Sampling In Frequency Domain Research Articles

Boosting transferability of adversarial samples via saliency distribution and frequency domain enhancement

Deep neural networks (DNNs) perform excellently in various vision tasks, however, they are susceptible to the adversarial samples. These samples are crafted by injecting subtle perturbations into benign images that are barely detectable to humans. Due to the differences between the substitute model and target model, the efficacy of black-box attack still yield suboptimal results. To address this issue, we propose a black-box attack approach, SD-FI2M, based on the momentum iterative fast gradient sign method, which can craft potent adversarial samples by inducing diverse inputs in the frequency domain and the saliency distribution of images. To enhance the diversity of input samples in frequency domain, a spectrum transformation technique utilizing the discrete cosine transform was used to craft adversarial samples with higher transferability. Further, Guided Grad-CAM was employed to obtain the saliency distribution of benign images, reducing indiscriminate damage to image features and alleviating the overfitting problem in adversarial samples. Extensive experiments on ImageNet have verified the superior attack performance and transferability of the proposed method.

Noise Reduction Method for Fiber Optical Current Transformer Based on Mode Decomposition-Sample Entropy

When a fiber optic current transformer (FOCT) malfunctions, its output signal is usually accompanied by a large noise, which affects the subsequent fault diagnosis results of the transformer. In response to the above noise issues, this paper constructs a new noise reduction algorithm by combining Complete Ensemble Empirical Mode Decomposition Adaptive Noise (CEEMDAN) and Sample Entropy (SE). Above all, the output signal of FOCT is decomposed into multiple fault components using the CEEMDAN algorithm. Then, by comprehensively considering the frequency domain characteristics and sample entropy calculation value of each component, the characteristics of each component are determined, so the power frequency and low-frequency components are selected as the feature vectors representing the fault mode. Finally, the noise components with high sample entropy are eliminated, and the power frequency current components with medium sample entropy and the low-frequency fault feature components with low sample entropy are recombined and superposed to build a fault signal dataset after noise reduction.

Intelligent Fault Diagnosis With Noisy Labels via Semisupervised Learning on Industrial Time Series

Deep neural networks (DNNs) excel at industrial fault diagnosis. Their performance heavily relies on the quality of human-annotated labels. Due to perception limitations of annotators, industrial time series samples (such as vibration and voltage signals) are frequently mislabeled in several conditions, such as samples with frequency domain feature differences and samples on class borders. Hence, an annotated industrial dataset will inevitably contain noisy labels at a certain level, leading to over-fitting and poor generalization of DNNs. In this work, we introduce an Industrial Noisy Label Semi-Supervised Learning (INL-SSL) fault diagnosis approach, addressing the problem that a certain number of samples in an industrial dataset are mislabeled. The proposed INL-SSL architecture simultaneously trains two DNNs, which cross-train on each other to filter noisy label errors. In particular, a fitted Gaussian mixture model divides time series samples of each DNN flow into an unlabeled set with samples likely to be noisy and a labeled set with samples likely to be clean. Given the labeled and unlabeled data, we proposed a time series MixMatch semi-supervised learning strategy to train the diagnostic model. Ablation study verifies the benefit of the proposed time series augmentation techniques for semi-supervised training. Extensive experiments on a benchmark industrial dataset of rolling element bearings (REB) reveal that the INL-SSL outperforms state-of-the-art approaches. On another self-collected REB dataset, the proposed approach also exceeds other comparison methods under noise ratios from 20% to 90%, validating the model's generalizability.

Frequency-domain adaptive MIMO filter with fractional oversampling using stochastic gradient descent for long-haul transmission over coupled 4-core fibers.

We propose a fractionally spaced frequency-domain adaptive multi-input multi-output (MIMO) filter architecture in which the sampling rate of input signals is below 2× oversampling with a non-integer oversampling factor for mode demultiplexing in long-haul transmission over coupled multi-core fibers. The frequency-domain sampling rate conversion to the symbol rate, i.e., 1× sampling, is placed after the fractionally spaced frequency-domain MIMO filter. The filter coefficients are adaptively controlled by stochastic gradient descent and gradient calculation with back propagation through the sampling rate conversion from the output signals on the basis of deep unfolding. We evaluated the proposed filter through a long-haul transmission experiment of 16-channel wavelength-division multiplexed and 4-core space-division multiplexed 32-Gbaud polarization-division-multiplexed quadrature phase shift keying signals over coupled 4-core fibers. The fractional oversampling frequency-domain adaptive 8×8 filter with 9/8× oversampling provided little performance penalty after 6240-km transmission compared to the conventional 2× oversampling frequency-domain adaptive 8×8 filter. The computational complexity in terms of the required number of complex-valued multiplications was reduced by 40.7%.

