China
Africa
Explore chapters and articles related to this topic
An efficient hybrid direction of arrival estimation scheme for massive multiple-input multiple-output systems
Published in Artde D.K.T. Lam, Stephen D. Prior, Siu-Tsen Shen, Sheng-Joue Young, Liang-Wen Ji, Engineering Innovation and Design, 2019
Ann-Chen Chang, Wei Jhang, Shiaw-Wu Chen
The DOA estimation has been of interest to the signal processing community for decades. Enormous interest has been raised within the array processing community in the high-resolution subspace-based angle estimation algorithms, which include Multiple Signal Classification (MUSIC) (Yang et al., 2014), Estimation of Signal Parameters via Rotational Invariance Technique (ESPRIT) (Hu et al., 2014), and their variants (Yao et al. 2017). However, these conventional MUSIC and ESPRIT algorithms are not suitable for the massive MIMO systems. They both need Eigenvalue Decomposition (EVD) of the autocorrelation matrix and/or spectrum peak search, which have high computational complexity. The subspace-based methods require the formulating of a reliable signal-subspace or noise-subspace method. However, a huge amount of computation would be involved, where a large array element size is required in applications. A possible alternative to the subspace-based method for DOA estimation with computational saving is the Orthogonal Projection (OP) method (Chang & Shen, 2013). Meanwhile, the computational requirements are very high for the searching-based estimators, where the complexity and estimation accuracy strictly depend on the grid size used during the search. It is time-consuming and the search grid is ambiguous.
A Review of Parametric High-Resolution Methods
Published in Yingbo Hua, Alex B. Gershman, Qi Cheng, High-Resolution and Robust Signal Processing, 2017
In Section 1.3, we present methods that exploit large sample theorems in statistics. In particular, we focus on data of multiple independent measurements. The key data structure is captured by the dominant (principal) subspace of the data matrix or the dominant eigenvectors of the data covariance matrix. The principal subspace is referred to as signal subspace. The orthogonal complement of the signal subspace is referred to as noise subspace. Exploitation of the orthogonality between the signal subspace and the noise subspace results in a method known as MUSIC. For some data models such as the superimposed exponentials, the signal subspace possesses an invariance structure that can be exploited in a matrix pencil fashion also known as ESPRIT. A number of variations of MUSIC and ESPRIT are described there. We also present the maximum likelihood methods and their asymptotic (large sample) performances. The bilinear nature of the sensor array model is exploited in the conditional maximum likelihood method, unconditional maximum likelihood method and the so-called MODE method. The situation of coherent signals is also discussed.
Hybrid Methods for Localization
Published in Prabhakar S. Naidu, Distributed Sensor Arrays Localization, 2017
The cross-spectral matrix is next subjected to eigenvalue decomposition. The signal subspace is identified by large eigenvalues and the noise subspace by small eigenvalues. To compute two correlations defined in Equations 5.23 and 5.24, Corr1 and Corr2, we compute the inner product of signal/noise subspace with computed basis vectors for assumed transmitter parameters. The results are shown in Table 5.1. Each table shows scan results of one parameter keeping others fixed and known. The maximum (M = 32) of Corr1 and the minimum (zero) of Corr2 are at the correct value of the unknown parameter. It was noticed that the results showed some dependence on the actual distribution of sensors. Data length played a significant role in being able to measure the Doppler effect due to the transmitter motion. It is noticed that a data length of 20 or more times N is required for localization but a length over 70 times N would be required for velocity estimation (see Figure 5.3).
Accurately estimated the complex relative permittivity of materials using a super high-resolution algorithm at X-band microwave propagation
Published in Electromagnetics, 2020
Manh Cuong Ho, Trong-Hieu Le, Le Cuong Nguyen
The multiple signal classification algorithm was proposed by R. Schmidt (Schmidt 1986). The basic approach of this algorithm is that from the received signal, the covariance matrix is calculated and then eigenvectors decomposition is carried out. The signal subspace and noise subspace are determined based on eigenvectors and eigenvalues. The results showed that the M – D dimensional subspace spanned by the M – D noise eigenvectors as the noise subspace and the D dimensional subspace spanned by the incident signal parameter vectors as the signal subspace; they are disjoint. The signal and the noise subspaces are calculated by the matrix algebra and they are found to be orthogonal to each other. Therefore, the signal and noise subspaces are isolated by the orthogonal property of this algorithm. Thus, the complex relative permittivity of the material sample is estimated by combining the autocorrelation and MUSIC function of the received signal.
A Novel Approach to Improve the Speech Intelligibility Using Fractional Delta-amplitude Modulation Spectrogram
Published in Cybernetics and Systems, 2018
Arul Valiyavalappil Haridas, Ramalatha Marimuthu, Basabi Chakraborty
There are few challenges that is existed in all the speech enhancement procedures. The main challenge is regarding the quality of the speech-enhanced signal obtained using the existing approaches. In the Wiener filter, spectral approximation used eliminates the spectral details of the speech harmonics that introduces the residual noise in the enhanced speech signal. This is the major challenge and to solve this problem, the order of the AR-model is increased that leads to the computational complexity (He, Bao, and Bao 2017). Signal subspace approach (SSA) suffers from the problems like the high computational complexity, and lack of robustness when applied to some forms of real noise (Lev-Ari and Ephraim 2003). Another major challenge of the speech enhancement method is about dealing with the different types of the non-stationary noises. Unlike the stationary noises, predicting and estimating the spectral properties of the stationary noises is a risky task (Sun et al. 2016). The EMD-based filtering (EMDF) method of removing the noise from the speech signal contributes poor performance in the presence of the babble noise (Upadhyay and Pachori 2017). The autoregressive hidden Markov models (ARHMM) used to suppress the noise in the non-stationary environments failed to model the random variations of speech and noise (Deng, Bao, and Kleijn 2015). The segment-based methods like the least mean square (LMS) based methods and the neural network-based methods offer more robustness, but they seek the knowledge of the noise to perform the speech enhancement (Ming and Crookes 2017).
Hyperspectral anomaly detection: a performance comparison of existing techniques
Published in International Journal of Digital Earth, 2022
Noman Raza Shah, Abdur Rahman M. Maud, Farrukh Aziz Bhatti, Muhammad Khizer Ali, Khurram Khurshid, Moazam Maqsood, Muhammad Amin
The autocorrelation matrix of the background pixels can be expressed as where represents the autocorrelation matrix of . In practice, this correlation matrix would be estimated from the hyperspectral data. Signal subspace techniques typically rely on the eigenspace of the correlation matrix to identify the background subspace. The eigenvectors of , , corresponding to the p-largest eigenvalues provide a basis for the background subspace while the remaining eigenvectors provide a basis for the orthogonal subspace. Using these eigenvectors, matrices and can be defined as In Complementary Subspace Detector (CSD) (Schaum 2007), the test statistic is defined as where and are the projections of on column space of and , respectively. Similarly in the Subspace RX detector (Schaum 2004), the test pixel is first projected onto the background subspace to obtain . The RX detector is then applied on .