scispace - formally typeset
Search or ask a question
Author

Wenyi Liu

Bio: Wenyi Liu is an academic researcher from Xi'an Jiaotong University. The author has contributed to research in topics: Covariance matrix & Oblique projection. The author has an hindex of 2, co-authored 5 publications receiving 31 citations.

Papers
More filters
Journal ArticleDOI
TL;DR: An oblique projection based alternating iterative scheme is presented to improve the accuracy of the estimated location parameters and shows that the LPATS provides good estimation performance for both the DOAs and ranges compared to some existing methods.
Abstract: This paper investigates the localization of multiple near-field narrowband sources with a symmetric uniform linear array, and a new linear prediction approach based on the truncated singular value decomposition (LPATS) is proposed by taking an advantage of the anti-diagonal elements of the noiseless array covariance matrix. However, when the number of array snapshots is not sufficiently large enough, the “saturation behavior” is usually encountered in most of the existing localization methods for the near-field sources, where the estimation errors of the estimated directions-of-arrival (DOAs) and ranges cannot decrease with the signal-to-noise ratio. In this paper, an oblique projection based alternating iterative scheme is presented to improve the accuracy of the estimated location parameters. Furthermore, the statistical analysis of the proposed LPATS is studied, and the asymptotic mean-square-error expressions of the estimation errors are derived for the DOAs and ranges. The effectiveness and the theoretical analysis of the proposed LPATS are verified through numerical examples, and the simulation results show that the LPATS provides good estimation performance for both the DOAs and ranges compared to some existing methods.

38 citations

Proceedings ArticleDOI
01 Sep 2019
TL;DR: A framework of DNN where a regression layer is utilized to address the problem of near-field source localization and a novel form of feature representation to take full advantage of the convolution networks is proposed.
Abstract: Source localization for near-field narrowband signal is an important topic in array signal processing. Deep neural network (DNN) based methods are data-driven and free of pre-assumptions about data model and are expected to learn the intricate nonlinear structure in large data sets. This paper proposes a framework of DNN where a regression layer is utilized to address the problem of near-field source localization. Unlike previous studies in which DOA estimation is modeled as a classification problem and have a relatively low resolution, we exploit a regression model and aim to improve the estimation accuracy. In the training stage, we propose a novel form of feature representation to take full advantage of the convolution networks. In addition, the architecture of deep neural networks is well designed taking in to consideration the trade-off between the expression ability and under-training risks. The simulation results show that the proposed approach has a rather high validation accuracy with a high resolution, and also outperforms some conventional methods in adverse environments such as low signal to noise ratio (SNR) or small number of snapshots.

12 citations

Proceedings ArticleDOI
01 Sep 2018
TL;DR: This paper investigates the problem of localizing multiple narrowband near-field signals impinging on a symmetrical uniform linear array (ULA) by exploiting the anti-diagonal elements of the array covariance matrix and presents an alternating iterative scheme to improve the estimation accuracy of the location parameters.
Abstract: Recently many subspace-based localization methods were developed for estimating the directions of arrivals (DOAs) and ranges of multiple narrowband signals in near-field. However, most of them usually encounter “saturation behavior” in estimation performance regardless of the signal-to-noise ratio (SNR) when the number of array snapshots is not sufficiently large enough. In this paper, we investigate the problem of localizing multiple narrowband near-field signals impinging on a symmetrical uniform linear array (ULA). Firstly, by exploiting the anti-diagonal elements of the array covariance matrix, a new linear prediction approach with truncated singular value decomposition (SVD) is proposed to estimate the location parameters (i.e., DOA and range) of the incident signals. Secondly, as a measure against the impact of finite array data, an alternating iterative scheme is presented to improve the estimation accuracy of the location parameters, where the “saturation behavior” encountered in most of localization methods is solved effectively. Furthermore, the statistical analysis of the proposed method is studied, and the asymptotic mean-squared-error (MSE) expressions of the estimation errors are derived for two location parameters. Finally, the effectiveness and the theoretical analysis are substantiated through numerical examples.

2 citations

Patent
24 Aug 2018
TL;DR: In this article, the authors proposed a near-field signal source positioning method based on interceptive singular value decomposition (ISMVD) for a single-antenna near field signal.
Abstract: The invention discloses a near-field signal source positioning method based on interceptive singular value decomposition. The method comprises the steps of 1, estimating the initial value of the electric angle of a near-field signal, and calculating the covariance matrix R of the initial value; 2, calculating the oblique projection operator of the initial value of the electric angle; 3, accordingto the oblique projection operator, obtaining an updated covariance matrix; 4, according to the initial value of the electric angle and the updated covariance matrix, updating the value of the electric angle; 5, if the difference value between the updated value of the electric angle and the initial value of the electric angle is smaller than a set threshold value, obtaining the final value of theelectric angle and completing the location of a near-field signal source; otherwise, adopting the updated value of the electric angle as an initial value, and repeating the steps 2 to 5 for iteration.The floating error caused by the characteristic value of the noise subspace is eliminated. Meanwhile, the prediction precision is improved. The mutual influence between a plurality of near-field signals is effectively eliminated. A real value can be approximated within a certain allowable error range after several times of iteration.

