Active Noise Control for Augmented Reality Glasses using Virtual Sensing with Directional Robustness
Active noise control (ANC) is an established technique for mitigating acoustic noise, successfully implemented in various commercial applications. However, the application of ANC to augmented reality (AR) glasses remains an emerging area, where a key challenge is the lack of a microphone at the user's ear. This paper first validates the necessity of virtual sensing techniques for this specific geometry and investigates the Auxiliary Filter (AF) method as a potential solution. We present a theoretical analysis of the method's directional sensitivity, enabling the prediction of the system's robustness when the noise direction deviates from the calibration angle. Simulation results confirm that while a fixed AF is effective under matched conditions, it suffers significant degradation under mismatch, consistent with the theoretical predictions. To address this limitation, we propose and evaluate a solution using a finite bank of auxiliary filters that are selected based on the noise source direction of arrival. The results demonstrate that this filter bank approach significantly recovers performance compared to a fixed filter, offering a practical alternative to continuous recalibration in dynamic acoustic environments.