A switching variable step size affine Projection adaptive algorithm for acoustic echo cancellation

摘要

As the technology advances, adaptive filtering algorithms used for the implementation of Acoustic Echo Cancellation (AEC) have a wider choice. The performance of AEC mainly depends on the effectiveness of the adaptive filter. The signals associated with such applications are highly correlated and its characteristics changes quite fast, which demands that the adaptive filters used should have high convergence rate and low misalignment with lesser execution time and it should also track the optimum solution. In this paper, a Switching Variable Step Size Affine Projection Adaptive Algorithm (SVSS APA) is proposed, which is an efficient combination of Variable Step Size Affine Projection Algorithm (VSS APA) and Affine Projection Algorithm (APA). Algorithm starts with VSS APA. As the weight converges, it switches on to APA. Thus in the new algorithm execution time can be minimized. SVSS APA is applied to a typical AEC scenario where environment slowly changes with time. Simulation results show that the new algorithm has performance almost equal to VSS APA with lesser execution time. In AEC scenario during the double talk condition, where talkers on the both sides talks simultaneously, double talk detector is used to control the adaptation.