UWB GPR Receiver Detects Waterpipe Leaks

Nov. 12, 2008
This novel ground-penetrating-radar-system takes advantage of ultrawideband signals to 2 GHz to probe beneath the surface of the Earth in search of leaky and damaged water-supply pipes.

Ground-penetrating-radar (GPR) systems based on ultrawideband (UWB) generation and reception of reflected signals are often associated with military applications such as land-mine detection. But because of aging waterpipe infrastructure in many major cities within the European Union (EU), a UWB-based GPR system also has application for detecting leaks in underground waterpipes. The cost of replacing these aging waterpipe systems, if left to deteriorate beyond a state of repair, is prohibitive for many EU communities. A more practical alternative is the detection of leaks in infrastructure before major damage has occurred, in order to effect repairs and rehabilitate the waterpipe system. For that reason, a UWB-based GPR system is attractive for monitoring and detecting problems in underground pipes.

The authors report on work that was part of their proposal/ contract (No. 036887) with the Waterpipe European Commission research project, "Integrated High Resolution Imaging Ground Penetrating Radar and Decision Support System for Waterpipe Line Rehabilitation." The proposed system operates in the frequency range from 100 MHz to 10 GHz and includes active UWB antennas, low-noise, variablegain amplifiers (VGAs), high-speed analog-to-digital converters (ADCs), and high-output-power configurable pulsed UWB signals.

Because of the expense of replacing entire waterpipe systems, resources are being allocated throughout the EU for waterpipe rehabilitation. With this emphasis on sustainable management, it has been necessary to develop risk-based approaches for the rehabilitation management of the water-supply network. Rehabilitation decisions should be based on inspection and evaluation of pipeline conditions, but this implies the capability of performing such inspections in a cost-effective and timely manner. In some cases, utility companies cannot locate a number of the old water supply pipes and current inspection technologies typically do not provide the needed detailed information on pipeline damage.

The authors present an approach for the effective inspection of underground waterpipes by means of a high-resolution ground penetration imaging radar (GPIR) system. The system is capable of detecting water pipes of different dimensions, of detecting leaks and damage in the pipes, and providing high-resolution images of the damaged areas. The proposed GPIR system can detect waterpipes made of various different types of materials, and detect leaks and damage in waterpipes of those different types of materials. It has a ground penetration depth to 2 m, and the imaging resolution of the damaged pipe is 5 cm or better. The time to perform a survey, that is, complete the detection and imaging of a section of waterpipe, is 10 s/m along the pipe axis. Thus, the survey time for a 1-km section of waterpipe will be about 3 h.

For transportability and mobility, systems the GPIR is mounted on a four-wheel vehicle. For programmability and signal processing, a laptop computer will also be on the vehicle, while the antenna system and transceiver analog/ digital signal processing units will be placed in the front side of the vehicle. The antenna system consists of 10 UWB bow-tie patch antennas with periodic reception. The main subunits of the proposed GPIR system are

  • a reconfigurable short-pulse transmitter;
  • a fixed UWB transmit antenna;
  • a scanning UWB receive antenna;
  • a receiver front end;
  • a digital correlator receiver; and a laptop personal computer to perform signal processing and image reconstruction.

A pulsed signal with zero baseband spectral components is generated by a digitally controlled generator. The generator is capable of creating a variety of UWB signals, such as one-period sinusoid or two opposite and inverted triangular pulses. The short-pulse generator is designed to produce output amplitudes on the order of 10 V and will drive a fixed-position UWB antenna (Fig. 1). In order to achieve synchronization between the receiver and transmitter units, a sample of the reference signal is fed to the correlator receiver.

The receiver chain consists of a mechanically scanned antenna measuring 1 x 1 m. The pulse repetition frequency (PRF) is selected as 10 MHz in order to achieve short scanning times for all 10 UWB antennas, including averaging of the received signals. Received signals are amplified and then correlated with the reference pulse signal to obtain the envelope of the signals reflected by the ground medium. The use of UWB signals provides extensive information for the imaging of a wide range of underground target media. Each receiver scanning antenna collects a number of reflected signal waveforms for the processing unit. The reconstruction algorithm for the complex dielectric constant of the underground medium (two real numbers for each frequency spectral component) is obtained by means of a time-domain version of the Method of Auxiliary Sources, developed previously.1 The complex dielectric permittivity of the underground target, the waterpipe, is reconstructed by means of solving an inverse scattering problem.

