DOA estimation in structured phase-noisy environments: technical report

TL;DR

This paper introduces a Bayesian variational mean-field approach for estimating directions of arrival in environments with phase noise, improving accuracy over traditional noise-only models.

Contribution

It presents a novel Bayesian methodology that explicitly models phase noise in DOA estimation, addressing a gap in existing methods.

Findings

Enhanced DOA estimation accuracy with phase noise modeling

Effective Bayesian variational mean-field approximation

Simulation results demonstrate improved performance

Abstract

In this paper we focus on the problem of estimating the directions of arrival (DOA) of a set of incident plane waves. Unlike many previous works, which assume that the received observations are only affected by additive noise, we consider the setup where some phase noise also corrupts the data (as for example observed in atmospheric sound propagation or underwater acoustics). We propose a new methodology to solve this problem in a Bayesian framework by resorting to a variational mean-field approximation. Our simulation results illustrate the benefits of carefully accounting for the phase noise in the DOA estimation process.