Radar Drone Identification via Spectral Feature Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional radar target classification methods, particularly those using neural networks, operate as 'black boxes,' making it impossible to understand the reasoning behind their decisions and relying on limited parameters, which hinders accurate and transparent identification of drones and other targets in radar signals.
Innovation Solution
An apparatus and method utilizing a visualization module to extract feature-specific structures from spectral data by modifying spectra and defining thresholds, combined with a classification module that performs predefined tests based on specific features to identify drones, providing a transparent and reliable classification process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If neural networks are used for radar target classification, then classification capability is achieved, but transparency and interpretability of the classification decision are lost
Solution Approach 1:
The patent segments the classification process into distinct modules: a visualization module that extracts and displays specific spectral features, and a classification module that applies predefined rules. This segmentation makes the decision process transparent by showing which features were extracted and how they led to the classification, rather than using a monolithic neural network black box.
Solution Approach 2:
The patent introduces an intermediary visualization module between the raw radar signal and the classification decision. This module extracts and presents specific spectral features (such as peak frequencies, amplitude values, and spectral patterns) that mediate between the complex input signal and the final classification, making the reasoning process visible and interpretable.
2Adaptability or versatility
If conventional classification algorithms are used, then classification is possible, but the features are not specific enough to distinguish drones from other targets
Solution Approach 1:
The patent applies local quality by focusing on specific local features in the spectral domain that are characteristic of drones. Instead of using general classification features, the visualization module extracts localized spectral patterns, peak structures, and frequency-specific characteristics that are particular to drone signatures, thereby improving discrimination precision while maintaining versatility through multiple testable features.
3Measurement precision
If complex learning algorithms are used, then classification accuracy may improve, but computational complexity and processing time increase
Solution Approach 1:
The patent employs self-service by using predefined spectral features and simple comparison tests that do not require external training data or complex learning algorithms. The system serves itself by having the visualization module extract features that are inherently discriminative, and the classification module applies straightforward rule-based tests, eliminating the need for computationally intensive training phases while maintaining high accuracy.
Data Source
Figure 1
Figure 2~3B
Figure 4A~4B
AI summary
An apparatus for automatic identification of a drone in radar signals based on features of spectral data of the radar signals is disclosed. The apparatus comprises: a visualization module (110) configured to visualize feature-specific structures in the spectrum (spe) and a classification module (120), configured to identify the drone based on predefined tests testing a presence of at least one feature and being performed on a result of visualization module (110).