多输入多输出(Multiple In Multiple Out,MIMO)雷达的阵元故障会导致其协方差矩阵出现整行整列数据缺失,从而降低其角度估计性能.为此,提出一种对抗自编码填补网络(Adversarial Autoencoder Imputation Network,AAEIN)来重构故障阵元的缺失数据.该网络由负责重构缺失数据的自动编码(Autoencoder,AE)网络和负责分辨数据来源的鉴别器组成.在二者的对抗训练中,AE网络的重构能力和鉴别器的分辨能力不断得到提升,直至两者收敛.为避免网络训练过程中参数量大和计算复杂度高的问题,文中结合MIMO雷达协方差矩阵的Hermitian特性,使用协方差矩阵上三角部分构建数据集用于网络训练.仿真结果表明,文中方法可以有效地重构故障阵元的缺失数据且具有较高的重构精度.
Reconstruction of Missing Data of Faulty Array Elements in MIMO Radar Based on Adversarial Autoencoder Imputation Network
Faulty array elements in Multiple-Input Multiple-Output(MIMO)radar result in entire rows and columns of missing data in its covariance matrix,thereby degrading its angle estimation performance.To address this issue,an Adversarial Autoencoder Imputation Network(AAEIN)is proposed for recon-structing missing data of faulty radar elements.The network comprises an Autoencoder(AE)responsible for reconstructing missing data and a discriminator responsible for distinguishing data sources.During ad-versarial training,the reconstruction capability of the AE network and the discriminative ability of the dis-criminator continually improve until both converge.To mitigate the issues of large parameter count and high computational complexity during network training,this paper exploits the Hermitian properties of the MIMO radar's covariance matrix.The upper triangular part of the covariance matrix is used to construct a dataset for network training.Simulation results demonstrate the effectiveness of the proposed method for reconstructing missing data of faulty radar elements with high precision.
MIMO radararray element failureadversarial autoencoder networkHermitian matrixda-ta reconstruction
NETL
NSTL
万方数据
