Virtual Sensor ANC with the Noise Source Searching

Active noise control (ANC) with a virtual sensor is an attractive method because in most case we cannot put sensors at the target point. However, the conventional method is basically specified for a one-dimensional case, and the distance between real and virtual microphones is small. To overcome this problem, this paper presents ANC using a virtual sensor for a two-dimensional case and the method for expanding the distance between the real and virtual microphones. The mechanism of the proposed method is as follows. Firstly, two microphones are used to search where the noise comes from. Then, using the objective frequency which is assumed to be known, the phase and the amplitude those are related to a disturbance source are estimated. Furthermore, the phase and the amplitude at the target point is computed based on the assumption that the noise sound decreases in inverse proportion to the bth power of the distance from a noise source point (the variable b changes from 0 to 1). Finally, the adequate output for suppressing the noise at the virtual sensor point is generated from the control source speaker, based on the adaptive filtering scheme such as a filtered-x LMS algorithm. In this paper, mathematical description of the above procedure is discussed, and the some numerical results are shown that clarify the validity of the proposed method.