The GPIR system consists of several subsystems, including:

  • a transmitter unit (including the transmit antenna);
  • a receiver front end (analog) and the scanning receive antenna;
  • a signal digitizer (ADC);
  • a laptop computer and image display; and a structure for mounting the GPIR on a vehicle for mobility.

The transmitter unit, including the transmit antenna, employs an UWB signal generator to produce very short pulse width signals (500 to 2000 ps). The peak pulse amplitude of these signals is on the order of 10 V. Pulsed signals are sent to the antenna, which is a transverse electromagnetic (TEM) flared horn antenna, for transmission. While the receiving antenna is a scanning type, the transmit antenna is fixed. The transmitted pulse rate is 10 MHz, driven by a temperature-compensated crystal oscillator (TCXO). The use of the TEM horn antenna ensures wideband impedance matching of the antenna to the solid-state pulse generator.

Table 1 shows the specifications for the high-voltage, short-pulse generator. It consists of a crystal oscillator at 10 MHz that sets the system PRF with minimum jitter. Clock signals are channeled through a single-pole, fourthrow (SP4T) switch to control four parallel step-recovery- diode-based subsystems to produce short pulses with pulse widths ranging from 0.5 to 2.0 ns. The pulses have amplitude of typically 1 V. Following this, a broadband, high-power amplifier is used to boost these short pulsed signals to an amplitude of at least 10 V. Inputs signals to the high-power amplifier are selected by means of a SP4T switch (Fig. 2).

Figure 3 shows the pulser system, fabricated on RO4350 printed-circuit-board (PCB) material with a thickness of 0.5 mm from Rogers Corp. (www.rogerscorp.com). The PCB material, which has a dielectric constant of 3.48, is a glass-reinforced hydrocarbon/ceramic laminate with loss. The pulse tested was constructed, simulated, and fully tested. 2 The prototype unit featured the integrated high-power, output-stage amplifier.

Figure 4 shows the TEM horn transmit antenna.3-7 Computer-aided-engineering (CAE) simulations were also conducted on the TEM horn antenna in order to model its return-loss performance and radiation pattern, and the results are shown in Fig. 5. The simulated results indicate that the antenna's directivity is about 9 dBi and the mean return loss (S11) is better than 8 dB.

The receiver front end consists of a broadband low-noise amplifier (LNA) capable of operating from 100 MHz to 2 GHz and an anti-aliasing filter. The LNA provides maximum gain of 40 dB, which can be throttled down to 0 dB by means of analog control. The function of the LNA is to boost received echo pulse signals reflected by the underground structures to a level sufficient for digitization by a highspeed ADC. An active type antenna is used for the receive antenna, in order to provide broadband coverage in a relatively small form factor (the antenna occupies less than 10 x 5 cm in size). The LNA is based on a high-electronmobility- transistor (HEMT) device that is directly matched to the receive antenna. A patch bow-tie antenna is used as the receive antenna. A total of 10 similar antennas are set parallel to the Earth's surface and approximately 20 cm above the ground. They are scanned mechanically through the use of an electromechanical single-pole, ten-throw (SP10T) switch. Raster-type scanning is used to scan a 1 x 2 m horizontal area by means of the Cartesian coordinate system. Table 2 provides a summary of the receiver's basic performance specifications.

Continue to page 2

Page Title

Figure 6 depicts the receiver architecture with 10 active antennas serving to feed 10 input antennas and bandpass filters (BPFs) along with the VGAs. The bandpass filters span 100 MHz to 2 GHz in order to bound the input frequency spectrum of the echo signals from the ground. The gains of the 10 VGA channels are independently controlled by means of the system's laptop computer. The SP10T switch routes the incoming echo signals from the VGAs to the highspeed ADC for further processing. The SP10T switch is also controlled by the laptop computer. The physical layout of one of the 10 receivers in the signal-processing chain is shown in Fig. 6. The receivers were fabricated on a base layer of the 0.5-mm-thick Rogers' RO4350 laminate material, and microstrip technology was used extensively in the design of the receiver. Each separate receiver sub-unit is housed in an aluminum enclosure to provide proper shielding and guard the receiver against reception of unwanted signals.

