a) seems to have a spreading yellow color at all scanning positions. On the other hand, the spreading of yellow color in the GPR radargram of Figure 6(b) to Figure 6(g) seems to spread on the scanning position of 5 to 11. This indicator can be used to interpret that there is an embedded iron object in these Figures. As referring to the frequency samples in the GPR radargrams, the spreading of yellow color in the image has been reduced as the position of the embedded iron object in the simulation increases. This indicator could be further analysed in order to correctly estimate the depth position of the embedded iron object in the simulation.

Simulation result based on phase of FFT calculation
The image of the GPR radargram generated from the PM GPR system simulation processed using phase values of FFT calculation can be seen in Figure 7(a) to Figure 7(g). The position of the iron object only can be estimated at the depth of 900 mm and 1000 mm. This is referring to the color pattern of the GPR radargram produced. Referring to the yellow color in the radargram of Figure 7(a), the spreading is at a scanning position of 4 until 12. The orange color in this figure seems to have some correlation with the yellow color which can be assumed as some part of the underground which does not contain any embedded object.
Based on Figure 7(b) until Figure 7(e), the spreading of yellow color is almost the same which is at the scanning position of 3 until 15. The spreading pattern of the yellow color is quite wide which seems almost same as the spreading of the yellow color in Figure 7(a). In contrast, the yellow color spreading in Figure 7(f) and Figure 7(g) seems to focus on the scanning position from 6 until 12 and from 7 until 11 respectively. This can be used as an indicator to estimate the embedded iron object. Based on these figures, the depth of the iron object in the PM GPR radargram is said to have been embedded at frequency samples of 4 and 5. The object seems to appear at a scanning position of 8 until 10.

CONCLUSION
The system of PM GPR has been developed and simulated using signal processing techniques in the frequency domain successfully. From the images of the GPR radargram that has been generated, the position of the embedded iron object in the simulation has been successfully detected for all depths using the magnitude of FFT calculation. However, by using the phase of FFT calculation the embedded iron object can only be detected at a depth of 900 mm and 1000 mm. The position of the embedded object can also be predicted correctly based on the scanning position. The estimated of the embedded object position based on depth need to be further studied especially using the magnitude of the FFT calculation technique. As this study is focusing on the iron object, in order to enhance the capabilities of the signal processing techniques developed, the PM GPR system simulation can be designed using other materials as ground and embedded objects.