Hardware Implementation of Real-Time 2D DOA Estimation Algorithm Using an L-Shaped Array

This paper presents a method for two-dimensional (2-D) direction-of-arrival (DOA) estimation, focusing on both elevation and azimuth angles, using an L-shaped sensor array. The proposed approach employs a power spectrum search algorithm combined with LU decomposition to improve the accuracy of 2-D DOA estimation while significantly lowering computational complexity compared to methods like the Capon algorithm. Key advantages of the proposed method include: 1) It uses LU decomposition (LUD) on the covariance matrix rather than eigenvalue decomposition (EVD) or singular value decomposition (SVD), reducing complexity from $O\left(M^{3}\right)$ for EVD/SVD to $O\left(M^{3} / 3\right)$ for LUD; 2) It simplifies the data by converting complex data to real data through unitary transformation, cutting computation costs by $75 \%$; and 3) It does not require prior knowledge of the number of signal sources. The method’s effectiveness is verified through both software simulations and real-time experiments on a prototype system built with a software-defined radio (SDR) platform. The setup includes an Analog Devices FMCOMMS5 SDR and an AMD Zynq 7000 SoC ZC702 evaluation board, which achieve internal time and phase synchronization through hardware. MATLAB is used to interface with the FMCOMMS5 boards, providing real-time spectrum and angle estimations. Results from both MATLAB simulations and real-time tests strongly support the effectiveness of the proposed method.