Low-frequency ground-penetrating radar (GPR) systems operating below 200 MHz are crucial for deep subsurface investigations, targeting large-scale features such as cavities, bedrock structures, and riverbeds. The integration of these systems with UAV drone platforms promises unparalleled survey efficiency and access to challenging terrain. However, transitioning from ground-coupled to air-coupled operation introduces significant electromagnetic wave propagation losses, drastically reducing penetration depth. Since propagation loss in media is proportional to frequency, developing efficient air-coupled antennas capable of operating at very low frequencies, down to 20 MHz, becomes paramount for UAV drone-based deep sensing.
The antenna is arguably the most critical component of a GPR system, dictating its radiated power, receiving sensitivity, and ultimately, its probing depth. Among various antenna types, the bow-tie dipole is widely favored in GPR for its symmetric radiation pattern, broadband characteristics, and relatively high gain. A fundamental challenge arises at low frequencies: a standard half-wave bow-tie antenna for 20 MHz would be approximately 7.5 meters long, which is impractical for mounting on a UAV drone. Existing research on dipole antennas for GPR often focuses on impedance matching, back-cavity design, or resistive loading for bandwidth enhancement, with fewer studies dedicated specifically to structural miniaturization techniques that preserve low-frequency performance. This work addresses this gap by introducing a novel antenna topology and an efficient loading methodology tailored for compact, low-frequency, high-gain applications on UAV drone platforms.
The design philosophy stems from analyzing the current distribution on a linear dipole antenna. The current distribution on a thin, center-fed linear dipole of length \(2l\) can be approximated as a standing wave:
$$I(z) = I \sin[k(l – |z|)] \quad \text{for} \quad |z| \leq l$$
where \(k = 2\pi / \lambda\) is the phase constant, \(l\) is the single-arm length, and \(z\) is the position along the arm from the feed point. This equation reveals that the current magnitude is minimal near the open ends (tails) of the dipole arms, especially at the lowest operating frequency where the distribution approaches a quarter-wave sinusoid. This insight forms the theoretical basis for miniaturization: the sections of the radiating conductor carrying relatively small currents can be folded or meandered with minimal impact on the overall radiation efficiency, particularly at the low-frequency end. The proposed strategy involves replacing the low-current tail sections of a conventional bow-tie antenna with a compact, folded linear dipole structure, creating a Hybrid Folded Dipole Patch Antenna (HFDPA).
To counteract the inevitable gain reduction from shortening the antenna, dielectric loading is employed. The wavelength within a dielectric medium is given by:
$$\lambda_D = \frac{c}{f\sqrt{\varepsilon_r}}$$
where \(c\) is the speed of light, \(f\) is the frequency, and \(\varepsilon_r\) is the relative permittivity of the dielectric. By overlaying the antenna with a high-permittivity, low-loss dielectric substrate, the effective electrical length of the radiating structure is increased without extending its physical dimensions, thereby recovering some of the lost gain. This is essential for maintaining adequate signal strength for the UAV drone-mounted, air-coupled system.
For impulse GPR applications, minimizing late-time ringing caused by internal reflections within the antenna is critical to resolve shallow subsurface targets. Resistive loading is a standard technique to dampen these reflections and achieve an ultra-wideband (UWB) impedance match. The classical Wu-King distribution provides the foundation for distributed loading to approximate a traveling-wave current. The original formulation gives the distributed impedance per unit length \(z_i(z)\) as:
$$z_i(z) = \frac{15\psi}{h – |z|}$$
where \(\psi\) is a constant related to the antenna’s characteristic impedance, making direct application difficult. To derive a practical design formula, we discretize the loading into \(n\) lumped resistors placed at equal intervals along a single arm of length \(l\). The resistance value at the \(i\)-th position (\(R_i\)) can be expressed relative to the first resistor \(R_1\):
$$R_i|_{i=2 \sim n} = \frac{n}{n – i + 1} R_1$$
This formula allows for a systematic design approach starting from an initial value for \(R_1\). However, conventional application of this distribution often leads to high resistance values near the feed point (where current is high), resulting in excessive ohmic loss and low radiation efficiency.
