Underdetermined passive localisation of emitters based on multi-dimensional spectrum estimation techniques

One-step passive localisation methods are verified to outperform the conventional two-step methods in terms of estimation accuracy and identifiability. However, the existing one-step algorithms still cannot work in the scenario where the number of emitters exceeds the total number of sensors from all the base stations. Based on a spatio-temporal processing procedure, the authors propose a novel localisation model, which has the form of multi-dimensional harmonic retrieval. By utilising multi-dimensional spectrum estimation techniques, the underdetermined localisation problem can be handled thanks to the increase of model's degree of freedom. To further improve the identifiability, a nested-array-based localisation model is also given based on the multi-dimensional processing framework. Simulation results demonstrate that the novel-model-based Capon algorithm can achieve higher localisation accuracy without the prior knowledge about the number of emitters, and is robust to the correlated noise.