Enhancement the DOA for 2D Coprime Array Using CNN

Abstract

Direction of arrival is the most important issue in the array signal processing the use of traditional technique for the DOA estimation would require large number of array sensor this lead to unfeasible and in the same time very costly .so the use of coprime array will use less number of array with the same result of the use of traditional method the cooperation of this method with the deep learning method that suggested in this paper will provide efficient DOA estimation .the suggested scenario is to use 2D coprime array with convolutional neural network (CNN) to analyze the properties of the covariance matrixes of the incoming signals. The dataset use depends on the signal detected by the sensor this approach in the simulation show demonstration on the traditional method such as MUSIC and ESPRIT with error only 0.04 or the arrival angle.