Robust Beamforming Against Direction-of-Arrival Mismatch Using Subspace-Constrained Diagonal Loading

Yueh‐Ting Tsai, Borching Su, Yu Tsao, Syu‐Siang Wang

Abstract

In this study, a new subspace-constrained diagonal loading (SSC-DL) method is presented for robust beamforming against the issue of a mismatched direction of arrival (DoA), based on an extension to the well known diagonal loading (DL) technique. One important difference of the proposed SSC-DL from conventional DL is that it imposes an additional constraint to restrict the optimal weight vector within a subspace whose basis vectors are determined by a number of angles neighboring to the estimated DoA. Unlike many existing methods which resort to a beamwidth expansion, the weight vector produced by SSC-DL has a relatively small beamwidth around the DoA of the target signal. Yet, the SSC-DL beamformer has a great interference suppression level, thereby achieving an improved overall SINR performance. Simulation results suggest the proposed method has a near-to-optimal SINR performance.

Introduction

In array signal processing, beamforming is a critical technique used to focus on a desired signal while suppressing interference from other directions. The Minimum Variance Distortionless Response (MVDR) beamformer, also known as the Capon beamformer, is a standard approach but is highly sensitive to model mismatches. Specifically, a small error in the estimated Direction-of-Arrival (DoA) can lead to "signal cancellation," where the beamformer mistakenly treats the desired signal as interference and aggressively suppresses it.

Existing solutions like Diagonal Loading (DL) improve robustness by adding a regularization term to the signal covariance matrix. However, choosing an appropriate loading factor is challenging, and these methods often struggle to effectively suppress strong interference while maintaining signal integrity. Other approaches widen the main beam to accommodate DoA errors, but this typically increases sensitivity to noise and nearby interference.

Proposed Method: SSC-DL

The authors introduce Subspace-Constrained Diagonal Loading (SSC-DL) to address these limitations. The core innovation is restricting the beamformer's weight vector to a specific signal-centric subspace.

The SSC-DL framework: constructing a signal-centric subspace from sampled angles to derive the target weight vector.

The SSC-DL algorithm follows a three-step process:

  1. Subspace Construction: A range of possible angles is defined around the estimated DoA. By sampling $M$ angles within this range, the algorithm calculates optimal weight vectors for each, which then serve as the basis for a specialized subspace ($\mathcal{W}$).
  2. Constrained Optimization: The optimization problem seeks to minimize output variance while strictly constraining the weight vector to lie within the constructed subspace.
  3. Optimal Loading Factor: A specific diagonal loading factor $\gamma = -(\sigma_n^2 + P \cdot N)$ is derived to stabilize the solution, where $P$ is signal power and $N$ is the number of sensors.

By design, this subspace is nearly orthogonal to the directions of potential interference. This allows the beamformer to naturally reject interference without needing to broaden the main beam.

Performance Evaluation

The performance of SSC-DL was evaluated using synthetic simulations with a Uniform Linear Array (ULA). Scenarios included various DoA mismatches and high-power interference sources (up to 40dB above noise).

Output SINR versus SNR with DOA mismatch, showing SSC-DL closely tracking the optimal performance curve compared to other robust methods.

Key results from the simulations include:

  • Near-Optimal SINR: SSC-DL consistently outperformed standard MVDR-DL and other robust methods, tracking the "Optimal" performance curve (which assumes perfect DoA knowledge) very closely across a wide range of SNR levels.
  • Interference Rejection: In high-interference environments, SSC-DL maintained superior SINR. While standard methods often failed to distinguish mismatched signals from strong interference, the subspace constraint in SSC-DL ensured the signal was preserved while interference was suppressed.
  • Sample Efficiency: The method demonstrated robustness even with a small number of data snapshots (as few as 20-30), whereas traditional methods often require significantly more data to converge.

Conclusion and Future Work

SSC-DL provides a robust framework for beamforming by leveraging signal-centric subspaces. It avoids the pitfalls of beamwidth expansion while providing superior interference suppression.

Future research may focus on developing data-driven methods to automatically determine the optimal subspace dimension $M$ and extending the subspace constraint to handle other steering vector errors, such as sensor gain or phase calibration issues.

Summary updated

Abstract

In this study, a new subspace-constrained diagonal loading (SSC-DL) method is presented for robust beamforming against the issue of a mismatched direction of arrival (DoA), based on an extension to the well known diagonal loading (DL) technique. One important difference of the proposed SSC-DL from conventional DL is that it imposes an additional constraint to restrict the optimal weight vector within a subspace whose basis vectors are determined by a number of angles neighboring to the estimated DoA. Unlike many existing methods which resort to a beamwidth expansion, the weight vector produced by SSC-DL has a relatively small beamwidth around the DoA of the target signal. Yet, the SSC-DL beamformer has a great interference suppression level, thereby achieving an improved overall SINR performance. Simulation results suggest the proposed method has a near-to-optimal SINR performance.
BibTeX
@misc{tsai2016robust,
  author = {Yueh‐Ting Tsai and Borching Su and Yu Tsao and Syu‐Siang Wang},
  title = {{Robust Beamforming Against Direction-of-Arrival Mismatch Using Subspace-Constrained Diagonal Loading}},
  year = {2016},
  eprint = {1602.02690},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1602.02690},
}
Figure