A numerical investigation on direct and data-driven flutter prediction methods

Three alternative algorithms for dynamic aeroelastic stability analysis are investigated. They are based either on direct search of oscillatory conditions of the coupled system or on eigenvalue analysis from frequency-domain sampling of the unsteady aerodynamics. The aerodynamic forcing is obtained with a harmonic-balanced solver developed for a general-purpose finite-volume fluid dynamics solver. This new implementation is first verified against experimental flutter results from the literature. A wing with relatively low bending stiffness is then used to explore the relative performance of each approach, both in terms of the numerical robustness and of their usability to support aeroelastic design.

Device-Free WLAN Based Indoor Localization Scheme With Spatially Concatenated CSI and Distributed Antennas

Various machine learning (ML) based localization schemes using channel state information (CSI) in wireless local area networks (WLANs) have been investigated recently. Adopting a proper feature selection technique is important to achieve further improvement in detection accuracy. As described herein, we propose a device-free indoor localization scheme using a lightweight ML model with compressed spatially concatenated CSI in WLAN systems with distributed antennas. In this scheme, feedback beam-forming weights (BFWs) are collected at a CSI capture terminal. Then, current and past BFWs are concatenated as accurate feature data to characterize the object behavior. Additionally, we propose the use of a frequency-domain sampling scheme for a low-complexity real-time target position detection with a small number of datasets. Using ray-trace based simulation analysis and experimentally obtained results from an indoor environment, we demonstrate that the proposed scheme using the concatenated CSI is effective not only for achieving more accurate real-time detection, but also for reducing the necessary complexity for both off-line training and on-line classification compared with other reference schemes.

Fully connected aperture array design of the segmented planar imaging system

Compared with the traditional imaging systems, segmented planar imaging technology has the advantages of low mess, small size, and low power in the same resolution situation. To obtain relatively complete frequency domain coverage, the lenslet array requires a large number of lenslets, and the photonic integrated circuit board requires a large number of optical devices, which limits the application and development of the segmented planar imaging technology. In this paper, we introduce a novel, to the best of our knowledge, design of the photonic integrated circuit to ensure that each lenslet in the lenslet array can form a baseline with any other lenslets. This breaks the barrier between segmented planar imaging technology and the traditional synthetic aperture, giving segmented planar imaging technology a sufficient number of frequency domain samples and a concise photonic integrated circuit structure.

Phase Demodulation of Rotor Torsional Vibration Measurement under Time-Varying Speed

When torsional vibration is measured with the zebra-tape method, the modulation signals can be contaminated by electrical noise, environmental noise, step noise, sensor nonlinearities, and so on. Although the amplitudes of these types of noise are very weak relative to the amplitude of the ideal pulse test signal, they may introduce large-scale random broad-spectrum noise to the signal phases. Such noise is difficult to remove using frequency domain, time-frequency-domain, or threshold sampling methods and seriously affects the accuracy of torsional vibration measurements. This paper presents a phase demodulation algorithm based on downsampling and local resampling (DSLR) to improve the accuracy of torsional vibration measurements. To verify the proposed DSLR algorithm, torsional vibration simulations and experiments are conducted under a time-varying rotation speed. The results show that the angular displacement signal obtained from the torsional vibration test using zebra tapes is the superposition of the dynamic and static angular displacements, with the latter generated because of the varying rotation speed with respect to time. The DSLR algorithm effectively reduces the phase noise and demodulates the pulse signal phases. The work in this paper provides a method for obtaining reference torsional vibration measurements of rotors based on the zebra-tape method under time-varying speeds.

An Algorithm for Solving the Phase Ambiguity Problem of SAW Delay-Line Sensor Systems.

Solving the phase ambiguity problem is crucial to achieving a wide-range and high-precision measurement for the frequency-domain sampling (FDS)-based surface acoustic wave (SAW) delay-line sensor systems. This study proposes an improved phase estimation algorithm called dual-band phase estimation (DBPE) to solve the problem. By using DBPE, the SAW sensor system can obtain an extensive and alterable measuring range without further requirements for sensor design or transmitted signals. Thus, it can be widely used in various FDS-based SAW delay-line sensor systems. Monte Carlo simulations and temperature measuring experiments, based on a YZ-cut LiNbO3 SAW delay-line sensor and a switched frequency-stepped continuous wave (S-FSCW) reader, are performed to demonstrate the algorithm's validity. The Monte Carlo simulations show that DBPE can effectively solve the phase ambiguity problem and has better performance than frequency estimation in measuring precision at a low signal-to-noise ratio (SNR). The temperature-sensing experiments show that DBPE has a good performance in measuring range and precision, serving as a phase ambiguity solver in the temperature sensor system.

