Active control of outgoing noise fields in rooms

Abstract

Active noise control is a strategy to suppress a noise by superimposing it with a carefully designed secondary noise. The strategy has been under research over the past half century with active noise control aided devices surging on the market over the last decade. However, up to now, the most successful applications of active noise control are still limited to the single channel systems, where noises propagate in ducts or in the human ear canals.

Many researchers attempted to extend the application of active noise control to spatial noise fields, such as controlling the tire rolling noise in cars, the ventilation noise in workplaces, or the pump engine noise outdoors, which account for the majority of noises we encounter in our everyday lives. They developed spatial active noise control systems based on room modes, spherical modes, or the Helmholtz integral equation. The attempts have found limited success in the real world because of two problems. The first is that a spatial noise field is the complicated interaction of a number of noise sources with the environment, both of which can be non-stationary and time-varying. This problem makes it extremely difficult to obtain clean reference signals for spatial active noise control systems. The second is that due to the lack of a time-domain spatial sound field control theory, the existing spatial active noise control systems process the acoustic quantities in the time-frequency domain. The time-frequency domain processing introduces the frame delay and thus probably makes the systems violate the causal control constraint.

This thesis proposes an outgoing noise field control system based on the frequency-domain sound field separation method. The method decouples the outgoing field (due to the noise sources) from the incoming field (due to the environment) on a sphere surrounding the noise sources. By canceling the outgoing field only, the proposed system reduces the noise entirely in a room without estimating the secondary paths in real-time and with negligible influence on the desired sound field in the room.

This thesis further derives a time-domain sound field separation method, based on which a low latency outgoing field control system with random noise field cancellation capacity is developed. Multiple circular arrays of vector sensors for three-dimensional sound field analysis are developed based on the time-domain method. The designed arrays have a compact geometry, and thus can be integrated with small sized wearable devices and provide them with real-time sound field analysis capacity.