I. Introduction
A Sonar System comprises beamforming and signal processing procedures to provide target detection and parameter estimation (e.g., bearing). The optimum sonar processor depends on the modeling assumptions that the environment admits. For example, it has been shown that the conventional line-array beamformer without shading is optimum for bearing estimation of a signal in a spatially white noise field [1] and that the variance of this estimate achieves the theoretical Cramer–Rao lower bound [2]. Studies have also been undertaken to determine the optimum configuration for the estimation of target bearing [3], [4] or detection [5] of a coherent signal in the presence of an incoherent noise field. It has been shown, in this instance, that a nearly optimal technique is the splitbeam processing method which provides improved broadband bearing resolution over conventional full-array beamforming [3], [6]. Nevertheless, the detection performance of the splitbeam processor is reduced as compared to the conventional full-array beamformer.