Rotor-Body Echo Separation Using a Cyclic-Power-Guided Soft Mask from UAV Radar Signals

Area of Science:

• Radar signal processing
• Unmanned Aerial Vehicle (UAV) characterization
• Micro-Doppler signature analysis

Background:

• Rotor-induced micro-Doppler signatures are crucial for radar-based rotary-wing UAV identification.
• Strong body returns near zero Doppler often obscure these signatures, especially in hovering or low-speed scenarios.
• Existing methods like EMD and VMD struggle with mode mixing and leakage under low signal-to-noise ratio (SNR) conditions.

Purpose of the Study:

• To develop a robust framework for separating rotor signatures from body returns in radar echoes.
• To enhance the continuity and reduce leakage in micro-Doppler structures.
• To improve the stability and accuracy of rotor-speed estimation for UAVs.

Main Methods:

• Proposed a Cyclic-Power-Guided Soft Mask (CPGSM) framework utilizing cyclostationary periodicity as a prior for rotor-body separation.
• Implemented a CPS-guided soft masking procedure to identify a DC-dominant overlap band.
• Employed cyclic power spectrum analysis within the overlap band to guide time-frequency allocation, combined with deterministic assignment elsewhere.

Main Results:

• Demonstrated improved micro-Doppler continuity and reduced body leakage compared to hard DC isolation, EMD, and VMD.
• Achieved more stable performance across a range of 5-30 dB SNR.
• Provided consistent rotor-speed estimates across different sensing configurations.

Conclusions:

• The CPGSM framework offers a significant advancement in radar-based UAV characterization by effectively separating rotor and body signatures.
• This method enhances the reliability of micro-Doppler analysis, particularly in challenging low-SNR environments.
• CPGSM provides a physically grounded approach for improved rotor-speed estimation and overall UAV monitoring.