With the rapid advancement in unmanned aerial vehicle technology, drones have become increasingly prevalent in various civilian and industrial applications. One notable area is power line inspection, where UAVs have replaced traditional manual inspections due to their cost efficiency, flexibility, and safety. However, the proliferation of drones also brings significant security concerns, particularly the intrusion of malicious UAVs into power inspection zones. Such intrusions can lead to collisions with critical infrastructure, theft of sensitive inspection data, and even threats to human life. Existing countermeasures, such as electronic geofencing, suffer from limitations in real-time response, adaptability, and obstacle avoidance. In this context, we propose a novel drone regulation method that leverages satellite navigation spoofing to actively prevent malicious UAV intrusion. Our approach establishes a virtual control zone around the power inspection area, detects approaching UAVs via radar, generates deceptive navigation signals, and induces the intruder to fly away. We validate the effectiveness of this method through field experiments and simulations, demonstrating superior performance in terms of response speed, flexibility, and obstacle avoidance compared to conventional geofencing techniques.
The term drone regulation encompasses a broad spectrum of strategies aimed at managing UAV behavior in restricted airspaces. Among these, geofencing is the most widely adopted technique in commercial drones. Geofencing creates a virtual perimeter based on GPS coordinates, and the UAV is programmed to avoid entering or to land automatically if it breaches the boundary. However, this method relies on the drone’s onboard software and compliance by the operator. Malicious actors can bypass such restrictions by disabling geofencing, using custom firmware, or simply ignoring warnings. Moreover, geofencing is static and cannot dynamically adapt to changing environments or multi-drone scenarios. Our proposed solution overcomes these drawbacks by actively interfering with the drone’s navigation system, thereby enforcing drone regulation even against non-cooperative targets.
We begin by defining the control zone for drone regulation. The power inspection area is surrounded by a circular or polygonal virtual boundary. For simplicity and ease of implementation, we adopt a circular control zone with radius R centered at the radar station. The detection range extends to a warning circle of radius Rw where Rw > R. When a UAV crosses the warning circle, the system initiates the spoofing procedure. The condition for a point P(x, y) to be inside the circular control zone is:
$$(x – x_0)^2 + (y – y_0)^2 \leq R^2$$
where (x0, y0) is the center of the circle. For a polygonal zone, we use the ray-casting algorithm: a horizontal ray from the point to infinity is drawn; if the number of intersections with polygon edges is odd, the point lies inside. This method is robust for convex and concave shapes alike. Table 1 summarizes the zone parameters used in our experiments.
| Parameter | Value | Unit |
|---|---|---|
| Control zone radius R | 27 | m |
| Warning circle radius Rw | 46 | m |
| Radar detection range | 100 | m |
| GPS satellite count | 9+ | – |
| Wind speed | 0.3–1.5 | m/s |
Detection of the intruding UAV is accomplished using a radar system. The radar transmits electromagnetic waves and receives echoes from the UAV. The maximum detection range Rmax is given by the radar equation:
$$R_{\text{max}} = \left[ \frac{P_t G_t G_r \tau \sigma \lambda^2 F_t^2 F_r^2}{(4\pi)^3 k T_s D_0 C_b L} \right]^{1/4}$$
where Pt is the transmitted power, Gt and Gr are antenna gains, τ is pulse width, σ is radar cross section of the UAV, λ is wavelength, k is Boltzmann constant, Ts is system noise temperature, D0 is detectability factor, Cb is bandwidth correction factor, L is system loss, and Ft, Fr are propagation factors. Assuming a Swerling I fluctuation model and Gaussian noise, the instantaneous detection probability Pd can be expressed as:
$$P_d = \left[ 1 + \frac{1}{1 + (R_0 / R)^4} \right]^{-N}$$
where R0 is the range at which SNR = 1, and N is the number of integrated pulses. In our setup, we achieve reliable detection up to 100 m with a probability exceeding 0.9.
Once the UAV is detected and its position (xu, yu, zu) is known, we generate a deceptive target point E (the spoofing waypoint). This point lies on the line from the UAV to the control zone center, extended far beyond the center. The planar coordinates of E are calculated as:
$$(x_e, y_e) = (2x_0 – x_u, 2y_0 – y_u)$$
which is the symmetric point of the UAV about the center. To make the spoofing more convincing, we often choose a real remote location, for example, the island of Saipan. These Cartesian coordinates must be converted to geodetic latitude B, longitude L, and height H for GPS signal generation. The conversion uses the reference ellipsoid parameters:
$$x = (N + H) \cos B \cos L$$
$$y = (N + H) \cos B \sin L$$
$$z = \left[ N(1 – e^2) + H \right] \sin B$$
with N = a / √(1 – e² sin²B), where a = 6378.137 km is the semi-major axis, and e² = 1 – (b/a)² with b = 6356.752 km. The spoofing signal is generated using a software-defined radio (HackRF) and processed by GPS-SDR-SIM. The navigation data file (ephemeris) is downloaded from NASA in real time to ensure authenticity of the counterfeit signals.
