A survey on active noise control techniques - Part II: Nonlinear systems

Lu Lu, Kai-Li Yin, Rodrigo C. de Lamare, Zongsheng Zheng, Yi Yu, Xiaomin Yang, Badong Chen

Abstract

Most of the literature focuses on the development of the linear active noise control (ANC) techniques. However, ANC systems might have to deal with some nonlinear components and the performance of linear ANC techniques may degrade in this scenario. To overcome this limitation, nonlinear ANC (NLANC) algorithms were developed. In Part II, we review the development of NLANC algorithms during the last decade. The contributions of heuristic ANC algorithms are outlined. Moreover, we emphasize recent advances of NLANC algorithms, such as spline ANC algorithms, kernel adaptive filters, and nonlinear distributed ANC algorithms. Then, we present recent applications of ANC technique including linear and nonlinear perspectives. Future research challenges regarding ANC techniques are also discussed.

The Challenge of Nonlinearity

While the Filtered-x Least Mean Square (FxLMS) algorithm is the industry standard for linear noise cancellation, its performance degrades significantly when encountering real-world nonlinearities. These distortions typically arise from three primary sources: the primary acoustic path, the secondary hardware path, and the internal components of the ANC system itself.

Table 2: Types and examples of nonlinearities encountered in ANC systems.

As shown in the taxonomy above, nonlinearities can range from high sound pressure levels in ducts to the saturation of power amplifiers and the chaotic nature of noise sources like grinders or fans. Linear models cannot effectively predict or cancel these complex signals, necessitating the development of Nonlinear ANC (NLANC) architectures.

System Architecture

A typical NLANC system modifies the standard linear control loop by introducing a "Nonlinear Expansion" block. This block processes the reference signal $x(n)$ into a higher-dimensional vector $\mathbf{x}_e(n)$, allowing the adaptive filter to capture nonlinear relationships between the noise source and the error microphone.

Figure 1: Schematic of a nonlinear active noise control system featuring nonlinear expansion.

In this architecture, the adaptive algorithm updates the digital filter weights based on the error signal $e(n)$ and the expanded reference signal. The goal is to generate an anti-noise signal $y(n)$ via the active loudspeaker that cancels the primary noise $d(n)$ at the error microphone location.

Algorithmic Families

Researchers have developed several distinct families of algorithms to handle these expansions:

  • Volterra Filters: These act as universal approximators using polynomial expansions. While mathematically robust, they are computationally expensive due to the "curse of dimensionality."
  • Functional Link Artificial Neural Networks (FLANN): These use trigonometric expansions (sine/cosine) to capture nonlinearities in a single-layer structure, offering a computationally efficient alternative to deep networks.
  • Kernel Adaptive Filters (KAF): By employing the "kernel trick," these algorithms map data into high-dimensional feature spaces where nonlinear noise can be treated linearly.
  • Spline ANC: This approach utilizes adaptive look-up tables and polynomial interpolation, making it highly suitable for real-time hardware implementation.
  • Heuristic Algorithms: Nature-inspired strategies like Particle Swarm Optimization (PSO) and Bacterial Foraging Optimization (BFO) are used to navigate complex error surfaces and avoid local minima.

Key Results and Applications

NLANC techniques have demonstrated significant success in high-stakes environments:

  • Medical Imaging: Hybrid algorithms have achieved over 40dB of attenuation for fMRI machine noise, significantly improving patient comfort.
  • Industrial Infrastructure: Specialized FxLMS variants have reduced the low-frequency hum of large power transformers by up to 15dB.
  • Hardware Efficiency: Modern VLSI implementations of NLANC structures have shown a 41% reduction in energy consumption per sample compared to traditional designs.

Future Research Challenges

Despite these advances, several frontiers remain. Computational complexity continues to limit the deployment of high-order Volterra or kernel filters in low-power consumer electronics. Furthermore, there is a need for stronger theoretical proofs regarding algorithm convergence in the presence of non-Gaussian, "spiky" noise. Future work will likely focus on "sparsification" techniques to reduce mathematical overhead and the integration of ANC into distributed Internet of Things (IoT) networks.

Summary updated

Abstract

Most of the literature focuses on the development of the linear active noise control (ANC) techniques. However, ANC systems might have to deal with some nonlinear components and the performance of linear ANC techniques may degrade in this scenario. To overcome this limitation, nonlinear ANC (NLANC) algorithms were developed. In Part II, we review the development of NLANC algorithms during the last decade. The contributions of heuristic ANC algorithms are outlined. Moreover, we emphasize recent advances of NLANC algorithms, such as spline ANC algorithms, kernel adaptive filters, and nonlinear distributed ANC algorithms. Then, we present recent applications of ANC technique including linear and nonlinear perspectives. Future research challenges regarding ANC techniques are also discussed.
BibTeX
@misc{lu2021survey,
  author = {Lu Lu and Kai-Li Yin and Rodrigo C. de Lamare and Zongsheng Zheng and Yi Yu and Xiaomin Yang and Badong Chen},
  title = {{A survey on active noise control techniques - Part II: Nonlinear systems}},
  year = {2021},
  eprint = {2110.09672},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2110.09672},
}
Figure