DOA Tracking of Two-Dimensional Coherent Distribution Source Based on Fast Approximated Power Iteration

Tao Wu,Yiwen Li,Chaoqi Fu,Pengtao Zhang,Yangjun Gao,Bo Feng,Wang Zheng

doi:10.1155/2020/3219516

Abstract

Aiming at two-dimensional (2D) coherent distributed (CD) sources, this paper has proposed a direction of arrival (DOA) tracking algorithm based on signal subspace updating under the uniform rectangular array (URA). First, based on the hypothesis of small angular spreads of distributed sources, the rotating invariant relations of the signal subspace of the receive vector of URA are derived. An ESPRIT-like method is constructed for DOA estimation using two adjacent parallel linear arrays of URA. Through the synthesis of estimation by multiple groups of parallel linear arrays within URA arrays, the DOA estimation method for 2D CD sources based on URA is obtained. Then, fast approximated power iteration (FAPI) subspace tracking algorithm is used to update the signal subspace. In this way, DOA tracking of 2D CD sources can be realized by DOA estimation through signal subspace updating. This algorithm has a low computational complexity and good real-time tracking performance. In addition, the algorithm can track multiple CD sources without knowing the angular signal distribution functions, which is robust to model errors.

Highlights

In the field of underwater acoustic and radar detection, since the position of target is constantly changing, it is necessary to obtain the target orientation information in real time. e traditional direction of arrival (DOA) tracking methods assume that the target is a point source
Based on a coprime array, authors of [4] have proposed an estimator where the Toeplitz matrix functions is utilized to resolve off-grid sources. e authors of [5] have realized the estimation of 2D DOA, receiving and transmitting angle simultaneously in the electromagnetic multiple input multiple output (MIMO) system. e authors of [6] have presented direction of departure and DOA estimation for the MIMO radar where a reduced dimension multiple signal classification algorithm requiring one-dimension search is derived. e authors of [7] have proposed an estimator for DOA and mutual coupling self-calibration in the uniform linear arrays-based bistatic MIMO radar where a two-step framework is proposed for DOA estimation, and mutual coupling coefficients are obtained via the least square method
(i) e rotation invariance relations within the uniform rectangular array (URA) with respect to nominal elevation and nominal azimuth are deduced under small angular spreads assumption (ii) e 2D DOA estimation algorithm of coherent distributed (CD) sources is derived based on URA, which does not need to know the angular signal distribution function (ASDF) form of sources and need not spectral peak search and can be applied for DOA estimation for point sources under the background of the nonmoving target (iii) Based on the fast approximated power iteration (FAPI) framework, a DOA tracking method is proposed. e algorithm has good tracking accuracy and can track multiple 2D CD sources in real time

Summary

Introduction

In the field of underwater acoustic and radar detection, since the position of target is constantly changing, it is necessary to obtain the target orientation information in real time. e traditional DOA tracking methods assume that the target is a point source. (i) e rotation invariance relations within the URA with respect to nominal elevation and nominal azimuth are deduced under small angular spreads assumption (ii) e 2D DOA estimation algorithm of CD sources is derived based on URA, which does not need to know the ASDF form of sources and need not spectral peak search and can be applied for DOA estimation for point sources under the background of the nonmoving target (iii) Based on the FAPI framework, a DOA tracking method is proposed. Under the premise of small angular spreads of sources and the distance between sensors, d is set as the half of the wavelength λ, generalized manifold coefficients of the (m, k)th, the (m−1, k)th, and the (m, k−1)th, and the (m−1, k−1)th sensors have the rotating relations as follows (Appendix): ηmk θi, φi􏼁 ≈ ejπ cos θi sin φi η(m−1)k θi, φi􏼁,

DOA Estimation under URA

DOA Tracking Based on FAPI

Results and Discussion

60 Azimuth

Conclusions