TEFISTA-Net: A learnable method for high-resolution range profile reconstruction with low-frequency ultra-wideband radar

• Published in: Signal processing Vol. 214; p. 109257
• Main Authors: Li, Rui, Wang, Xueqian, Li, Gang, Zhang, Xiao-Ping
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.01.2024
• Subjects: Compressive sensing (CS); Deep learning; Geometrical theory of diffraction (GTD); Low-frequency ultra-wideband (LFW) radar; Deep learning; Low-frequency ultra-wideband (LFW) radar; Geometrical theory of diffraction (GTD); Compressive sensing (CS)
• ISSN: 0165-1684; 1872-7557
• DOI: 10.1016/j.sigpro.2023.109257

•The article proposes a learnable method named TEFISTA-Net for high-resolution range profile reconstruction (HRRP) with low-frequency ultra-wideband radar, which is an unrolling network with target enhancement capability.•The method achieves lower computational complexity and reduced difficulties in algorithm parameter tuning, compared with existing competitors based on compressed sensing.•The proposed network structure can effectively reduce background artifacts and improve the robustness of HRRP reconstruction.•Simulated and measured radar data demonstrate the performance advantages of the proposed method, in terms of the reconstruction accuracy and target saliency of HRRP. In low-frequency ultra-wideband (LFW) radar, the reconstruction of high-resolution range profile (HRRP) is an important task. The state-of-the-art compressive sensing (CS) method for HRRP reconstruction based on geometrical theory of diffraction (GTD) requires manual algorithm parameter tuning and is computationally expensive. In this paper, we propose a deep learning-based CS method for LFW radar HRRP reconstruction. We design a neural network architecture, i.e., target enhancement-based FISTA-Net (TEFISTA-Net), by unrolling the fast iterative shrinkage thresholding algorithm (FISTA). A new loss function based on the target-to-background ratio (TBR) is introduced for network training with the target enhancement capability in low signal-to-noise ratio (SNR) scenarios. The algorithm parameters in the traditional CS methods are substituted by network parameters learned from training data, getting rid of manual parameter tuning. In addition, simple convolution operations in our new method lead to lower computational complexity compared with existing methods. Experimental results based on diverse data show that the proposed method leads to higher computational efficiency with similar or better performance in comparison with existing state-of-the-art methods.