K-Means Clustering of Arrival Estimates in a 2D Signal Using DCONVNET

This research presents a fast two-dimensional direction-of-arrival (DOA) estimation approach based on a deep learning convolutional neural network (DCONVNET) for low-elevation targets of very-high-frequency (VHF) array radar. Initially, the technique uses the azimuth and pitch angle uncoupling properties of a uniform planar array to transform the two-dimensional angle estimation problem into two one-dimensional DOA estimation problems. Digital beamforming is used to retrieve the target information in the azimuth and elevation dimensions. Finally, the azimuth and pitch angles are estimated using the alternating direction method of multipliers. This approach significantly reduces complexity and eliminates the need for eigenvalue decomposition during operation, which further enhances efficiency by avoiding complicated two-dimensional joint estimation computation.