Radar Cross Section Reduction of Antenna Array With LK-ResNet

In this letter, a new method is proposed for reducing the radar cross section (RCS) of an antenna array. The RCS is reduced by optimizing the spatial arrangement of the antenna array, which is treated as a classification problem. To achieve this, an RCS excitation matrix is generated using radar signal direction of arrival and array information. This matrix is then input into a carefully designed light kernel residual network (LK-ResNet) to produce a switch matrix. Light kernels can increase the number of channels and nonlinearity with lower time complexity. The numerical and simulated results show that the proposed RCS excitation matrix provides much higher classification accuracy than the initial data. The LK-ResNet offers superior stealth performance when compared with other networks. Furthermore, the proposed method is also validated with measurement.