Development of a Linear Frequency Modulation-Based Ground-Penetrating Radar System for Crevasse Detection

Abstract

For the purpose of ensuring safe exploration in Antarctica, a ground penetrating radar (GPR) has been developed with the primary objective of detecting hidden crevasses. The previously GPRs predominantly employed impulse methods, which, however, concentrated the energy of the signal in a short duration, potentially causing saturation of the transmitter amplifier due to the high signal power required for acquiring echo signals from deep targets. In contrast, linear frequency modulation (LFM) method disperses the signal power in the time domain, thereby allowing the prevention of transmitter amplifier saturation by employing lower amplitudes, assuming the same signal energy. Thus, we developed LFM-based GPR. The operation frequency range was from 0.5 GHz to 2.0 GHz. The developed GPR comprises a system control unit with field programmable gate array (FPGA), power circuit, radio frequency (RF) transceiver, data converter and a control computer. To achieve high-quality crevasse imaging, two types of noise reduction circuits were designed. The voltage-controlled oscillator (VCO) noise reduction circuit was designed to remove noise from the VCO output signal and stabilize the VCO drive voltage. Using a low dropout (LDO) circuit, the unstable VCO drive voltage was i stabilized, and a two-stage filtersection was designed to eliminate noise from the VCO output signal. Subsequent performance verification involved comparative experiments before and after the adoption of the circuit, demonstrating improvements in VCO signal quality. The I/Q Mixer noise reduction circuit was designed to stabilize the unstable drive voltage of the I/Q mixer. A three-stage Cascaded-LDO circuit was configured, and experiments confirmed improvements in power noise within the I/Q mixer. To validate the final design of the ground-penetrating radar, range accuracy and Bscan experiments experiment were conducted. The range accuracy experiment aimed to determine if frequency differences corresponding to a 10cm change in target distance could be distinguished, and the experiment successfully observed frequency differences corresponding to a 10 cm change. The B-scan experiments, conducted in an environment resembling crevasse characteristics, aimed to distinguish targets with range-walk and two layers with different permittivity. Experimental results confirmed the range-walk of targets in air and sand and obtained subsurface profile data distinguishing layers of Styrofoam and sand with different permittivity.