基于神经网络的多波束天线自适应调零

Neural network-based adaptive nulling with multiple-beam antennas

具有一定的干扰自适应调零能力是对卫星通信系统的基本要求。本文结合多波束天线特性,提出了一种基于RBF神经网络的MVDR波束合成器干扰抑制方案。该方法可有效避免采样矩阵求逆带来的影响,收敛速度更快,对所需信号参数具有一定的顽健性。计算机仿真验证了该方法的可行性。

It is essential to satellite communication system with the ability of nulling interferenceadaptively. In this paper, radial basis function (RBF) neural network is applied to minimum variancedistortionless response (MVDR) beamformer for interference suppression problem with multiple beamantennas (MBA). This approach can effectively avoid the effect of direct sample matrix inversion (SMI), and owns faster convergence speed. At the same time, it is robust to the variation of desired signalparameter. Simulation results show the effectiveness of the proposed method.