A Cascaded Semi-Blind Source Separation Method for Joint Acoustic Echo Cancellation, Interference Suppression, and Noise Reduction

Acoustic echo cancellation (AEC), interference suppression, and noise reduction play important roles in full-duplex communication. However, conventional systems that cascade adaptive filters and beamformers often experience a degradation in performance during doubletalk situations. To tackle this issue, this paper presents a multichannel semi-blind-source-separation (SBSS) method that combines the element-wise iterative source steering (EISS) AEC algorithm with a geometrically constrained independent vector analysis source extraction algorithm for full-duplex communications. Simulation results confirm the effectiveness of the proposed method.