Low Complexity Single Source DOA Estimation Based on Reduced Dimension SVR

TL;DR

This paper introduces a low-complexity, reduced-dimension support vector regression method for single-source DOA estimation, combining azimuthal SVR with a closed-form elevation approach, achieving similar accuracy to MUSIC with less computational effort.

Contribution

It proposes a novel sequential DOA estimation technique that reduces computational complexity by focusing training on azimuthal angles and integrating a closed-form elevation estimation.

Findings

Significant complexity reduction compared to MUSIC.

Achieves similar root-mean-square error performance.

Effective for real-time DOA estimation.

Abstract

Conventional direction of arrival (DOA) estimation algorithms suffer from performance degradation due to antenna pattern distortion and substantial computational complexity in real-time execution. The support vector regression (SVR) approach is a possible solution to overcome those limitations. In this work, we propose a sequential DOA estimation technique that combines the reduced dimension SVR (for the azimuthal plane) with a closed form approach (for the elevation plane). Thus, the training and testing are only required for the azimuthal angles which makes it very attractive from the implementation complexity point of view. Our analysis demonstrates that the proposed algorithm offers significant complexity gain over the popular MUSIC algorithm while exhibiting similar root-mean-square error performance.