Nonlinear Active Noise Control using partial-update Filtered-s LMS algorithm

Active Noise Control (ANC) has been gaining an increasing interest in recent years. Much attention has been devoted to design of efficient control algorithms, enabling noise reduction at a high level, with computational load acceptable by currently available electronics. Among different approaches to noise control, employment of vibrating plates as secondary sources or as active barriers is particularly attractive. This idea can also be applied to control vibration of a whole device casing in order to reduce noise propagating from the device. A system using the vibrating plates may have a nonlinear nature. Linear control algorithms commonly used for ANC are exposed to a decrease of performance, when facing the nonlinearities. This happens due to impossibility of reducing generated additional frequeny components, not present in the reference signal. In the literature there are many nonlinear algorithms, mostly based on the popular LMS algorithm with a reference signal filtering (Fx-LMS). These algorithms involve a significantly greater computational complexity, and thus their use requires a more efficient hardware. It can still be insufficient in case of multichannel structures. This paper focuses on reducing the computational complexity of the Filtered-s LMS algorithm (Fs-LMS) by using a partial update of the control filter weights. The proposed solution is then verified in simulations by using data from a real active casing.