Efficient maximum likelihood angle estimation for signals with known waveforms in white noise

A large-sample decoupled maximum likelihood (ML) angle estimator, referred to as WDEML, for signals with known waveforms is presented herein by exploiting the a priori knowledge that the additive noise can be modeled as spatially and temporally white. We show that incorporating this additional knowledge improves angle estimation accuracy significantly over existing angle estimators for signals with known waveforms, especially in some difficult scenarios such as when the snapshot number is small and/or the signal-to-noise ratio (SNR) is low. Moreover, we show that WDEML achieves similar angle estimation performance as the optimal exact ML method but enjoys the benefit of a much simpler computational demand