Skip to Main Content
An important problem in adaptive beamforming algorithm design is robustness to steering vector uncertainty. A white-noise gain constraint (WNGC) has historically been an effective approach. Recently, the robust Capon beamforming class of algorithms has been developed to provide robustness through the use of a steering vector uncertainty region and an implicit steering vector estimation step as part of the beamformer. Like WNGC, the RCB algorithm can be shown to produce weight vectors based upon a diagonal loading of the covariance matrix. This paper provides some direct comparison of WNGC and RCB approaches for scenarios of interest. Geometrical interpretations of RCB algorithms provide insights to their optimality. Both approaches have a parameter than can be used to tune the performance from aggressive to conservative. Major observed differences are that RCB provides more accurate power estimates, focuses its high loading more locally, but exhibits higher white noise gain near sources.