Comparison of Impulse, Stepped-Frequency and Chirp Signals in Terms of Measurement Efficiency for Ground-Penetrating Radar Systems

Ground Penetrating Radar (GPR) is a non-invasive soil investigation tool that uses electromagnetic waves to probe the subsurface and determine the distribution of the electrical permittivity and conductivity. Traditional systems use impulse radar, which transmits short duration waveforms. Advances in high frequency electronics, such as high sampling rate converters, allow greater flexibility in waveform design and signal processing. This work focuses on analysing impulse, stepped frequency continuous wave (SFCW) and chirp waveforms to improve the signal-to-noise ratio (SNR) and measurement speed of GPR systems. Impulse radars use high peak to average power ratio (PAPR) waveforms. To increase SNR for a given maximum voltage, multiple measurements are averaged (stacked). SFCW GPR systems offer higher average power, but require further processing due to “ringing” introduced after frequency to time conversion. In some cases, chirp waveforms have been successfully implemented in GPR systems. Both SFCW and chirp waveforms can be designed to have the desired frequency spectrum, reducing ringing in the reconstructed time domain signal. In a model-based approach, we compared a Ricker wavelet pulse with a center frequency of 600 MHz, an SFCW and a chirp signal with non-linear frequency modulation. The spectrum of the SFCW waveform was shaped by varying the duration of transmission (dwell time) of each frequency. The signals were fed through a gprMax-simulation with an on-ground GPR setup. The modelled soil consists of two homogeneous layers. The upper layer has a conductivity of 10 mS/m and a relative permittivity of 7, while the lower layer has a conductivity of 20 mS/m and a relative permittivity of 20. The boundary between the two layers is at a depth of 30 cm below the antennas. The received SFCW and chirp signals were finally transformed into time domain for comparison with the impulse signal. In the experiment described, chirp and SFCW techniques provide 10 dB increase in SNR in comparison to the impulse technique. The analysis confirms that shaping the spectrum by non-linear frequency modulation prior to transmission reduces ringing in the reconstructed time domain signals. For the described setup, switching from impulse to frequency modulated waveform based techniques is beneficial for systems requiring a high signal to noise ratio for a limited transmission voltage. The efficiency of the waveforms can be increased by using a variable transmission time that depends on the frequency. The choice between chirp and SFCW depends on hardware and measurement time requirements.