A cascaded IIR–FIR adaptive ANC system with output power constraints

• In our paper, a new power constrained algorithm, consisting of a cascaded IIR and FIR filter, was developed.
• The filter gradient equations, and system requirements for convergence, were presented.
• Additionally, a simple power control recursive equation, based on the principle of minimal disturbance, was developed.
• The performance of the new algorithm was compared to the leaky LMS.
• New algorithm provides faster convergence, improved frequency response performance, and explicit control of output power constraint level.

This paper develops a new power-constrained adaptive filter architecture and algorithm intended for use in active noise control systems where the noise source consists of a narrowband (periodic) component and an uncorrelated wideband component (e.g., background noise). In addition to using a transducer to generate an anti-noise output to destructively cancel the noise source, we impose the additional requirement of limiting the maximum output power level to prevent system overdrive. This is accomplished by separating the narrowband and wideband input components, and then using a penalty function formulation to place output power constraints on the narrowband portion. The new filter architecture offers improved performance compared to previous methods, including explicit output power control and frequency-selective power limiting.