1 citations

Patent
03 Dec 2019
TL;DR: In this article, a near-field signal source positioning method based on a deep neural network regression model is proposed, where the estimation precision of the direction of arrival angle is improved by ten times under the conditions that the signal-to-noise ratio is lower than 15dB and the snapshot number is smaller than 200.
Abstract: The invention discloses a near-field signal source positioning method based on a deep neural network regression model. The method comprises the following steps: carrying out calculation based on a covariance matrix R to obtain a feature extraction matrix r; constructing a deep neural network regression model; generating a training set of the deep neural network regression model; determining various parameters required by training the deep neural network regression model; training the constructed deep neural network regression model by using the determined parameters and the training set to obtain a trained deep neural network regression model; and inputting the feature extraction matrix r into the trained deep neural network regression mode, and outputting the direction of arrival and thedistance of the near-field signal through the deep neural network regression model to complete near-field signal source positioning. According to the method disclosed by the invention, with introduction of the deep neural network regression model, the estimation precision of the direction of arrival angle is improved by ten times under the conditions that the signal-to-noise ratio is lower than 15dB and the snapshot number is smaller than 200; and the estimation precision of the distance is close to a theoretically optimal solution.

Cited by
More filters
Book ChapterDOI
11 Dec 2012

1,704 citations

01 Jan 2016
TL;DR: This statistical signal processing detection estimation and time series analysis will help people to read a good book with a cup of tea in the afternoon, instead they juggled with some malicious bugs inside their laptop.
Abstract: Thank you for reading statistical signal processing detection estimation and time series analysis. Maybe you have knowledge that, people have look hundreds times for their chosen novels like this statistical signal processing detection estimation and time series analysis, but end up in harmful downloads. Rather than reading a good book with a cup of tea in the afternoon, instead they juggled with some malicious bugs inside their laptop.

146 citations

Journal ArticleDOI
TL;DR: A source localization scheme under the simultaneous existence of near-field and far-field sources using a symmetric double-nested array (SDNA) that significantly outperforms the existing methods for both DOA and range estimations.
Abstract: Nested arrays have recently attracted a growing concern owing to their ability of achieving extended aperture and enhanced degrees of freedom (DOFs) compared to conventional uniform linear arrays (ULAs). In this paper, we devise a source localization scheme under the simultaneous existence of near-field (NF) and far-field (FF) sources using a symmetric double-nested array (SDNA). First, the direction-of-arrivals (DOAs) of FF sources are estimated using 1-D MUSIC spectrum. Then, the NF components can be extracted by applying the oblique projection technique. By constructing a special NF cumulant matrix and performing vectorization to generate the virtual array signal, the DOAs of NF sources can be estimated by the spatial smoothing MUSIC (SS-MUSIC) algorithm. Finally, with the NF DOA estimates, the range estimates of NF sources are obtained via 1-D peak searching. We also derive the consecutive range of the difference coarray and optimum array configurations for the SDNA, under a given number of sensors. The proposed algorithm exploits the large coarray aperture to enhance the localization performance and can distinguish correctly the types of the sources. The numerical results demonstrate that our proposed method significantly outperforms the existing methods for both DOA and range estimations.

64 citations

Journal ArticleDOI
TL;DR: In this paper, a new symmetric NLA, termed symmetric displaced coprime array (SDCA), was proposed to locate the near-field and far-field sources simultaneously.
Abstract: Conventionally, mixed near-field (NF) and far-field (FF) source localization is performed using a symmetric uniform linear array (ULA). Recently, two symmetric nonuniform linear arrays (NLAs), symmetric nested array (SNA) and compressed SNA (CSNA), have been developed to enhance the positioning performance by their advantages in the degrees of freedom and the array aperture over symmetric ULAs. In this article, we devise a new symmetric NLA, termed symmetric displaced coprime array (SDCA), to locate the NF and FF sources simultaneously. The SDCA configuration consists of three sparse ULAs with a certain displacement, implying there are two displaced coprime arrays with one common subarray. For a given number of sensors, the SDCA configuration is solely determined by a closed-form expression and its consecutive coarray ranges can also be analytically computed. In addition, we derive two optimum SDCA configurations by maximizing the number of the unique and consecutive lags in the difference coarray. Compared with the SNA and the CSNA, the SDCA provides more unique and consecutive virtual sensors, as well as a larger physical array aperture. Numerical experiments are presented to verify the superiorities of the SDCA configuration over the existing symmetric NLAs.

42 citations

Journal ArticleDOI
TL;DR: In this article, a DoA estimation framework based on complex-valued deep learning (CVDL) is proposed for the near-field region in short-range MIMO communication systems.
Abstract: The near-field effect of short-range multiple-input multiple-output (MIMO) systems imposes many challenges on direction-of-arrival (DoA) estimation. Most conventional scenarios assume that the far-field planar wavefronts hold. In this article, we investigate the DoA estimation problem in short-range MIMO communications, where the effect of near-field spherical wave is non-negligible. By converting it into a regression task, a novel DoA estimation framework based on complex-valued deep learning (CVDL) is proposed for the near-field region in short-range MIMO communication systems. Under the assumption of a spherical wave model, the array steering vector is determined by both the distance and the direction. However, solving this regression task containing a massive number of variables is challenging, since datasets need to capture numerous complicated feature representations. To overcome this, a virtual covariance matrix (VCM) based on received signals is constructed, and thus such features extracted from the VCM can deal with the complicated coupling relationship between the direction and the distance. Although the emergence of wireless big data driven by future communication networks promotes deep learning-based wireless signal processing, the learning algorithms of complex-valued signals are still ongoing. This article proposes a one-dimensional (1-D) residual network that can directly tackle complex-valued features due to the inherent 1-D structure of signal subspace vectors. In addition, we put forth a cropped VCM based policy which can be applied to different antenna sizes. The proposed method is able to fully exploit the complex-valued information. Our simulation results demonstrate the superiority of the proposed CVDL approach over the baseline schemes in terms of the accuracy of DoA estimation.

23 citations