Compressed Sensing-Based Terahertz Imaging Algorithm in 6G-ISAC Terminal Systems

As an important embodiment of integrated sensing and communication (ISAC), terminal imaging using orthogonal frequency division multiplexing (OFDM) communication waveforms in terahertz (THz) band is an important application component of future the sixth generation (6G) networks, because of the THz band has very small wavelengths and high imaging resolution combined with virtual array imaging technology. It means that THz terminal system can get high-resolution images as long as moving through space. However, OFDM-THz imaging system with fully sampling strategy exists an unbearable long-term time cost since the fully sampling spatial step length needs no more than half of the wavelength. How to achieve high-efficiency and high-quality image reconstruction under the premise of reducing the cost of sampling time is a difficult problem to overcome. Compressed sensing (CS) theory is a rapidly developing theory in recent years, which can reconstruct signal from under-sampled data by utilizing sparsity of signal. In this paper, a random under-sampling motion policy is applied for reducing time cost, and a CS-based algorithm is proposed for reconstructing image from under-sampled echo data. The algorithm utilizes the sparse prior information of image discrete gradient domain, decomposes sparse operators, and constructs several sub-problems that can be solved in parallel. A iterative residual correction term is used to realize fast iteration convergence of the algorithm, thus improving the convergence speed. The validity of the proposed algorithm is proved by theoretical analysis and experimental verification. With the comparison of fast parallel proximal algorithm (FPPA), it can be obvious that the proposed algorithm has both higher convergence speed and superior image quality under structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR).