Abstract
A new blind beamformer based on higher-order cumulant and neural network is presented in this paper. Its model and simulation are studied further. In this beamformer, the higher-order cumulant is used to estimate the steering vector of the desired signal and the linear programming neural network is employed to carry out the blind optimizing beamforming. The algorithm is not very dependent on the manifold of the array and adapts to the error easily. It is able to avoid computing the inverse of a matrix so as to cut down on computing amount and be easily implemented by hardware. Simulation proves efficiency and correctness of this beamformer model.