Grouping Parallel Detection Method of UAV Based on Multi Features of Image Transmission Signal

Yongxiao Xie, Peng Jiang, Xianyao XiaoRadioengineering 2021 · DOI

Abstract

The emergence of low, slow, and small civilian unmanned aerial vehicles (UAV) brings fun and convenience to life and work. However, with the widespread popularity of UAV, the illegal activities caused by them have gradually increased, causing great harm to social security. To solve this problem, in the paper, we propose a set of detection and recognition methods for UAV by UAV image transmission signal (ITS). The method is divided into two groups. In the first group, according to the signal characteristics in different transform domains such as spectrum and time-frequency spectrum, three sets of algorithms are proposed, which are time-frequency ridge double feature estimation (TFRDFE), segmented spectrum estimation (SSE) and cycle accumulation estimation of segmented spectrum (CAE-SS). Three sets of algorithms are estimated to perform blind detection on suspected UAV ITS. The second group uses the accurate recognition algorithm of UAV ITS to extract the periodic features in the signal, and completes the recognition of UAV through feature matching, decision criteria and other methods. The two groups of methods are implemented in parallel, and when the two groups both detect and recognize the flying target, it can be determined that there is UAV in the target airspace. The experimental results show that the recognition rate of the first group of suspected UAV ITS blind detection algorithm can reach 100% when the (signal-to-noise ratio) SNR is -22 dB. The second group of UAV ITS recognition algorithm can achieve 100% recognition rate when SNR is -4 dB. Therefore, this method can complete the multi-target recognition of UAVs and has practical application value.

Introduction

The proliferation of "low, slow, and small" civilian drones has created a significant security headache. Whether it's interfering with commercial flight paths, invading privacy, or posing threats to government facilities, illegal drone activity is on the rise. Traditional detection methods often fall short: radar struggles with small plastic frames and low-altitude clutter; optical systems are limited by weather and line-of-sight; and acoustic sensors are easily drowned out by urban noise.

Radio Frequency (RF) detection is the most promising alternative, but it faces its own hurdles. Drones operate in the crowded ISM bands (like 2.4GHz and 5.8GHz) alongside Wi-Fi and Bluetooth. Distinguishing a drone's video feed from a nearby router—especially when the signal is weak (low Signal-to-Noise Ratio)—is a complex problem.

Overview of the UAV detection system architecture, showing the acquisition and processing pipeline.

Proposed Methodology

The proposed system utilizes a Grouping Parallel Detection Method that essentially uses a "two-factor authentication" approach for signal processing. This parallel architecture ensures high sensitivity while maintaining specificity.

Group A: Blind Detection (The Scout)

This group focuses on finding any suspected UAV signal using three sub-algorithms designed for extreme sensitivity:

  1. TFRDFE (Time-Frequency Ridge Double Feature Estimation): Extracts "ridges" from the signal's spectrogram to identify wide-bandwidth, long-duration signals.
  2. SSE (Segmented Spectrum Estimation): Analyzes the signal's bandwidth and duration across different segments.
  3. CAE-SS (Cycle Accumulation Estimation of Segmented Spectrum): An optimized version of SSE that uses signal accumulation to "boost" the signal out of the noise floor.

Group B: Accurate Recognition (The Expert)

Once a signal is flagged, Group B identifies the specific drone model using SSCAFE (Sliding Shift Cyclic Auto-correlation Feature Extraction). Since most drone video feeds use Orthogonal Frequency Division Multiplexing (OFDM), they have unique "cyclic prefixes"—essentially digital fingerprints. By matching these periodic features against a pre-built library, the system can identify the specific UAV model.

Algorithm structure diagram and signal spectrograms illustrating the parallel processing flow.

Experimental Results and Discussion

The researchers evaluated their system using real-world signals collected from DJI Phantom 4Pro and DJI Inspire Two units, with Wi-Fi signals as a control group. The signals were sampled at 40 MSPS. To test the limits of their algorithms, varying levels of Gaussian white noise were superimposed onto the captured data, simulating long-distance detection scenarios.

The experimental results demonstrate impressive robustness:

  • Blind Detection (Group A): The CAE-SS algorithm achieved a 100% recognition rate at an SNR of -22 dB. This is significantly better than traditional methods and even some recent deep-learning approaches.
  • Model Identification (Group B): The system achieved a 100% recognition rate at -4 dB SNR. While less sensitive than Group A, it successfully distinguished between the Phantom 4Pro, the Inspire Two, and Wi-Fi signals.

By using traditional signal processing (FFTs and autocorrelation) rather than heavy deep learning models, the approach maintains lower computational complexity, making it more suitable for real-time hardware implementation.

Conclusion

By combining high-sensitivity blind detection with high-specificity model recognition in a parallel architecture, this system reliably distinguishes drones from background noise and Wi-Fi signals. While Group B is more susceptible to noise than Group A, the dual-group approach ensures high reliability. Future work will focus on hardware optimization for field deployment and expanding the feature library to include more UAV models.

Summary updated

Abstract

The emergence of low, slow, and small civilian unmanned aerial vehicles (UAV) brings fun and convenience to life and work. However, with the widespread popularity of UAV, the illegal activities caused by them have gradually increased, causing great harm to social security. To solve this problem, in the paper, we propose a set of detection and recognition methods for UAV by UAV image transmission signal (ITS). The method is divided into two groups. In the first group, according to the signal characteristics in different transform domains such as spectrum and time-frequency spectrum, three sets of algorithms are proposed, which are time-frequency ridge double feature estimation (TFRDFE), segmented spectrum estimation (SSE) and cycle accumulation estimation of segmented spectrum (CAE-SS). Three sets of algorithms are estimated to perform blind detection on suspected UAV ITS. The second group uses the accurate recognition algorithm of UAV ITS to extract the periodic features in the signal, and completes the recognition of UAV through feature matching, decision criteria and other methods. The two groups of methods are implemented in parallel, and when the two groups both detect and recognize the flying target, it can be determined that there is UAV in the target airspace. The experimental results show that the recognition rate of the first group of suspected UAV ITS blind detection algorithm can reach 100% when the (signal-to-noise ratio) SNR is -22 dB. The second group of UAV ITS recognition algorithm can achieve 100% recognition rate when SNR is -4 dB. Therefore, this method can complete the multi-target recognition of UAVs and has practical application value.
BibTeX
@inproceedings{xie2021grouping,
  author = {Yongxiao Xie and Peng Jiang and Xianyao Xiao},
  title = {{Grouping Parallel Detection Method of UAV Based on Multi Features of Image Transmission Signal}},
  booktitle = {Radioengineering},
  year = {2021},
  doi = {10.13164/re.2021.0556},
}
Figure