The spoofing procedure begins when the UAV enters the warning circle. Our system transmits a GPS L1 signal that simulates the same satellite constellation but with a modified pseudorange corresponding to the deceptive point E. The signal power is carefully calibrated to be slightly higher than the real GPS signals, ensuring that the UAV’s receiver locks onto the false signals. Once locked, the UAV’s navigation system reports the spoofed location, and the flight controller commands the drone to fly toward that distant point, effectively driving it away from the power inspection area. This drone regulation mechanism does not require cooperation from the UAV operator, making it robust against malicious intent.
We conducted field experiments using a DJI Phantom 3 Standard drone as the target. The experimental site was a school football field, free of electromagnetic interference and with clear line-of-sight. The control zone was marked on the ground with a 27 m radius, and the warning radius was set to 46 m (as shown in Table 1). The drone was flown manually toward the zone at speeds between 1.5 and 3.5 m/s and at an altitude of 8.2 m. As it crossed the warning circle, the spoofing system was activated. The deceptive waypoint was chosen as Saipan (15.087252°N, 145.504356°E). Within approximately 4 seconds, the drone’s GPS reported a location change to the Pacific Ocean, while the camera feed still showed the football field. The drone then entered attitude mode (no GPS hold) and began to drift away from the zone due to the spoofed velocity command. Table 2 shows the measured response times for different approach speeds.
| UAV speed (m/s) | Time to change location (s) | Time to exit control zone (s) |
|---|---|---|
| 1.5 | 2.1 | 3.8 |
| 2.5 | 1.8 | 2.9 |
| 3.3 | 1.5 | 2.4 |
| 3.5 | 1.3 | 2.2 |
To evaluate obstacle avoidance capability, we performed simulations in the ROS (Robot Operating System) environment. A UAV was programmed to follow a path from its entry point to a distant waypoint, with eight obstacles (spheres, cylinders) placed along the trajectory. Using our spoofing method, the operator could dynamically adjust the deceptive point via a smartphone app, sending 360-degree directional commands to steer the UAV around obstacles. The simulation recorded a 100% success rate in avoiding all obstacles, whereas the conventional geofencing approach (which forces the drone to land vertically) resulted in collisions when obstacles were dense. Table 3 compares the two methods in key performance metrics.
| Criteria | Our method | Traditional geofencing |
|---|---|---|
| Real-time response | Immediate spoofing upon detection | Delayed warning or automatic landing |
| Flexibility | Adjustable waypoint via app | Fixed no-fly zone; no remote override |
| Obstacle avoidance | Steerable path; avoids obstacles | Vertical descent; may hit obstacles |
| Cooperation required | No | Yes (drone must obey geofence) |
| Multi-UAV capability | Potentially scalable with multiple transmitters | Limited to individual UAV firmware |
The experimental results demonstrate that our drone regulation method achieves efficient and flexible control over malicious UAVs. The spoofing signal quickly overrides the real GPS, forcing the drone to turn away within seconds. Furthermore, the ability to modulate the deceptive point’s velocity (by gradually shifting the spoofed position) allows us to handle high-speed intruders. For a UAV flying at 3.3 m/s, the total time from spoofing activation to complete exit of the control zone was only 2.4 s, which is well within the safety margin for power facilities. In contrast, traditional geofencing would only trigger a landing command, which could take 5–10 seconds and may not prevent intrusion if the drone is already inside the zone.
We also addressed the limitation of GPS spoofing priority. In some drones, the GPS signal is not the highest-priority sensor; for instance, vision-based positioning may override. To enhance reliability, we recommend jamming the remote control signal simultaneously, so the drone cannot counteract the spoofed navigation. For multi-drone scenarios, one can deploy multiple spoofing transmitters or use a directional antenna array to cover wider areas. Further improvements include integrating acoustic or optical sensors for backup detection in case of radar failure.
In conclusion, we have presented a novel drone regulation method specifically designed to protect power inspection zones from malicious UAV intrusion. By combining radar detection, GPS spoofing, and real-time app control, our system offers superior real-time performance, flexibility, and obstacle avoidance compared to existing geofencing solutions. The experimental validation confirms that a typical consumer drone can be reliably driven away within seconds. Future work will focus on scaling the approach to counter coordinated drone swarms and to harden against anti-spoofing countermeasures. With the increasing reliance on UAVs for critical infrastructure monitoring, effective drone regulation is paramount, and our method provides a practical, non-destructive solution to ensure safety and security.