Controlling auxiliary-noise-power for efficient neutralization of acoustic feedback in active noise control systems

The focus of this paper is acoustic feedback neutralization during online operation of single-channel active noise control (ANC) systems. Invasive techniques, in which additional auxiliary noise is injected, for online feedback path modeling and neutralization (FBPMN) comprise of three adaptive filters: 1) ANC filter, 2) Linear prediction filter, and 3) Feedback path modeling filter. In the existing methods, for online FBPMN, auxiliary noise with fixed variance is injected in all operating conditions, thus degrades the NRP of ANC systems at steady-state. In this paper a gain scheduling scheme is proposed to vary the auxiliary-noise-power (ANP). In proposed scheme the ANP is varied based on the convergence of both FBPMN filter and linear prediction filter. Simulation results show that proposed scheme can better achieve the conflicting requirements of fast convergence of FBPMN filter, and reduced steady-state residual error power at error microphone, compared to existing methods.