To achieve efficient resistive loading, this work proposes a “Broadband Loading, Local Optimization” (BLLO) method. The core idea is to initially over-design the loading using a larger number of resistors \(n\) based on Eq. (3) to achieve a bandwidth significantly wider than the target. This “broadband loading” stage inherently results in smaller relative resistance values at high-current positions compared to a design targeting the exact bandwidth with fewer resistors. Subsequently, a “local optimization” is performed, tuning the resistor values to fine-tune the impedance match specifically within the desired target bandwidth (e.g., 20-140 MHz), often allowing for a further reduction in resistance values. This BLLO process effectively lowers the total parasitic power dissipation while maintaining excellent time-domain fidelity, a key requirement for the UAV drone‘s GPR payload.
The structural design of the HFDPA is driven by the platform constraints of a typical UAV drone, such as the DJI T60. To ensure aerodynamic stability and practical mounting, the antenna dimensions were set to 1800 mm × 200 mm × 1 mm. The radiating element is patterned on a thin FR-4 substrate. The antenna consists of two main sections integrated in series on each arm: a triangular bow-tie section (850 mm long, 200 mm wide at the base) responsible for the primary high-gain radiation, and a folded linear dipole section occupying the remaining 49.5 mm of length but providing an effective conductor length of nearly 1990 mm through strategic folding on both sides of the substrate. The feed gap is 1 mm. For dielectric loading, a layer of alumina ceramic (\(\varepsilon_r \approx 9.8\)) of the same planar dimensions is placed over the antenna.
To validate the design concept, a comparative simulation study was conducted using HFSS. Four antennas were modeled: a 7.5m Standard Bow-tie (S-Bowtie), a 3.75m Folded Bow-tie (F-Bowtie) from literature, the unloaded 1.8m HFDPA (H-Bowtie), and the dielectric-loaded 1.8m HFDPA (DH-Bowtie). Their simulated gain performance from 20 to 140 MHz is summarized below, demonstrating the effectiveness of the hybrid folding and dielectric loading approach.
| Antenna Type | Physical Length | Length (λ at 20 MHz) | Key Feature | Approx. Gain at 20 MHz (Simulated) |
|---|---|---|---|---|
| S-Bowtie | 7500 mm | 0.5 λ | Reference Standard | 2.3 dBi |
| F-Bowtie | 3750 mm | 0.25 λ | Full-arm Folding | -0.5 dBi |
| H-Bowtie | 1800 mm | 0.12 λ | Hybrid Folded Dipole | 0.2 dBi |
| DH-Bowtie | 1800 mm | 0.12 λ | Hybrid Folded + Dielectric Load | 1.8 dBi |
The results clearly show that the DH-Bowtie, with a physical length of only 0.12λ at 20 MHz, achieves a gain very close to the full-sized 0.5λ standard antenna, significantly outperforming other miniaturized counterparts. This confirms that selectively folding the low-current regions and applying dielectric loading is a highly effective strategy for low-frequency miniaturization on UAV drone platforms.
The BLLO resistive loading method was applied to the DH-Bowtie design. The antenna’s first resonant frequency was found at 61 MHz, corresponding to a quarter-wavelength of 1229.5 mm. Each arm was divided into 10 segments accordingly. Resistor values were initially calculated using Eq. (3) with \(n=9\) for broadband loading, then optimized for the 20-140 MHz band. A comparison of the loading resistance values and the calculated relative power dissipation (proportional to \(I_i^2 R_i\)) between a standard Wu-King distribution targeting the same bandwidth and the final BLLO-optimized distribution highlights the efficiency improvement.