Designing and building a system suitable for the WATERPIPE program requires that it meet a set of rigorous performance requirements, since the final demonstration is not a purely laboratory-based experiment. Series of outdoor measurements in real conditions and at various terrains have to be taken during the design and implementation whole procedures. The digital portion of the system must be fast, robust, and expandable while also meeting requirements for portability and compactness. One issue dealing with portability, for example, is the potential lack of adequate and efficient power sources in the test environment. As a result, the system should be designed with low power consumption, and power-saving digital circuitry, but with the performance and expandability to meet present and future system requirements.

The digital portion of the system must provide digitization of the RF signals presented to it, and additional signal processing using innovative algorithms to finally yield a three-dimensional (3D) representation of the waterpipe structure at a scanned area of interest. The heart of the overall digital system is a high-speed dataacquisition and control subsystem. It is combined with an Intel. x86 processor- based platform (a laptop computer) that is responsible for the post processing and analysis required to create a 3D image of the waterpipe.

The digital system (Fig. 7) was logically subdivided into two parts, consisting of a data acquisition and control module and the laptop computer. The data acquisition and control module must perform sampling at rates of at least 1 GSamples/s, properly buffering and transferring valid data to the second part of the digital system, the laptop computer. In order to achieve adequate signal-to-noise ratio (SNR) for proper application of the algorithms during the next phase of signal processing, the data acquisition and control module must deliver digital resolution of at least 8 b.

A properly configured laptop computer was used as the second part of the digital subsystem, to perform high-speed post-processing and generate a 3D visual output. The laptop computer also offers portability and low-power operation and, with the proper application software, ready communication with the data acquisition and control module. The laptop is equipped with an Intel Core 2 Duo, 1.8-GHz microprocessor with 2 GB of random-access memory (RAM) and a 120-GB hard-disk drive (HDD). The operating system is Microsoft Windows Vista Premium running with Version 8.1 of the LabView mathematical and signal-processing software from National Instruments (www.ni.com).8

The laptop computer provides control of the data acquisition system, the SP10T and SP4T switches in the RF section via a Universal Serial Bus (USB) to TTL input/output (I/O) interface, the gain control of the VGAs via an interface from USB to 16 DAC channels at 16-b resolution to provide analog control signals to the VGAs, and the stepper motor for the scanning receive antenna.

For image reconstruction, the laptop computer runs a virtual instrument (VI) program developed in LabView that executes the following algorithm:

1. Selection of the proper timing sequence from the pulser; calibration of amplifier gain; calibration of the stepper motor controller.

2. Movement of the scanner element to the correct position (scan line = 0) for the first scanning line.

3. Selection of the correct antenna on the scanning element (scan antenna = 0) for the first antenna.

4. Triggering, acquisition, an averaging of the 10 sampled signal snippets, at 16 samples per snippet.

5. Transferring the averaged snippet to the laptop computer and storing the data within a matrix.

6. Repeating step 3 until a full scan line is sampled.

7. Moving the scanner element to the next position and advancing the stepper motor the proper number of steps until the antenna element has advanced a total of 10 cm.

8. Repeating step 2 until 2 m are scanned, for a total of 20 scan lines with 10 signal snippets per line, with 16 samples of 8 b per snippet.

9. Exporting data to the digital-signalprocessing (DSP) and image-processing module.

10. Ending the program or repeating from the beginning as desired.

Each snippet contains 16 samples of data at 8 b per sample, derived from an averaging procedure. The number of samples and the number of averages taken can be adjusted: for a total of 400 snippets x 16 samples per snippet x 8 b per sample = a matrix with 51200 b results. Signal acquisition is performed by a PCI Signal Analyzer Platform card from Acquiris (now Agilent, www.agilent.com), specifically a model AP240 with two channels at 1 GSamples/s each at 8 b resolution. The analyzer also provides a single interleaved channel at 2 GSamples/s and 8-b resolution, combined with a special dithering mode that increases the dynamic range to 12 b.

