DOA estimation By ESPAR antennas using fourth-order MUSIC algorithm

Abstract

Direction of arrival (DOA) estimation technology plays an important role in enhancing the performance of adaptive arrays for mobile wireless communications. A number of DOA estimation algorithms have been developed. For the most recent ones being MUSIC and ESPRIT algorithms, who both utilizing subspace-based on exploiting the eigen structure of the input covariance matrix. Among many DOA algorithms, MUSIC algorithm which is based on the correlation matrix decomposition of the received signals has a good estimation performance and alleviates the system complexity comparing to other algorithms. As we all know, mobile terminals have rigorous limitation on hardware and algorithm complexity as well as power consumption. Unlike conventional antenna array systems which require one receiver chain per antenna branch, the ESPAR(Electronically steerable parasitic array radiator) antennas can be controlled by means of its electronically controllable reactance, which only need a single-port output and requires less complex hardware and low power consumption. Thus due to these advantages, the ESPAR antennas can be used for application to wireless communication system, especially to mobile user terminals. Recently, a great deal of effort has been focused on taking the advantages of ESPAR antennas to improve the performance of localization system. According to the ESPAR antennas configuration, the RF currents on the element, which depend on the values of the reactances, are not independent but mutually coupled with each other. The single output is a highly nonlinear of the reactances. And the antenna pattern is formed due to the reactance on the parasitic radiators. All of these ESPAR antennas features make it complex to adapt to conventional algorithms. In this paper, we combine a reactance domain MUSIC algorithm based on fourth-order MUSIC algorithm with a 7-element ESAPR antenna to evaluate and estimate the DOA performance. The main motivation is that fourth-order comulant is insensitive to additive Gaussian noise regardless of whether it is white or colored. The covariance matrix of the output can be computed by the fourth-order cumulants. The cumulants have an more important suppression property for not only Gaussian noise but also non-Gaussian noise. By using the reactance domain fourth-order MUSIC algorithm, we create the correlation matrix for the single-port output ESPAR antenna. Finally we evaluate the DOA estimation performance for both one and two sources cases in a feasible SNR environment and the results show a good performance.