Sie sind auf Seite 1von 3

Pulse versus Stepped Frequency Continuous Wave modulation for Ground

Penetrating Radar

A. Langman and M.R. Inggs


Radar Remote Sensing Group, Department of Electrical Engineering
University of Cape Town,Private Bag, Rondebosch 7701, South Africa
Email: mikings@eng.uct.ac.za Web: http://rrsg.ee.uct.ac.za
Tel: +27 21 650 2799 Fax: +27 21 650 3465

Abstract There has often been debate concerning the relative mer- only means of inversion and information extraction [3].
tis of pulsed (carrier-less) and Stepped Frequency Continuous Wave
(SFCW) radars used in Ground Penetrating Radar (GPR) applications. II. R ESOLUTION AND SPECTRAL SHAPE
It is sometimes claimed that SFCW has poor time sidelobe performance.
In this paper we tease out the definiton of resolution and bandwidth, and An important concept that appears frequently in radar the-
show that both systems have equivalent resolution and time sidelobe per- ory is the concept of resolution. Wehner [4] defines the range
formance if the spectral shapes are properly defined.
resolution in terms of the ability to resolve point targets sepa-
rated in slant range (along the line of sight) of the radar. If we
I. I NTRODUCTION assume that the received waveform has a bandwidth,  (see
The object of any Ground Penetrating Radar system (GPR) Figure 1a), the corresponding impulse response for the wave-
is to provide information that relates to the location and phys-  
form would have a 
 
form. The resulted time resolution
ical properties of targets that are to be detected. Hence, when
choosing a GPR modulation technique, it is essential to choose
   is given by the width of the mainlobe at the point
below the peak (a convenient criterion chosen to simplify
one that yields the most information for the specific applica- the formula). Hence using this minimum distance we would
tion. In some cases one might qualify the latter statement with be able to resolve the impulse response from two targets - as is
requirements for speed of operation, or regulatory control of shown in Figure 1(b)
radiated spectrum [2]. It is important to note that in most GPR scenarios we are in-
It is also important to choose the correct means for compar- terested in the impulse response of the target, since time side-
ing the performance of the waveforms against carefully spec- lobes can swamp, weak, close-by targets. The corresponding
ified operational requirements. In this paper, we will give an targets bandwidth (target reflectivity spectrum) may not nec-
example where we will compare impulse to stepped frequency essarily be rectangular i.e. with corresponding time domain
continuous wave GPR. The comparison criteria will be based response.
on both spectral shape and time domain waveform and how We use the term "resolution" to provide some upper limit
these relate to radar resolution and range sidelobes. to the resolving power of the waveform - but this is only an
The paper begins by providing a basic discription of impulse approximation, with the strong assumption that our target is a
and SFCW GPR. It then continues by defining the concept of point scatterer (with corresponding "infinite" reflectivity spec-
resolution and the impulse response of the radar. Finally the trum [5],[3]). When comparing resolutions of difference wave-
waveforms are compared on the basis of equivalent resolution forms, one must be careful of how one defines the bandwith
(bandwidth) and sidelobes, and they are seen to be the same. of the system. The concept of resolution can also be applied
Most commerical GPR systems use an impulse waveform, to cross-range resolution, depth resolution and even frequency
where the radar transmits a very narrow pulse (less than a few resolution. In this paper we will limit our discussion to depth
nano-seconds), with a large peak power, at a constant pulse profiles.
repetition frequency (PRF). This signal is then applied to the Recent papers have looked at the comparison between vari-
terminals of a broad-band antenna. The time delayed received ous waveforms, in particular impulse and SFCW types [1]. An
waveform is then sampled, usually integrated and then dis- important criterion used in these papers is the level of side-
played. We must also bear in mind [2] that the resulting spec- lobes. In the previous section we saw that the radar bandwith
trum consists of lines, spaced at the radar repetition rate. related not only to the radar resolution but also to the side-lobe
The stepped frequency continuous wave waveform is im- level, when the spectral shape is considered.

