Self-adapting variable step size strategies for active noise control systems with acoustic feedback

Abstract Self-adapting variable step size (SAVSS) normalized least mean square (NLMS) algorithms are derived and analysed for the active noise control systems with acoustic feedback path. The objective of the proposed SAVSS scheme is to resolve the conflicting requirements of rapid convergence and low mis-adjustment. A tuningless power scheduling scheme is proposed that varies the gain according to the output of the predictor filter, and switches to a low gain on the basis of a minimum criteria. In the steady state, a low gain auxiliary noise significantly improves noise reduction performance, and the update of the feedback compensation filter and predictor filter is discontinued to reduce computations. Simulations are performed with narrowband noises to validate the improved performance of the proposed method in comparison with established state-of-the-art methods.