Expectation Maximization-Based Target Localization From Range Measurements in Multiplicative Noise Environments

In this letter, target localization from range measurements corrupted by both additive and multiplicative noises is studied. First, the maximum likelihood estimator (MLE) is proposed. Then, the novel expectation maximization (EM) algorithm with guaranteed convergence is proposed to find the MLE. It is shown that the expectation step can be calculated componentwisely, and the maximization step can be solved via the Majorization-Minimization (MM) approach. The algorithm is termed as R-EM-MM. To provide a benchmark performance of R-EM-MM, the theoretical performance of unbiased estimator, i.e., Cramèr Rao bound (CRB) is derived. In addition, the enhanced R-EM-MM (E-R-EM-MM) is proposed to jointly learn the noise statistics and perform target localization. Finally, numerical simulations are conducted to demonstrate the near optimal performance, compared with CRB and other state-of-art method.