Achieving high-accuracy and non-contact heart rate monitoring via radar is a non-trivial task under the interference from respiration its harmonics. In order to eliminate interference, we propose weighted reconstruction method of second harmonic signal heartbeat for estimation. The proposed combines Variational Modal Decomposition (VMD) algorithm with prior information about heartbeat. Namely, after sophisticated endeavor preprocessing, chest wall displacement decomposed into several...

In order to achieve automatic behavioral monitoring of farming animals, this article proposes a convolutional neural network (CNN)-based recognition method for lactating sows. The data streams sows are collected by wearable sensors embedded with 3-D accelerometer and gyroscope used recognize six types behaviors, including movement, drinking, eating, nursing, sleeping, lying. Among these lying, sleeping can be classified as similar static behaviors. Based on the action images constructed...

The power of end-to-end deep learning techniques to automatically learn latent high-level features from raw signals has been demonstrated in numerous application fields, however, few studies systematically investigate how properly encode the time-series firings binary environment sensors that typically work an event-triggering scheme and have irregular sampling rates for in-home human activity recognition. To this end, we here propose two different methods process streaming sensor readings...

High-resolution and low-complexity direction of arrival (DOA) estimation deserves a warm welcome in real radar sensor. However, achieving this goal is nontrivial. For the DOA uniform linear array (ULA), we offer an efficient approach to reach good compromise between complexity, resolution accuracy. Under covariance fitting framework, construct sparse non-negative least squares (NNLS) that solved by alternating method multipliers (ADMM) at low-computational complexity. The regularization item...

Pervasive computing greatly advances the automatic recognition and understanding of human activities effectively bridges gap between low-level sensor signals high-level human-centric applications. The inherent complexity behavior, however, inevitably poses a huge challenge to design robust activity recognizer, especially in classifying similar activities. In this study, we present hierarchical framework, named HierHAR, that infers on-going multi-stage process for better distinguishing...

Abstract In this article, we propose a weighted ℓ 2,1 minimization algorithm for jointly-sparse signal recovery problem. The proposed exploits the relationship between noise subspace and overcomplete basis matrix designing weights, i.e., large weights are appointed to entries, whose indices more likely be outside of row support jointly sparse signals, so that their expelled from in solution, small correspond solution prefers reserve indices. Compared with regular minimization, can not only...

Using multiple measurement vectors (MMV), we propose an algorithm based on weighted &#x2113;<inf>1</inf> minimization for direction- of-arrival (DOA) estimation, in which the weights are obtained by exploiting orthogonality between noise subspace and array manifold matrix. The proposed penalizes nonzero entries whose indices correspond to row support of jointly sparse signals smaller other more likely be outside larger weights, therefore it can encourage sparsity at true source locations....

The recognition of human activities plays a central role in bridging the gap between raw sensor signals and high-level pervasive applications. However, complex nature behavior presents great challenge to choice representative features discriminant classification model further makes it quite difficult develop an accurate robust activity recognizer. Random forest benefits from idea bootstrap random feature sampling, while only preliminary experimentation for simple has been previously reported...

Using wideband signals to improve the resolution performance of forward-looking imaging radar sensors is a very valuable topic. However, frequency point and angle information are coupled in array manifold for signal, making high-resolution estimation range extremely difficult under low complexity. In this paper, we propose novel algorithm radar. Firstly, time delays estimated by signal parameters via rotation invariance techniques (ESPRIT). Then, multiple sets filters constructed using...

In this paper, we focus on how to obtain a proper regularization parameter that should be properly selected for reasonable compromise between finding sparse solution and restricting the recovery error. An enlarged square of Frobenius norm noise can employed select parameter. methodology, exploit inverse cumulative distribution function (CDF) achieve ideal. The simulations demonstrate proposed method selecting has large dynamic range therefore effectively suppress spurious peaks.

Combining the covariance matching criterion with sparse representation, much effort was devoted to improve angular resolution. However, requirement of hyperparameters or high sidelobe makes them unsatisfactory in practical radar applications. In this paper, we propose a Sparse Capon-like Weighted Quadratic Estimator (SCWQE) hyperparameter-free and apply it 77GHz Frequency Modulated Continuous Wave (FMCW) for high-resolution Direction Of Arrival (DOA) estimation. Under assumption uncorrelated...

在信号源为非圆信号的情况下，该文提出了一种实值ESPRIT波达方向(DOA)估计方法。根据非圆信号为实值信号的特点，利用欧拉公式将接收数据转化为正弦和余弦数据，并将其加以拼接，从而虚拟地将阵元个数加倍，然后在此基础上构造一个旋转不变结构来估计波达方向，新算法可以处理的信号个数是传统ESPRIT算法的两倍。由于在实值基础上进行特征值分解，所以该文提出的算法可以有效地将运算量减少到相同维数复值运算量的1/4。仿真实验表明新算法不仅估计精度高而且能够处理的信号个数可大于阵元个数。

A fast sparse covariance-based fitting algorithm with the non-negative least squares (NNLS) form is proposed for direction of arrival (DOA) estimation. The Khatri-Rao product array manifold uniform linear arrays utilized to achieve dimension reducing transformation after vectorizing covariance matrix. Furthermore, DOA estimation problem derived as a NNLS by using property spatial spectrum, which can be solved some efficient solvers. Numerical experiments show that method obtain high...

Based on the covariance-like fitting criterion we propose a direction of arrival (DOA) estimation algorithm that embeds weighting scheme in objective function without selection any hyperparameters. With an assumption uncorrelated sources, formulate problem DOA as linearly constrained quadratic optimization under power constraints. Numerical results show proposed method not only is robust to this but also has superior performance comparison with some other sparsity-driven methods.

Based on the orthogonality between signal subspace and noise subspace, we propose a sparse recovery method for direction of arrival (DOA) estimation. With assumption uncorrelated sources, covariance matrix fitting is achieved by embedding MUSIC-like weights into quadratic minimization, which capable prompting sparsity solution. Numerical results show that proposed outperforms some other methods.

In this paper, the wideband Capon cepstrum weighted l2, 1 minimization algorithm (WB-CW 1) is presented for direction of arrival (DOA) estimation using an acoustic array. The problem DOA converted into sparsity pattern a jointly sparse signal and then solved by norm minimization. WB-CW uses to calculate support related weights. basic idea assign large weights those entries whose row indices are more likely outside signals, so that their expelled from solution. By doing so, further enhances...

We propose a sparsity-aware noise subspace fitting (SANSF) algorithm for direction-of-arrival (DOA) estimation using an array of sensors. The proposed SANSF is developed from the optimally weighted criterion. Our formulation leads to convex linearly constrained quadratic programming (LCQP) problem that enjoys global convergence without need accurate initialization and can be easily solved by existing LCQP solvers. Combining objective function, ℓ 1 norm, non-negative constraints, enhance...

We propose a weighted ℓ <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2,1</inf> minimization algorithm for retrieving high resolution range profile with the stepped frequency radar. Based on sliding window technique raw pulse train is split into subsequences to for-m jointly sparse signature. By using Hadamard product of two Pisarenko framework cepstrums support-related weighting scheme designed control priority class bases in reconstruction...