Automatic LPI Radar Waveform Recognition Method Based On Monobit Quantization

Radar signal reconnaissance equipment requires real-time interception of enemy Low Probability of Intercept (LPI) radar signals for analysis and identification. Monobit receivers, leveraging their monobit ADC quantization width, enhance sampling rates, reduce data volume, and accelerate processing speeds, guiding the future development trend of radar reconnaissance equipment. This paper proposes, for the first time, an automatic recognition method for LPI radar waveforms based on monobit quantization. Firstly, we analyze the exceptional suitability of the Wigner-Ville Distribution (WVD) time-frequency analysis (TFA) method for monobit signals. Subsequently, we construct an Efficient Monobit LPI Radar Waveform Recognition Network based on Spatially Separable Convolutions (EMLN-SSC). This network effectively extracts and reduces the dimensionality of edge details in the time-frequency spectrum of monobit signals, achieving superior recognition rates compared to current State-of-The-Art (SOTA) models. Meanwhile, it reduces parameter quantity and computational complexity by over 48% and 28%, respectively. Despite a slight decrease of 1.62% in average recognition rate compared to traditional non-quantized signal recognition systems, the proposed method significantly reduces computational complexity, facilitating hardware deployment, and holding great potential in electronic reconnaissance.