Angle of arrival and signal to noise ratio estimation for broadband signals by phased adaptive array antennas

Abstract

The MUltiple SIgnal Classification (MUSIC) algorithm is the most popular algorithm to estimate angle of arrival (AOA) for narrowband signals. The algorithm is based on using subspace noise Eigenvectors (Q n ) matrix which is related to the smallest set of Eigenvalues for the received signal covariance matrix (φ XX ). These sets of Eigenvalues are corresponding to the channel's thermal noise. It is found that, the distribution of Eigenvalues magnitudes vary according to the value of received signal's bandwidth. The Eigenvalues magnitudes which refer to the received sources are spreading on the account of noise subspace Eigenvalues. This new distribution causes a change in the related Eigenvectors, and the link of Eigenvalues to thermal noise is no longer exists. Since MUSIC algorithm is based on the use of these Eigenvectors relating to the thermal noise Eigenvalues, and all these thermal noise subspace Eigenvalues are occupied by wideband signals, the MUSIC algorithm will no longer work. In this paper a new approach was suggested to estimate the angle of arrival (AOA) as well as output signal to noise ratio (SNR) for wideband signals by using linear adaptive array antennas system with transversal filter as a wideband beam-former processor in conjunction with Linearly Constrained Minimum Variance beam-former algorithm (LCMV). Behind each sensor there is a general mathematical model for M array sensors and J taps with a delay time between taps equivalent to (90°) phase shift at center frequency. The proposed approach was simulated and tested in different cases with different input levels and different received signals relative to bandwidth (10%, 15%, 20% up to 40%). The results show that, with a proper number of taps behind each sensor, the system has excellent compensation the effect of signals bandwidth and estimate the angles of arrival as well the output SNR's for all received signals simultaneously with high accuracy. Under the same assumptions the MUSIC technique is tested with transversal filter as a broadband processor and the results were very poor.