The RF signal from the selected "read" antenna of the scanner element is driven via the SP10T

switch and sent to the RF input port of the AP240, while triggering is done via the external trigger of the card that is connected to a special signal generated directly from the pulser.

Extreme caution was taken in wire lengths used for triggering. Averaged data is stored in a matrix in the memory module of the PCI card and then read from the laptop at a much slower rate.

Outfitting the laptop to work with the PCI signal analyzer required some modification, since the AP240 required a laptop with an External PCI Expansion Chassis that could host the AP240 and provide the necessary bandwidth for the analzyer to transmit data to the laptop. The Magma PCI Expansion Chassis from Mobility Electronics proved to be the solution for this function, since it offers 2 PCI Slots and interconnection with the laptop through a PCMCIA card, effective drivers, and sufficient data transfer rate.

The GPIR sub-units were mounted on a small vehicle to achieve mobile operation (Fig. 8). This required the use of a special mounting structure to achieve simple and reliable operation. The receiver system can scroll in the x direction by means of its stepper motors run under laptop computer control, and the vehicle provides independent suspension on each wheel to adapt to rugged terrain when testing for waterpipe leaks.

In operation, the system was used with 500/2000-ps pulse width signals, shown on a commercial digitizing oscilloscope from Agilent Technologies. The scope screen shows signals with sharp rise and fall times, due to the performance of the step-recovery diodes used in generating the shortpulsed signals. The diodes are well matched to their amplifiers, with high suppression of any reverse signals. In another experiment, 500-ps pulses were transmitted by the pulser and its antenna and received at a 1-m distance by the receiver units without any processing. The display shows that the pulse width of these signals has not been altered but the signal reflections have increased because of some antenna mismatch, especially for frequencies below 400 MHz. This is not seen as a problem since the effects can be removed through processing afterwards. Work continues on the algorithms used in this system to recreate realistic 3D images of underground waterpipes.


1. R. Zaridze, G. Bit Babik, K. Tavarashvili, D. Economou and N. K. Uzunoglu, "Wave Field Singularity Aspects in Large Size Scattered and Inverse Problems," IEEE Transactions on Antennas & Propagation, Vol. 50, No. 1, 2002, pp. 50 58.

2. Agilent Technologies, Advanced Design System 2006 manuals (www.agilent.com).

3. Xialong Liu et al., "Design and Performance of TEM Horn Antenna with Low Frequency Compensation," Asia Pacific Conference on Environmental Electromagnetics, CEEM' 2003.

4. Picosecond Pulse Labs, Application Note AN-14a, "UWB Signal Sources, Antennas and Propagation," August 2003 (www.picosecond.com).

5. A. G. Yarovoy et al., "GPR Antenna Measurements in Time Domain," unpublished.

6. A. G. Yarovoy et al., "Ground Penetrating Impulse Radar For Landmine Detection," unpublished.

7. Z. Zhan and Y. H. Lee, "Ultra Wideband TEM Horn Design Using Multi Objective Evolutionary Algorithm," IEEE ATT 2005.

8. National Instruments, Labview 8.1 Manual (www.ni.com).

Sponsored Recommendations

Getting Started with Python for VNA Automation

April 19, 2024
The video goes through the steps for starting to use Python and SCPI commands to automate Copper Mountain Technologies VNAs. The process of downloading and installing Python IDC...

Can I Use the VNA Software Without an Instrument?

April 19, 2024
Our VNA software application offers a demo mode feature, which does not require a physical VNA to use. Demo mode is easy to access and allows you to simulate the use of various...

Introduction to Copper Mountain Technologies' Multiport VNA

April 19, 2024
Modern RF applications are constantly evolving and demand increasingly sophisticated test instrumentation, perfect for a multiport VNA.

Automating Vector Network Analyzer Measurements

April 19, 2024
Copper Mountain Technology VNAs can be automated by using either of two interfaces: a COM (also known as ActiveX) interface, or a TCP (Transmission Control Protocol) socket interface...