Estimate-maximize algorithms for multichannel time delay and signal estimation

Computationally efficient iterative algorithms are developed for the joint maximum-likelihood estimation of the time delays and the spectral parameters of stationary Gaussian signals radiated from a stationary source and observed in the presence of uncorrelated additive noise at two or more spatially distributed receivers. The first set of algorithms is based on the estimate-maximize approach, while the second set achieves superlinear performance by using a modified approach. These algorithms decouple the estimation of all the unknowns, separately estimating the signal, the signal spectral parameters, and each channel´s delay, attenuation, and noise power. This decoupling considerably simplifies the estimator´s structure and computation. Convergence is guaranteed to the set of stationary points of the likelihood function, and each iteration increases the likelihood. The convergence rates are analyzed theoretically, and the algorithm´s performance demonstrated via simulation