Off-Grid direction of arrival estimation in the presence of measurement noise and heavy cluttered environment
- PDF / 685,896 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 54 Downloads / 196 Views
ORIGINAL PAPER
Off-Grid direction of arrival estimation in the presence of measurement noise and heavy cluttered environment Sadeq Ebrahimi1 · Ghazaleh Sarbishaei1
· Ghosheh Abed Hodtani2
Received: 15 April 2020 / Revised: 30 August 2020 / Accepted: 16 September 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract In this paper, we focus on estimating Direction of Arrival (DOA) and removing heavy clutter embedded with measurement noise. A correlated Gaussian process is chosen to model destructive effects of clutter. Also, a white Gaussian process is selected to describe measurement noise caused by sensor array. After adding these distortions to the off-grid model, we utilize Sparse Bayesian Learning and principal component analysis (as a preprocessing stage) in order to remove these distortions as well as estimating of true DOAs. Finally, at the end we will show how ignorance of clutter from model or combine it with measurement noise degrade DOA estimation. This will be demonstrated by various numerical simulations. Keywords Direction of arrival estimation · Off-grid DOA estimation · Clutter · Sparse Bayesian learning
1 Introduction Multiple Signal Classification (MUSIC) is one of the most used subspace-based techniques for DOA estimation [1]. It is robust against noise [2] and makes almost maximum use of the structure of the sampled data [3]. Authors in [4] have studied the performance of MUSIC in computational time-reversal (TR) applications that the analysis builds upon classical results on first-order perturbation of singular value decomposition. MUSIC has been successful in wide range of applications in communication systems, from source localization [5] to radar imaging [6], [7], [8]. Despite its efficiency, subspace-based DOA estimation methods suffer from several major drawbacks such as necessity of knowledge about actual source numbers, sensitive to sensor calibration error and hard to detect correlated sources [9]. By utilizing spatial sparsity of targets, DOA estimation can be considered as a sparse reconstruction problem [10]. Detection of coherent sources, better resolution and reduction in snap-
B
Ghazaleh Sarbishaei [email protected] Sadeq Ebrahimi [email protected] Ghosheh Abed Hodtani [email protected]
1
Sadjad University of Technology, Mashhad, Iran
2
Ferdowsi University of Mashhad, Mashhad, Iran
shots is some of the benefits of the sparse DOA estimation [11]. Generally, sparse DOA estimation can be classified into three categories, namely, on-grid, off-grid and gridless methods [12]. In on-grid models, it is assumed that real DOAs lie on fix set of grid points [10]. However, grid mismatch is a major problem with these kind of approaches; that is to say, a target is not on the sampling grid. To prevent this problem, grid points should be as much as close together. But, a denser sampling grid increases the computational complexity of the algorithms. In addition, a denser sampling grid results in a higher mutual coherence between columns of the measurement m
Data Loading...
