Signal waveform estimation in sensor array processing

The problem of estimating the waveform of narrowband signals impinging on a sensor array is of importance in many engineering applications. A known look-direction for a signal-of-interest (SOI) is usually assumed and a particular performance measure is optimized. To achieve optimality, these methods require the SOI to be uncorrelated with the other signals as well as the additive noise. Herein, the signal parameters (directions) are assumed to be unknown and an optimal estimator of all the signal waveforms is derived. The signal waveforms can be correlated or even coherent. The methods is based on a stochastic signal model and uses optimal signal parameter estimes obtained from the measurements via maximum likelihood estimation or weighted subspace fitting. Based on these estimates, the emitter signal covariance is estimated and a structured maximum a posteriori estimate of the signal waveforms is obtained. Simulations are presented comparing this estimator to the deterministic maximum likelihood estimator and the unstructured stochastic estimator.