Analyzing Convergence of Distributed FxLMS Algorithm

分布式FxLMS算法的收敛特性分析

School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an 710072, China

Received: 27 June 2019

Abstract

The distributed FxLMS algorithm based on acoustic sensor networks can solve the problem of the increased computational complexity caused by a large-scale active noise control (ANC) system. Firstly, the distributed algorithm is deduced theoretically. In the ANC process based on sensor networks, the iteration of partial weight coefficients is replaced by the real-time communication of weight coefficients among nodes. Then the convergence characteristics of the distributed algorithm in its mean and mean square senses are analyzed and verified with simulation. Finally, the noise reduction effects of different algorithms are compared. The results show that the distributed algorithm, with its computational complexity reduced, can achieve the same noise reduction effects as the centralized ANC system.

摘要

基于声学传感器网络的分布式FxLMS算法可以解决大规模有源噪声控制系统带来的运算量和复杂度剧增的问题。首先对分布式算法进行了理论推导。在基于传感器网络的ANC过程中，通过各节点之间权系数的实时通信代替了部分权系数的迭代。分析了算法平均以及均方意义下的收敛特性，并利用仿真进行了验证。最后，比较了不同算法的降噪效果。结果表明，在减少运算量的基础上，分布式算法能够获得与集中式ANC系统一致的性能。