A low complexity delayless frequency domain feedback canceller for hearing aids

The adaptive feedback canceller (AFC) based on the least mean square (LMS) or normalized LMS (NLMS) algorithm exhibits poor convergence and high computational complexity when used for modelling an impulse response with large number of coefficients. Also, its convergence rate degrades for colored signal as input. The usage of frequency domain LMS (FLMS) algorithm for developing AFC results in enhanced convergence and low computational burden but it suffers from inherent block processing delay or latency. Though the partitioned block FLMS (PBFLMS) algorithm is successful in reducing the delay to some extent but cannot completely eliminate it. In order to overcome this problem, this paper introduces a low complexity delayless PBFLMS (D-PBFLMS) algorithm based AFC for hearing aids. The simulation results show the effectiveness of the proposed AFC in terms of the computational complexity, convergence rate and feedback cancellation accuracy in the presence of white noise and speech segment as input.