An Intelligent Anti-Interference Scheme for FDA-MIMO Radar Under Nonideal Condition

The realization of anti-interference technologies via beamforming for applications in frequency diverse arrays and multiple-input and multiple-output (FDA-MIMO) radar is a field that is undergoing intensive research due to its two-dimensional range-angle-dependent beampattern characteristics. To solve the missing covariance matrix problem and improve the anti-interference capability of FDA-MIMO radar, we present a two-stage based intelligent anti-interference scheme for FDA-MIMO radar under the nonideal condition. The scheme consists of two parts: signal covariance matrix missing data recovery and intelligent beamforming vector estimation. A dual-channel generation adversarial network (DC-GAN) structure is proposed to effectively recover both real and imaginary parts of data from a covariance matrix. Based on the recovered covariance matrix, the beamforming vectors are accurately estimated by constructing a one-dimensional convolution neural network (1D-CNN). Meanwhile, a multiple-target process scheme combined with the 1D-CNN is introduced to deal with multitarget situation. In the numerical simulation part, the simulation results reveal that the DC-GAN network can effectively recover the missing data from the covariance matrix, and the lower the missing data rate, the better the data recovery performance. In addition, in the simulation of beamforming vector estimation, the effects of two different input modes on network training performance are evaluated, and the performance differences between a fully connected neural network and 1D-CNN are analyzed and compared. The numerical simulation results verify the effectiveness of the proposed FDA-MIMO radar anti-interference scheme under different number of interference signal scenarios and improve the interference suppression capability of FDA-MIMO radar.