Semi-Blind Channel Estimation and Near-Field Source Localization with Extremely Large Antenna Arrays

This paper proposes a low-complexity semi-blind solution for channel estimation when using Extremely Largescale Antenna Arrays (ELAA), without prior knowledge of the considered channel model. More precisely, the new method combines pilots and a subspace approach for efficient and robust channel estimation, considering Intersymbol Interference (ISI). To reduce the costs, a Divide and Conquer (DAC) methodology is considered for a fast and parallelizable subspace estimation. For the positioning objective, physical localization parameters (i.e., angles and range) are accurately obtained using a simple leastsquares fitting applied to the first channel tap (i.e. the shortest path), assuming a line of sight situation. A simulation-based study is provided to assess the potential of our semi-blind method.