Comparing arctan and maximum likelihood methods for bearing estimation

One well-known system for estimating the bearing to an underwater acoustic source consists of two crossed dipole sensors and an omni-directional sensor. At DRDC Atlantic, several bearing estimation methods for this three-channel system in the presence of ocean noise have been examined with the goal of minimizing the bearing error. However, any practical bearing-estimation algorithm must remain efficient enough for implementation in a real-time sonar processor. In this paper, we discuss some of the different methods used and their trade-offs based on estimate accuracy and computational efficiency.