This paper presents an adaptive signal detection algorithm with dynamic dual-threshold for low probability of intercept (LPI) radar signals in complex electromagnetic environments. The proposed method addresses the challenge of detecting weak LPI radar signals buried in noise through a synergistic approach combining Neyman-Pearson criterion-based false alarm probability constraints with Lagrange multiplier optimization for detection probability enhancement. The algorithm dynamically estimates background noise levels and employs dual thresholds to effectively suppress interference spikes and signal splitting phenomena. Implementation on Xilinx V7 field-programmable gate array (FPGA) platform utilizing pipeline architecture achieves real-time processing with 1.2 μs latency while maintaining resource utilization below 0.7%. Experimental results demonstrate 82.6% detection probability at 0 dB signal-to-noise ratio (SNR), representing a 14.1% improvement over conventional cell averaging-constant false alarm rate (CACFAR) and 3.3% improvement over state-of-the-art censored mean-clutter map CFAR (CM-CM CFAR) algorithms, with false alarm rate deviation controlled within 12% while outperforming four recently proposed advanced CFAR methods. Testing across continuous wave (CW), binary phase shift keying (BPSK), and linear frequency modulation (LFM) signals shows detection accuracy exceeding 93% for SNR ≥ 6 dB. The proposed solution provides an efficient and reliable approach for LPI radar signal detection in challenging environments, balancing computational efficiency with detection performance for practical deployment in electronic warfare and spectrum monitoring applications.
Wei et al. (Thu,) studied this question.