Mixed Incoherent Far-Field and Near-Field Source Localization under Uniform Circular Array

Xiaolong Su, Zhen Liu, Xin Chen, Xiang Li

TL;DR

This paper introduces a novel phase-based algorithm for precisely locating multiple sound sources that are simultaneously near and far from a uniform circular array (UCA). By cleverly using phase differences between specific sensor pairs, the algorithm can accurately determine the direction and distance of these mixed sources, outperforming existing methods in both accuracy and computational efficiency.

What Problem Does This Paper Address?

Locating sound sources using sensor arrays is crucial in many applications, from radar and sonar to audio processing. Traditional methods often struggle when dealing with a mix of sources that are either close to the array (near-field) or very far away (far-field). This is because the wavefronts from near-field sources are curved, while those from far-field sources are essentially planar. Most existing super-resolution algorithms are designed for one type of source or the other, making them ill-suited for mixed scenarios. This paper tackles the challenge of accurately and efficiently locating multiple, independent (incoherent) sources that can be a combination of near-field and far-field types, all observed by a uniform circular array (UCA).

Methods

The proposed algorithm leverages the unique geometry of the UCA and a phase-based approach to overcome the limitations of existing methods. Here's a breakdown of the key steps:

  • Decoupling Angles: The core innovation lies in calculating the phase differences between "centro-symmetric" sensors (pairs of sensors diametrically opposite each other on the UCA). These phase differences are shown to be dependent only on the source's angle (azimuth and elevation) and not its range.
  • Least Squares for Angles: Since the phase differences form a set of equations with only angular unknowns, a least squares method is employed to estimate the 2D Direction of Arrivals (DOAs) for all sources simultaneously.
  • 1-D MUSIC for Range: Once the angles of the mixed sources are estimated, the algorithm uses these angles as input to a 1-D MUSIC (Multiple Signal Classification) method. This is applied to estimate the ranges of the near-field sources.
  • Range Spatial Spectrum for Classification: A "range spatial spectrum" is constructed. Peaks in this spectrum indicate near-field sources, allowing for their identification and range estimation. Far-field sources, by contrast, result in a divergent spectrum.
  • Improved Frequency Estimation: To enhance accuracy, the paper proposes using adjacent spectral lines from the Fast Fourier Transform (FFT) to refine frequency estimation, mitigating potential errors when a source frequency falls between FFT bins.

Datasets

Not discussed in the paper. The evaluation is based on simulated data.

Key Insights and Ideas

  • Centro-symmetric pairs are key: The clever use of diametrically opposite sensor pairs in a UCA provides a direct link to source angles, simplifying the problem.
  • Phase holds the angular information: The paper demonstrates that phase differences, not just amplitude, are critical for distinguishing and locating sources in mixed scenarios.
  • Separation of concerns: The algorithm elegantly separates the problem into angle estimation first, followed by range estimation, which is more manageable than trying to solve for all parameters simultaneously.
  • MUSIC for range is effective: Adapting the powerful MUSIC algorithm to a 1D search for range, after accurate angle estimation, proves to be a robust approach for near-field sources.
  • Range spectrum as a classifier: The divergence of the far-field spatial spectrum versus the distinct peak of the near-field spectrum provides a clear mechanism for distinguishing between source types.

Key Results

The paper presents simulation results demonstrating the effectiveness of the proposed algorithm.

  • Superior Accuracy: Compared to the TSMUSIC method (a state-of-the-art approach), the proposed algorithm shows higher accuracy in estimating azimuth and elevation angles, especially at lower Signal-to-Noise Ratios (SNRs).
  • Improved Range Estimation: The near-field source range estimation also outperforms TSMUSIC.
  • Computational Efficiency: The proposed method is significantly faster, with a CPU time of 0.26 seconds per run compared to 50.95 seconds for TSMUSIC, making it more practical for real-time applications.
  • Robustness to Frequency Error: The algorithm shows reasonable performance even with small frequency estimation errors (within 2 kHz), with angle errors remaining below 1 degree and range errors below 0.2 meters.

Limitations and Future Work

  • Frequency Estimation Dependency: The paper acknowledges that the localization performance is related to the accuracy of frequency estimation. While they propose an improvement, significant frequency errors can still impact results.
  • Phase Ambiguity: The paper mentions that phase ambiguity is assumed to be addressed by the condition R < c/4f, and that this has been handled in prior work. However, the inherent complexity of phase ambiguity in real-world scenarios might still pose challenges.
  • Future Work: The authors suggest that the algorithm can be extended to handle multiple near-field or multiple far-field sources, implying that the current work focuses on a mixed scenario with potentially multiple sources of each type. Further exploration into more complex interference scenarios or different array geometries could be future directions.

Summary updated

Abstract

A high-accuracy algorithm is presented for the localization of mixed incoherent near-field and far-field narrow-band sources under uniform circular array (UCA). Herein, considering that it is difficult to classify the mixed sources, we first decouple mixed sources’ angles and ranges by calculating centro-symmetric sensors’ phase differences. Then, as the phase differences including only sources’ angles can be transformed as indefinite equations, each source’s azimuth angle and elevation angle are obtained by performing the least squares method. After that, on the basis of the estimated angles of the mixed sources, one-dimensional (1-D) multiple signal classification (MUSIC) method and corresponding spatial spectrum are utilized to identify the mixed sources and estimate the ranges of the near-field sources. Finally, simulation and comparison results verify the superior performance of our proposed algorithm.
BibTeX
@inproceedings{su2018mixed,
  author = {Xiaolong Su and Zhen Liu and Xin Chen and Xiang Li},
  title = {{Mixed Incoherent Far-Field and Near-Field Source Localization under Uniform Circular Array}},
  booktitle = {Sensors},
  year = {2018},
  url = {https://openreview.net/forum?id=LCKrZkH-uX_},
}
Figure