This research paper presents a novel Riemannian Gradient Descent (RGD) method for efficiently solving the Joint Blind Super-Resolution and Demixing (JBSD) problem in Integrated Sensing and Communication (ISAC) systems, offering theoretical guarantees of linear convergence and robustness to frequency separation conditions.
The proposed subband splitting technique effectively solves the block permutation problem in determined blind source separation by splitting the entire frequency range into overlapping subbands and sequentially applying a blind source separation method to each subband.
Die vorgestellte Methode nutzt Multi-Encoder-Autoencoder für die Blind Source Separation ohne starke Anwendungsspezifische Voraussetzungen.
提案された方法は、単一チャンネルの非線形混合物に対する盲目的なソース分離を可能にしました。