Implementation of distributed multichannel active noise control with intermittent communication and coprocessor assisted data combination

Multichannel active noise control (MCANC) with distributed control offers a more computationally efficient alternative to traditional centralized schemes by leveraging multiple low-cost controllers. In this distributed MCANC (DMCANC) architecture, each node, comprising an error microphone, a loudspeaker, and a local controller, collaborates through mutual data exchange to refine global control filters and improve overall noise attenuation. However, communication network issues may affect the system’s stability. Therefore, an intermittent communication DMCANC (IC-DMCANC) system is developed to improve the system’s robustness, where each ANC node exchanges data at flexible intermittent and possibly heterogeneous communication instants, determined by hardware capability and network conditions. Moreover, with the aim of preventing instability caused by non-communication phases and transmission delays, the weight-constrained filtered reference least mean square (WCFxLMS) algorithm is used for iterative updating of each node. To further alleviate the computational and communication burden of each node, a coprocessor is introduced to be responsible for the data fusion and then send the combined result to each node for updating the parameters in the WCFxLMS. Thus, the proposed IC-DMCANC with coprocessor assisted (IC-DMCANC-CPA) system is more efficient and easier to implement. The numerical simulation results validate the satisfactory noise reduction performance of the proposed method compared to the centralized method. Furthermore, the real-time experiments demonstrate the effectiveness and robustness of the IC-DMCANC-CPA system in a real scenario.