Multipath parameters estimation with a reduced complexity unitary‐SAGE algorithm

Abstract

AbstractEstimating multipath parameters from multi‐dimensional measurement data using maximum‐likelihood methods can be a very time‐consuming process due to the iterative nature of these algorithms. Although the Space‐Alternating Generalized Expectation‐maximization (SAGE) algorithm has a higher convergence speed compared to the classical Expectation‐Maximization (EM) algorithm, it can still be computational‐intensive. Since most of the computations of the SAGE algorithm are concentrated in the maximization step (M‐step), we propose a new implementation of this algorithm, namely the Unitary‐SAGE (U‐SAGE) algorithm, where the entire M‐step in every iteration is evaluated in the real‐valued domain. This helps to reduce the processing time and memory requirements of the classical SAGE algorithm since all computations of the M‐step are performed using efficient matrix manipulation. Here, we present the general implementation of the U‐SAGE algorithm in the frequency domain when applied in both the element‐space (ES) and the newly developed beamspace (BS) domains. We show that the convergence characteristics and accuracy of the newly proposed U‐SAGE algorithm is similar to the classical SAGE algorithm, but with a significant reduction in overall computation time and memory usage. Copyright © 2004 AEI