Performance Comparison of APA, JO-APA, VSS-APA, and VR-APA for Acoustic Echo Canceller

Well-known normalized least mean square (NLMS) algorithm for system identification problems, does not provide a satisfactory convergence rate while dealing with correlated inputs. So, for the applications such as acoustic echo cancellation (AEC) where signals are highly correlated and have echo path of long duration, algorithms that can provide a better convergence rate and low misadjustment are preferred. In this paper, performance comparison of affine projection algorithm (APA) and its variants have been studied for AEC application. Simulation results for APA, joint-optimized APA(JO-AP A), variable step-size APA (VSS-AP A), and variable regularized APA (VR-AP A) at different values of projection order have been obtained. Based on the simulation study it is found that JO-AP A provides a better convergence rate while maintaining lower misalignment specifically at the higher values of projection order.