| Loading Position (i) | Norm. Current (I_i) | Wu-King R (Ω) | Wu-King Loss (I_i²R) | BLLO Optimized R (Ω) | BLLO Loss (I_i²R) |
|---|---|---|---|---|---|
| 1 (Near Feed) | 0.99 | 15.0 | 14.7 | 5.6 | 5.5 |
| 2 | 0.95 | 16.8 | 15.2 | 15.0 | 13.5 |
| 3 | 0.89 | 19.2 | 15.2 | 5.6 | 4.4 |
| 4 | 0.81 | 22.5 | 14.8 | 24.0 | 15.8 |
| 5 | 0.71 | 27.0 | 13.6 | 15.0 | 7.6 |
| 6 | 0.59 | 33.8 | 11.8 | 36.0 | 12.5 |
| 7 | 0.45 | 45.0 | 9.1 | 33.0 | 6.7 |
| 8 | 0.31 | 67.5 | 6.5 | 80.0 | 7.7 |
| 9 (Near End) | 0.16 | 135.0 | 3.5 | 80.0 | 2.1 |
| Total (Arbitrary Units) | – | – | 104.3 | – | 75.7 |
The total estimated loading loss for the BLLO design is approximately 27% lower than for the standard Wu-King distribution. Assuming a baseline radiation efficiency of 50% for the Wu-King loaded antenna, the BLLO method improves the overall efficiency by a significant margin, making it highly suitable for power-conscious UAV drone applications.
Prototype antennas were fabricated and tested. The time-domain performance was evaluated by measuring the S21 parameter between two identical antennas in a bistatic configuration using a vector network analyzer and transforming the frequency-domain data to the time domain. The comparison among the unloaded H-Bowtie, the BLLO-loaded H-Bowtie, and the BLLO-loaded DH-Bowtie clearly demonstrated the effectiveness of the BLLO method in suppressing late-time ringing. The first peak (direct coupling) remained strong, while subsequent peaks (ground reflections and antenna ringing) were markedly reduced. The dielectric-loaded version also showed an increased time delay for later arrivals, confirming its increased electrical length.
Further validation was conducted through a practical GPR survey in an underground parking garage, comparing the BLLO-loaded DH-Bowtie antenna against a commercial, wideband conformal bow-tie antenna. Both systems produced clear subsurface images with distinct reflections from overhead walkways. Critically, the DH-Bowtie antenna showed no visible degradation in image clarity or excessive ringing artifacts compared to the commercial antenna, proving its suitability for real UAV drone-based GPR surveys.
The final and most relevant test was a flight trial with the antenna mounted on a UAV drone. The DH-Bowtie antenna was successfully integrated onto a DJI T60 agricultural drone. A flight survey was conducted over a terraced field, maintaining a constant altitude of approximately 30 meters above the highest ground point. The drone maintained stable flight throughout the survey, confirming the aerodynamic compatibility of the low-profile antenna design. The collected GPR data successfully imaged the varying ground surface echo with time delays corresponding accurately to the known topography, demonstrating the system’s operational capability from a UAV drone platform. Strong surface returns were detectable even at altitudes exceeding 50 meters, affirming the antenna’s robust radiation performance.
In conclusion, this work presents a comprehensive solution for low-frequency GPR antenna miniaturization for UAV drone deployment. The proposed Hybrid Folded Dipole Patch Antenna (HFDPA) combines a bow-tie radiator with a folded linear tail section, achieving a compact size of 0.12λ at 20 MHz. When combined with strategic dielectric loading, it recovers gain performance comparable to a standard half-wave antenna. The novel “Broadband Loading, Local Optimization” (BLLO) resistive loading technique provides an efficient method to achieve wide impedance bandwidth with minimal ringing and reduced ohmic loss compared to classical approaches. Laboratory measurements, field surveys, and successful UAV drone flight trials collectively validate the design. The HFDPA antenna, with its low-frequency capability, compact form factor, and robust performance, represents a significant step forward in enabling deep subsurface sensing from agile UAV drone platforms, opening new possibilities for efficient and extensive geophysical exploration.