FXECLMS Algorithm for Active Acoustic Windows: an Experimental Report

Abstract

Active noise control (ANC) systems have become attractive solution for controlling the environmental noise in open windows. However, few works have demonstrated the benefits of using ANC systems in active acoustic windows since their implementation depends on several variables, such as the control source configuration (quantity and position of speakers), the adaptive control algorithm and the computational capabilities of the embedded devices. Specifically, the computational complexity of the adaptive control plays an important role in the development of practical real-time ANC applications. Until date, several works have presented theoretical models for ANC systems. Nevertheless, they do not consider the computational capabilities of the embedded devices to efficiently implement their adaptive algorithms, which limits its use in practical ANC applications. This paper presents for the first time a multichannel adaptive algorithm based on a filtered-x error coded least mean square (FXECLMS) and its efficient implementation to be used in an open-window ANC system. From the experimental results, we discuss the limitations, challenges and provide potential solutions in practical ANC implementations using advanced embedded devices.