separated in frequency by

plemented by transmitting a number of single frequency tones
hertz. At each frequency
For a SFCW waveform, the shape of the radiated spectrum
is typically limitted (truncated) by the frequency range of the
the amplitude and phase of the received data is sampled and frequency generator - hence the spectral shape is considered
recorded. Most implementations then use the Inverse Discrete to be rectangular. It is also not continuous, due to the discrete
Fourier Transform (IDFT) to transform the data into the spatial step nature of the modulation.
domain. This yields a synthesised pulse. Clearly this is not the If we now consider the case of an impulse waveform. The

0-7803-7031-7/01/$17.00
0-7803-7031-7/01/$10.00(C)
(C)2001
2001IEEE
IEEE 1533

0-7803-7033-3/01/$10.00 (C) 2001 IEEE


(a) Frequency/Time transform for a pulse waveform Figure 2. Spectra of a Gaussian pulse and conventional SFCW
waveform.

(b) Resolution of two pulses

Figure 1. Relationship between bandwidth, impulse response


and resolution.

bandwidth of the resulting transmitted pulse is typically trun- Figure 3. Impulse response of a Gaussian pulse and SFCW
cated and shaped by the antenna response. The waveform waveform.
bandwith is then defined (typically) as the width of the wave-
form spectrum at 3dBs below the peak response in the band-
width of the antenna. If we wish to compare the resolution and dow to the frequency domain data ensuring that the band-
sidelobe level for an impulse and SFCW waveform (using the width points are the same as that of an equivalent impulse radar
above definitions for pulse bandwith), one would be comparing waveform.
waveforms with spectra as shown in Figure 2. Here we show
a Gaussian shaped impulse (which has Gaussian shaped spec- III. C ONCLUSIONS
trum) with a SFCW operating over the same 3dB bandwidth.
The resulting impulse response for the two waveforms is When comparing the impulse response of two waveforms, it
shown in Figure 3, and the SFCW impulse response is indeed is important to take care in how one defines the spectra. The
dismal. bandwidth alone is not sufficient. The GPR literature is littered
The sidelobes of the impulse response of the SFCW can be with papers comparing the impulse response of the baseband
reduced by weighting the spectrum. In fact, if we weight the pulse and SFCW modulation techniques, however most miss
SFCW spectra with a window that has the same envelop as the the simple point that theoretically the waveforms can be made
envelope of the spectrum of the impulse waveform and we en- identical. The difference lie more in the implementation of the
sure that both spectra cover the same carrier frequency span, hardware and the algorithms to extract the information from
then both waveforms will have the same impulse response. the radar waveforms.
There is thus no difference between the waveforms, based on With the growing interest by regulatory authorities [2] in the
the restricted criteria of resolution and side lobes. The results EMC of GPR, the impulse, repetive nature of pulsed systems
will be the same as shown in Figure 3, except the synthetic may not be as attractive as the equivalent SFCW system, which
range response is now generated by applying a Guassian win- has possibilities of sophisticated spectral control.

0-7803-7031-7/01/$17.00 (C) 2001 IEEE 1534

0-7803-7033-3/01/$10.00 (C) 2001 IEEE


R EFERENCES

[1] Noon, D.A. & Stickley, G.F., The Relationships between Ring-down,
Range-sidelobes and Quality Factor of GPR Antennas, 7th International
Conference on Ground Penetrating Radar (GPR98), Lawrence, Kansas,
pp. 63-68, May 27-30, 1998.
[2] Gary Oeloft, http://www.g-p-r.com/regulato.htm
[3] R.T. Lord, Aspects of Stepped-Frequency Processing for Low-Frequency
SAR Systems, PhD Thesis, University of Cape Town, Rondebosch, South
Africa, 2000.
[4] D.R. Wehner, High-Resolution Radar Artech House, Dedham, Ma, USA,
1995.
[5] A.J. Wilkinson, R.T. Lord and M.R. Inggs, Stepped-Frequency Process-
ing by Reconstruction of Target Reflectivity Spectrum, IEEE Proc. of the
South African Symp. on Communications and Signal Processing, COM-
SIG98, Cape Town, South Africa, pp. 101104, September 1998.

0-7803-7031-7/01/$17.00 (C) 2001 IEEE 1535

0-7803-7033-3/01/$10.00 (C) 2001 IEEE

Das könnte Ihnen auch gefallen