A New Threshold-Based JADE-MUSIC Algorithm for DOA Estimation of Unknown Signal Groups
Shahriar Shirvani Moghaddam, Akbar Keshavarz Nasab · 2013
JADE-MUSIC algorithm combines joint approximate diagonalization of eigenmatrices (JADE) and multiple signal classification (MUISC) methods to estimate direction of arrival (DOA) of noncoherent signal groups which consist of coherent signals. Knowing the number of noncoherent sources, JADE algorithm separates the steering vectors of different signal groups using fourth order cumulants (FOC) and MUISC algorithm estimates DOAs. In real applications, achieving zero values of eigenvalues which show coherent signals is not practical. Also, the number of noncoherent sources should be determined. In this paper, a threshold-based method is proposed that first estimates the number of non-coherent sources based on eigenvalue gradient method (EGM) and then separates noncoherent groups using JADE algorithm. Finally, DOAs of coherent signals in each group are estimated using forward-backward spatial smoothing (FBSS) based MUSIC algorithm. Previous work uses a table to find the best threshold values, but according to the proposed method, threshold values can be obtained on-line using a second order statistics that introduces lower complexity. In this research, 3 noncoherent sources which each of them includes 4 coherent signals are simulated. Simulation results prove the effectiveness of proposed algorithm.