Low Complexity With MSF Approach to Acoustic Echo Canceller

In this paper we present the Multiple SubFilters (MSF) parallel structure, LMS adaptive algorithm, for acoustic echo cancellation. The performance of the MSF based echo canceller has been compared with conventional echo canceller which is a Single Long Filter (SLF) used to model the impulse response of the acoustic echo channel. The convergence performance of the MSF parallel structure is studied for common error and different error adaptive algorithms. Simulation results show that MSF with both adaptation algorithms provide better convergence over the SLF. However, the steady state error performance of the different error adaptation algorithm is poor as compared to that of common error as well as that of SLF adaptation algorithm. The sacrifice in steady state error performance can be compensated to achieve fast convergence by introducing independent adaptation step size as there is decoupled weight updation equation of each sub-filter. The combination of both the adaptation algorithm in MSF can achieve a trade-off between steady state and convergence rate. A scheme named as selective coefficient update (SCU) algorithm is proposed in which only few coefficients are updated at each iteration.