Sparse DOA Estimation for Directional Antenna Arrays: An Experimental Validation

Abstract

In the literature, antenna arrays for direction of arrival (DOA) estimation are being theoretically optimized to maximize the number of sources that can be estimated for a given number of isotropic elements. In this work, coprime arrays with monopole and patch elements are designed and fabricated to evaluate the impact of antenna directivity and realistic antenna behavior on DOA estimation accuracy. Two DOA estimation algorithms, namely: MUSIC and Lasso are applied where the complex radiation patterns are incorporated within the steering matrix and then the received signal is modified accordingly. With four sources, a root mean square error (RMSE) of around 4 and 1 degrees is achieved at 5 dB with monopole and patch elements, respectively. Furthermore, a software-defined radio (SDR) platform is utilized to experimentally evaluate the real-time performance in realistic environments. It is shown that when the DOA deviates from the boresight, the estimation error increases due to antenna directivity. The performance of the patch array is better than its monopole counterpart due to the inherent multipath mitigation in the directive antennas and polarization purity of the two radiated electric filed components. Experimental results show that the RMSE in the DOA estimation of two sources using coprime array with monopole and patch elements is around 4 and 1 degrees, respectively.