Implementation of a microphone array for improved noise path estimation in a feed-forward active noise control system

Abstract

Active noise control (ANC) systems detect incoming sound and generate an anti-noise signal to attenuate it through destructive interference. ANC systems are generally limited to low-frequency sounds because the sensors and actuators are close together, leaving insufficient time to react to high-frequency or impulsive sounds. To compensate, common applications, such as noise-cancelling headphones, rely on complete ear coverings to block out higher-frequency sound indiscriminately, potentially hindering communication and situational awareness. The present work proposes to use an external array of microphones surrounding the user to increase the distance between the sensors and the user to afford more time to generate anti-noise signals and provide better path estimation. Using time delay of arrival (TDOA), the array tracks the locations of the user and the noise sources in real-time. The ANC system uses a feed-forward Filtered-x Least Mean Squared (FxLMS) algorithm that adjusts the weights of the controller based on the TDOA path estimation instead of adapting the filters with an error microphone and feedback loop. Simulations of the proposed system and feedback FxLMS path estimation were conducted in MATLAB. Compared to the feedback FxLMS algorithm, the TDOA system yielded 11 ± 1 dB less root-mean-square error in the generated anti-noise signals.