Abstract

All adaptive algorithms suffer stability issues when employed for the impulsive noise control under the domain of active noise control (ANC) systems. There is a dire need of investigations to overcome this limitation for the impulsive noise, a robust adaptive algorithm is proposed in literature. In the first part of paper, this robust adaptive algorithm is tested for the first time under ANC environment for impulsive noise cancellation and thus, a new ANC algorithm named filtered-x least cosine hyperbolic (FxLCH) algorithm is presented. Simulations are carried out to validate the improved performance of proposed FxLCH algorithm where the impulsive noise realizations are generated by symmetric α-stable distributions. Moreover, the proposed solutions perform better than the standard filtered-x least mean square (FxLMS) algorithm including its variants, and it shows better stability and converges faster than its competitors. Robustness of the algorithm is a constraint in the presence of high impulsive noise. To overcome this problem and to enhance the robustness of proposed FxLCH algorithm, two modifications are suggested. First proposed modification clips the reference and error signals (CFxLCH algorithm), while the second modification integrates already reported normalized step size with FxLCH (MFxLCH) algorithm. The performance of suggested MFxLCH algorithm is validated by extensive simulations. The results exhibited that MFxLCH algorithm acts as a trade-off between FxLMS and filtered-x recursive least square (FxRLS) family algorithms. It has shown better convergence speed than that of FxLMS family algorithms and can approach steady state error as of FxRLS family with almost same computational complexity as of FxLMS family algorithms.

中文翻译：

---

主动噪声控制脉冲噪声的复杂性较低的解决方案

摘要

当在有源噪声控制（ANC）系统的域内用于脉冲噪声控制时，所有自适应算法都会遇到稳定性问题。迫切需要进行研究以克服这种对脉冲噪声的限制，文献中提出了一种鲁棒的自适应算法。在论文的第一部分中，首次在ANC环境下测试了该鲁棒的自适应算法以消除脉冲噪声，因此，提出了一种新的ANC算法，称为滤波x最小余弦双曲（FxLCH）算法。进行仿真以验证所提出的FxLCH算法的改进性能，该算法通过对称的α稳定分布生成脉冲噪声。此外，提出的解决方案比包括它的变体的标准滤波x最小均方（FxLMS）算法性能更好，并且与竞争对手相比，它具有更好的稳定性和收敛速度。该算法的鲁棒性是存在高脉冲噪声时的一个约束条件。为了克服此问题并增强所提出的FxLCH算法的鲁棒性，提出了两种修改方法。第一个提出的修改是剪辑参考信号和误差信号（CFxLCH算法），而第二个修改则是将已报告的归一化步长与FxLCH（MFxLCH）算法集成在一起。通过广泛的仿真验证了建议的MFxLCH算法的性能。结果表明，MFxLCH算法是FxLMS和滤波x递归最小二乘（FxRLS）系列算法之间的折衷方案。