A Channelization-Based DOA Estimation Method for Wideband Signals

Rui Guo, Yue Zhang, Qianqiang Lin, Zengping Chen2016 · DOI

TL;DR

This paper introduces a novel "channelization-based" approach for estimating the Direction of Arrival (DOA) of wideband signals. By digitally splitting the wideband signal into multiple narrowband sub-channels, the method improves signal-to-noise ratio (SNR) and allows for more accurate DOA estimation compared to existing techniques, especially in challenging real-world environments.

What Problem Does This Paper Address?

Locating the direction of wideband signals is crucial in applications like radar, sonar, and communications. Traditional narrowband DOA estimation methods (like MUSIC and ESPRIT) fail with wideband signals because the phase difference between sensors depends not only on the DOA but also on the signal's frequency. Existing wideband methods, such as Incoherent Signal Subspace Method (ISM) and Test of Orthogonality of Projected Subspaces (TOPS), have limitations. ISM can struggle with low SNR and coherent sources, while both methods can be sensitive to environmental conditions and require careful parameter tuning. The paper aims to develop a more robust and hardware-friendly wideband DOA estimation method.

Methods

The core of the proposed method is a digital channelization receiver. This receiver takes the wideband signal from an array of sensors and splits it into multiple narrowband sub-channels. The key steps are:

  • IF Digitization and Digital Channelization: The wideband radio frequency (RF) signals are digitized directly. A digital channelization receiver then partitions these signals into several frequency channels. This process is illustrated in Figure 2, where a filter bank is used to achieve this separation.
  • Channel Selection: Not all sub-channels will contain the wideband signal of interest. The method identifies the relevant sub-channels by comparing their power output, assuming signals within the wideband signal's bandwidth will have higher power.
  • DOA Estimation: Two channelization-based methods are proposed:
  • Channelization-ISM: This method applies standard narrowband ISM to each selected sub-channel independently. The final DOA is then obtained by averaging the DOA estimates from all selected sub-channels (arithmetic or geometric mean).
  • Channelization-TOPS: This method leverages the orthogonality between signal and noise subspaces within each sub-channel. It uses a transformation matrix to align array manifolds across different frequencies and then estimates DOA by finding the angle that minimizes the singular value of a constructed matrix.

The paper emphasizes that this approach isolates signals in different bandwidths effectively, improves the output SNR by filtering out noise outside the signal bandwidth, and its parallel processing architecture makes it suitable for hardware implementation.

Datasets

The paper does not introduce new datasets. Instead, it relies on simulated data for its experiments and uses a custom-built wideband digital array radar (DAR) test-bed for experimental validation.

Key Insights and Ideas

  • Channelization as a Pre-processor: The digital channelization receiver acts as a powerful pre-processing step, effectively converting a wideband DOA problem into multiple narrowband DOA problems. This simplifies the estimation process.
  • Improved SNR: By selecting only the sub-channels containing the signal and discarding others, the method inherently improves the SNR, leading to better performance, especially at lower SNRs.
  • Hardware Efficiency: The parallel nature of channelization makes it well-suited for implementation on hardware platforms like FPGAs, enabling real-time processing.
  • Robustness to Environment: The channelization approach appears to be more robust to environmental factors compared to some conventional methods, as demonstrated by experimental results.
  • Dynamic Range Improvement: The method shows better performance in resolving weak signals, indicating an improved dynamic range compared to standard ISM and TOPS.

Key Results

  • Simulation Results: Both Channelization-ISM and Channelization-TOPS consistently outperformed their conventional ISM and TOPS counterparts in terms of Root Mean Square Error (RMSE) across various SNR levels and numbers of snapshots. The proposed methods approached the Cramér-Rao Lower Bound (CRLB) more closely.
  • Resolution Probability: For weak signals, the Channelization-ISM method showed significantly better resolution probability compared to the standard ISM method, especially at lower SNRs.
  • Experimental Results: Experiments conducted in a microwave anechoic chamber validated the simulation findings. The channelization-based methods achieved lower RMSEs and provided DOA estimates closer to the true DOA compared to conventional ISM and TOPS. Table 2 summarizes the mean deviation, showing Channelization-TOPS with an RMSE of 0.6792 degrees and Channelization-ISM with 0.9103 degrees, both significantly better than TOPS (0.6121 degrees) and ISM (3.2509 degrees).

Limitations and Future Work

  • Hardware Constraints: While the architecture is hardware-friendly, the number of channels used in digital channelization is ultimately limited by hardware resources, which can affect performance.
  • Array Manifold Imperfections: The paper assumes a deterministic and constant array manifold. Performance degradation can occur due to mutual coupling, sensor position errors, or gain/phase uncertainties. The authors mention that calibration algorithms exist for these issues but do not implement them in this work.
  • Future Work: The authors suggest that the channelization receiver architecture can be extended to isolate channels with different bandwidths using multistage DFT filters, which would be beneficial for more complex scenarios with coexisting signals of varying bandwidths. Further research could also explore the impact of array manifold imperfections and develop compensation strategies.

Summary updated

BibTeX
@misc{guo2016channelization,
  author = {Rui Guo and Yue Zhang and Qianqiang Lin and Zengping Chen},
  title = {{A Channelization-Based DOA Estimation Method for Wideband Signals}},
  year = {2016},
  doi = {10.3390/s16071031},
}

Loading related content...

Figure