A Lower Dimensional MLE Algorithm for Distributed Source Parameter Estimation

In the area of the distributed sources parameter estimation, the conventional maximum likelihood estimation (MLE) algorithm, which being called 4D-MLE here, is a four dimensional nonlinear optimization problem, therefore its computation cost is very large. A lower dimensional MLE algorithm, which being simplified to three dimensional nonlinear optimization problem, is proposed. It is called 3D-MLE. Both of the two algorithms use the Newton type search algorithm to find the globe optimum. In a single search process, the 3D-MLE can reduce the inverse computation of covariance matrix for 51 times and matrix multiplication for 87 times than the 4D-MLE. The Cramer-Rao Bound (CRB) of the new MLE algorithm is shown, its computation cost is reduced also. The computer simulation validates that, the estimation accuracy of the 3D-MLE is similar to the 4D-MLE. The new algorithm not only reduces the computation cost but also avoids the loss of performance.