Active Noise Control: Architectures, Algorithms and its Parameters

This paper notes the widespread usage of the digital signal processing in equipment of modern communication systems, industrial electronic, and in medical and consumer devices. One of the areas of such processing is the active noise control (cancellation). It is used to cancel the acoustic noise in air by generating and radiation of similar noise with the opposite phase at a space point, where the noise cancellation is required. Today, there are three types of active noise cancellation devices: without feedback, with feedback, and with both feedbacks simultaneously. The paper considers an active noise cancellation device without feedback, as the simplest one which is designed to the cancellation of both the narrowband and broadband noises. The operation principles of such a device with traditional and modified architecture are described. It is shown that in the traditional architecture, only adaptive filters with simple adaptive algorithms of gradient descent based on the least square criterion can be used with a small convergence step-size, which guarantees a stable operation, but provides a slow convergence of the adaptive filter. In addition, such an architecture cannot use the adaptive filters with efficient recursive least square criterion adaptive algorithms. On the contrary, in the modified architecture the gradient descent algorithms can use an increased convergence step. This ensures a stable operation of the adaptive filters and their increased convergence. Also, in this architecture it is possible to use the recursive least squares adaptive algorithms. The simulation results confirm the operability of the considered devices of the active noise control and demonstrate the comparative efficiency of their operation. The best performance is demonstrated by the modified architecture with the adaptive filter based on the recursive least square algorithm, which provides the shortest transient response compared to the traditional and modified architectures with adaptive filters based on the gradient descent algorithms, and also provides a noise suppression of 51 dB under the considered simulation conditions, while in the architectures with adaptive filters based on the gradient descent algorithm this suppression reaches only 39...43 dB.