Microwave Photonic Radar Lost Bandwidth Spectrum Recovery Algorithm Based on Improved TSPN-ADMM-Net

Compared with conventional microwave regime radars, microwave photonic (MWP) radar is capable of transmitting extremely large bandwidth signals, wherein the frequencies of such signals distribute across multiple bands. In practical applications, the large bandwidth of MWP radar may be split into multiple discrete subbands due to various considerations such as antijamming, resource-saving, and communication band avoidance. Nonetheless, it leads to the fact that MWP radar suffers from the challenging problems of sidelobes’ elevation and main lobes’ broadening. These problems will affect the image quality seriously. To address this issue, a spectrum recovery algorithm based on an improved truncated Schatten-
$p$
norm and sparse regularizer-alternating direction method of multipliers (TSPN-ADMM) network is proposed in this article. This algorithm can efficiently recover the lost spectrum in MWP radar applications and further improve the imaging quality of the MWP radar. In the lost spectrum recovery problem, the parameters of the recovery algorithm directly determine the recovery performance. The different forms of lost spectrum possessed by MWP radar make the selection of parameters for the spectrum recovery algorithm extremely difficult. As a consequence, in this article, the spectrum recovery problem for MWP radar can be reformulated into a matrix completion problem by exploiting its joint sparsity and low-rankness. Based on the traditional TSPN-ADMM algorithm, an improved TSPN-ADMM-Net approach is proposed using the algorithm unrolling technique, wherein the hyperparameters in the TSPN-ADMM algorithm are optimized in an end-to-end training manner. Consequently, the algorithm proposed in this article can achieve excellent recovery results when dealing with the multiple spectrum missing situations existing in MWP radar. The effectiveness of the algorithm is verified by a combination of numerical simulations and actual MWP radar data.