Modular Performance Analysis of a Cascaded TDM-MIMO FMCW Radar for Short-Range Counter-UAV Sensing.

Area of Science:

• Radar Systems Engineering
• Counter-UAV Technology
• Signal Processing

Background:

• Small unmanned aerial vehicles (UAVs) present significant challenges for short-range radar detection due to their low radar cross-sections (RCS) between -10 and -25 dBsm.
• Millimeter-wave radar systems are crucial for detecting these small targets, necessitating robust sensing frameworks.

Purpose of the Study:

• To provide a modular, analytical, and simulation-based benchmark for a cascaded 77 GHz TDM-MIMO FMCW radar system designed for counter-UAS applications.
• To evaluate the performance of the radar system across key sensing functions including target detection, direction-of-arrival (DoA) estimation, and micro-Doppler analysis.

Main Methods:

• A 77 GHz TDM-MIMO FMCW radar with 12 transmitters and 16 receivers was simulated, creating a 192-element virtual uniform linear array (ULA) over a 40 m range.
• The benchmark framework included range-Doppler processing, Doppler-dependent TDM phase compensation, virtual-array snapshot formation for DoA estimation, and micro-Doppler analysis using a nearest-centroid Mahalanobis classifier.
• Four specific benchmarks were analyzed: detection under Swerling fluctuation models, residual TDM phase error impact, DoA estimation accuracy, and micro-Doppler separability between UAVs and birds.

Main Results:

• Detection probability above 0.9 was maintained for targets with -10 dBsm RCS throughout the 40 m window under Swerling I; targets with -20 and -25 dBsm RCS showed reduced detection ranges (approx. 28 m and 21 m, respectively).
• The Time-Delay-of-Arrival (TLS-ESPRIT) algorithm for DoA estimation yielded the lowest conditional Root Mean Square Error (RMSE), but results were limited by short-range conditions and potential TDM phase errors.
• Micro-Doppler analysis achieved high per-class accuracy for birds (over 95% at 20 dB SNR), but overall four-class accuracy saturated around 72-75%, with significant confusion between micro-quadrotor and fixed-wing UAV classes.

Conclusions:

• The study identifies specific performance margins and limitations of the radar architecture, crucial for guiding future development and field validation.
• The findings highlight challenges in achieving high detection and classification accuracy for small UAVs, particularly those with low RCS and similar micro-Doppler signatures to natural targets like birds.
• Accurate TDM phase compensation and consideration of short-range operating conditions are critical for reliable DoA estimation in this radar system.