Semi-Blind Localization Using a Single Vector Sensor in Underwater Acoustic OFDM Systems

Using a single acoustic vector sensor, we address the 3D localization of an OFDM transmitter based on a three-path channel model typical in shallow water. To fully leverage the position information involved in the frequency-domain input-output relationship of pilot subcarriers, we first introduce the well-known matched field processing (MFP). Given the sensitivity of MFP to model mismatches, we also propose a novel semi-blind localization method by relaxing rigid assumptions on complex path amplitudes. We show that it is equivalent to a conditional maximum likelihood (CML) estimator and thus achieves an error performance comparable to the derived Cramér-Rao bound. Moreover, to alleviate the intensive computational burden of 3D search, we develop an iterative framework in which azimuth angle estimation and range-depth localization are performed alternately. For the proposed CML-type method, we execute simulations to verify its statistical efficiency and its robustness over MFP under various conditions. Tank experiments are conducted to further demonstrate algorithm practicality.