Real-time signal processing of UWB fuze based on K-VSD and cross-frame analysis

To meet the real-time processing requirements of target signals for ultra-wideband (UWB) fuze in interference environments, this paper proposes a target detection method based on K-SVD sparse dictionary learning. The output signals from the UWB fuze correlation receiver are collected and subjected to frame processing and target segment screening to construct a target dictionary. By introducing dictionary regularization and a low-usage atom replacement strategy, the method effectively reduces overfitting risks and enhances the dictionary's generalization capability. During the testing phase, the input signals are segmented, and sparse coding is performed on each frame using the trained dictionary. The Orthogonal Matching Pursuit (OMP) algorithm calculates the reconstruction error of each frame to identify continuous low-error intervals. For these intervals, a cross-frame coding fluctuation metric is proposed, which measures the dynamic changes in coding through the overlap ratio of activated atom sets between adjacent frames. This enables the differentiation between target signals and interference signals under a single dictionary framework, thereby improving the accuracy and robustness of target recognition. The simulation results demonstrate that this method achieves efficient and real-time target identification even under low signal-to-noise ratio (SNR) and multiple interference scenarios.