Geometrically Constrained Blind Moving Source Extraction based on Constant Separation Vector and Auxiliary Function Technique

This paper focuses on the permutation problem in Blind Source Extraction (BSE) by employing beamforming-based Geometrical Constraints (GC). Specifically, we focus on the recently proposed auxiliary function-based Independent Vector Extraction (Aux-IVE) with Constant Separation Vector (CSV) mixing model. Building upon this mixing model, we provide the source signals’ spatial information, utilizing GC based on beamforming. We facilitate the extraction of moving target source and improve the extraction performance. Furthermore, we discuss the effect of the number of constraints on the extraction performance. Experimental results confirmed that our proposed method, GC-CSV-Aux-IVE, exhibits superior extraction performance and achieves nearly 100% accuracy in extracting the moving target source.