Preventing output saturation in active noise control: An output-constrained Kalman filter approach

Junwei Ji, Dongyuan Shi, Boxiang Wang, Xiaoyi Shen, Zhengding Luo, Woon-Seng Gan

TL;DR

This paper introduces a novel Kalman Filter with Output Power Constraint (KF-OPC) for active noise control (ANC). By intelligently re-scaling the disturbance signal, KF-OPC effectively prevents the control signal from exceeding hardware limitations, thus avoiding output saturation and its associated non-linear distortions, while maintaining fast convergence.

What Problem Does This Paper Address?

Active Noise Control (ANC) systems aim to cancel unwanted noise by generating an anti-noise signal. While Kalman Filter (KF)-based ANC offers superior tracking and faster convergence compared to traditional methods like LMS, it faces a critical challenge in environments with extremely high noise levels. The power of the generated anti-noise signal can exceed the physical limitations of the audio hardware (e.g., amplifiers, speakers). This "output saturation" leads to clipping of the control signal, introducing undesirable non-linear distortions and potentially causing the entire ANC system to become unstable or diverge. Existing methods to address this in LMS-based systems often compromise performance or introduce other issues. This paper seeks to provide a robust solution for KF-based ANC that prevents saturation without sacrificing its inherent advantages.

Methods

The core innovation is the Kalman Filter with Output Power Constraint (KF-OPC). The standard KF-based ANC system uses a state-space model where the control filter is estimated. The disturbance signal, treated as a measurement, is used to update this estimate.

The KF-OPC approach modifies this by introducing a constraint factor, denoted by 'α'. This factor is applied to the disturbance signal before it's used as a measurement in the Kalman filter update equation:

ŵ(n|n) = ŵ(n|n-1) + K(n) [αd(n) - x'ᵀ(n)ŵ(n|n-1)]

The constraint factor 'α' is derived based on the system's rated output power (p₀), the secondary path gain (G_s), and the disturbance signal's power (δ). The goal is to ensure that the power of the anti-noise signal, after passing through the secondary path, does not exceed the rated output power. By re-scaling the measurement (the disturbance signal), the KF-OPC indirectly constrains the output power of the control signal, thereby preventing saturation. When the system's output power is sufficient, α is set to 1, effectively reverting to the standard KF-ANC.

Datasets

Not discussed in the paper. The paper relies on simulated primary and secondary paths and noise signals for evaluation.

Key Insights and Ideas

  • Saturation is a practical bottleneck: While KF-ANC is theoretically powerful, its practical application is limited by hardware constraints that can lead to saturation and instability.
  • Measurement re-scaling is key for KF: Unlike FxLMS methods that modify the update gradient, KF-ANC's measurement-driven update allows for a more direct approach: re-scaling the measurement itself to control output power.
  • Intelligent constraint factor: The proposed constraint factor 'α' is not arbitrary but derived from system parameters, aiming for an optimal balance between constraint and performance.
  • Stability and linearity preserved: The KF-OPC not only prevents saturation but also ensures the anti-noise signal remains linear with the interference signal, maintaining system stability.

Key Results

The paper presents simulation results demonstrating the effectiveness of KF-OPC:

  • Tonal Noise: In a scenario with a 400Hz tonal noise, the conventional KF-ANC produced significant harmonics due to output saturation, leading to inferior noise reduction. KF-OPC, however, successfully suppressed these harmonics and achieved better noise reduction. Figures show the control signal exceeding the rated power in KF-ANC, leading to clipping, while KF-OPC keeps it within limits.
  • Broadband Noise: For broadband noise (200-2000Hz), KF-OPC maintained fast convergence, comparable to standard KF-ANC, and significantly better than MOVFxLMS. Crucially, it kept the output power within the rated limit (0.8), preventing saturation and producing a more linear anti-noise signal.
  • Real Noise and Path: When tested with a real primary path, secondary path, and compressor noise, KF-OPC achieved good noise reduction while respecting the output power constraint. It sacrificed a small amount of noise reduction compared to unconstrained KF-ANC but ensured stability and was faster to converge than MOVFxLMS.

Limitations and Future Work

  • Real-world path estimation: The paper uses simulated or simplified models for primary and secondary paths. Further research could explore the impact of imperfect real-world path estimation on the KF-OPC's performance.
  • Computational complexity: While KF-OPC is presented as practical, the computational cost of calculating the constraint factor and its integration into the KF update should be considered for real-time embedded systems.
  • Generalizability: While demonstrated for tonal and broadband noise, exploring performance with more complex and non-stationary noise environments would be valuable.
  • Adaptive secondary path: The current approach assumes a known or estimated secondary path gain. Investigating how KF-OPC performs with an adaptive secondary path would be a logical extension.

Summary updated

Abstract

The Kalman filter (KF)-based active noise control (ANC) system demonstrates superior tracking and faster convergence compared to the least mean square (LMS) method, particularly in dynamic noise cancellation scenarios. However, in environments with extremely high noise levels, the power of the control signal can exceed the system's rated output power due to hardware limitations, leading to output saturation and subsequent non-linearity. To mitigate this issue, a modified KF with an output constraint is proposed. In this approach, the disturbance treated as an measurement is re-scaled by a constraint factor, which is determined by the system's rated power, the secondary path gain, and the disturbance power. As a result, the output power of the system, i.e. the control signal, is indirectly constrained within the maximum output of the system, ensuring stability. Simulation results indicate that the proposed algorithm not only achieves rapid suppression of dynamic noise but also effectively prevents non-linearity due to output saturation, highlighting its practical significance.
BibTeX
@inproceedings{ji2025preventing,
  author = {Junwei Ji and Dongyuan Shi and Boxiang Wang and Xiaoyi Shen and Zhengding Luo and Woon-Seng Gan},
  title = {{Preventing output saturation in active noise control: An output-constrained Kalman filter approach}},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
  year = {2025},
  doi = {10.1109/icassp49660.2025.10888338},
}
Figure