Weighted coherent processing on sparse volumetric vector sensor arrays

A network of drifting sensors, such as vector sensors mounted to freely drifting buoys, can be used as an array for locating acoustic sources underwater. Localizing a source using traditional coherent processing methods has been improved through arbitrary selection of element-wise weights of the covariance matrix [Nichols and Sabra, JASA 2015, Vol. 138]. However, the selection of weights offers an opportunity to optimize localization performance measures such as accuracy or precision. Here, the performance of source localization is compared between optimal weightings and traditional weightings for both simulated and at-sea data collected from a freely-drifting vector sensor array deployed in the Long Island Sound.