Single photonic integrated circuit imaging system with a 2D lens array arrangement.

The segmented planar imager is an advanced optical interferometric telescope with a photonic integrated circuit (PIC). It provides a significant reduction in size, weight, and power consumption as compared to traditional optical interferometry. In this article, we propose the combination of a single PIC with a two-dimensional (2D) lens array to achieve single-PIC imaging. Unlike previous designs which require a large number of PIC arrangements in different directions for imaging, a single-PIC imaging system requires only one PIC for 2D frequency domain sampling and imaging. In addition, the single-PIC imaging system can form a larger equivalent aperture through modularization. Since PIC can be mass-produced, the modularization ability of the single-PIC imaging system greatly shortens the production and development cycle of large-aperture telescopes.

Sparse sampling in frequency domain and laser imaging

The monochromaticity of the laser and the characteristics of the natural image’s spectrum, including sparsity and concentrating in the low frequency range, make it possible to sample the image spectrum sparsely. Based on small-scale laser detectors and the introduced laser reference signals, a method of laser imaging with sparse sampling in frequency domain is proposed in this paper. The principle of frequency sparse sampling laser imaging and the imaging system structure are introduced. The simulation results of spectrum and complex images reconstructed are given. Both the effects of the signals’ parameters, such as the ratio of the reference laser signal amplitude to the laser echo spectrum amplitude and the initial phase of the laser reference signal, on reconstruction results are investigated. The reconstruction results are evaluated by correlation coefficient, mean square error (MSE), and structural similarity index (SSIM). For the strong correlation between phase and amplitude of the laser echo complex image, the amplitude image and the phase image are both set to be 256 × 256 diagram. The sparse laser detector plane array consists of 5 64 × 64 frequency domain laser detector arrays, which form a cross and make a sparsity rate of 31.25%(5/16). The simulation results show that the correlation coefficient, MSE and SSIM of the spectrum reconstructed are 0.96, 22.14, 1.00 and those of the complex image reconstructed are 0.96, 1857.25 and 0.67 respectively. The simulation results indicate that the method proposed is effective. However, the method requires the laser reference signal amplitude to be about 30 times the mean value of the laser echo spectrum amplitude, which reduces the dynamic range of the detectors. The initial phase of the laser reference signal has no obvious effect on the reconstruction results.

Clustering-based CLEAN algorithm in ghost imaging with sparse spatial frequencies.

When insufficient samples in the spatial frequency domain could be effectively compensated by the modified CLEAN algorithm, a novel aperture-synthetic scheme of ghost imaging takes advantage of a superior speed of modulation and an enhancement on the spatial resolution. However, there still exist some imperfections in the modified CLEAN reconstructions, such as the fact that some omitted scatter noise still remains or the object contour may be incomplete. Therefore, we optimize the modified CLEAN algorithm by proposing a density clustering algorithm to overcome these drawbacks and improve the visual quality.

Investigation of the low-complexity Hilbert FIR filter enhanced 112-Gbit/s SSB 16-QAM transmission with parallelized Kramers-Kronig reception over 1440-km SSMF.

The Hilbert transform links the log-magnitude and the phase of the field modulated signals as long as the minimum phase condition is satisfied in the Kramer-Kronig (KK) receiver. In discrete-time signal processing, the Hilbert transform is generally replaced by a finite impulse response (FIR) filter to reduce the computational complexity, that is the so-called Hilbert transform FIR (HT-FIR) filter. The performance of the HT-FIR filter is extremely important, as the in-band flatness, the ripple, the group delay, the Gibbs phenomenon, and the edge effect, which indeed impair the phase retrieval. Hence, we investigate four different HT-FIR filter schemes that are in the form of type III and type IV based on the frequency-domain (FD) sampling approach and the time-domain (TD) windowing function approach. Also, we analyze the performance for each filter under different digital upsampling scenarios and conclude that a trade-off between the reduced inter-symbol-interference (ISI) and the Gibbs phenomenon is essential to obtain an optimal sampling rate and an improved KK performance when the HT-FIR filter with a short length is adopted. The results show that the FD-based HT-FIR filter can relax the upsampling requirement while having a better in-band flatness and a lower edge effect. The experiment is conducted in the parallelized block-wise KK reception-based 112-Gbit/s SSB 16-QAM optical transmission system over a 1920-km cascaded Raman fiber amplifier (RFA) link to investigate the limit transmission performance of the practical KK receiver. The experimental results show that when the transmission distance is up to 1440-km, the BER of the FD-based HT-FIR filter can be lower than the soft decision-forward error correction (SD-FEC) threshold of 2 × 10-2 with only 3 samples per symbol (3-SPS) upsampling rate and 8 non-integer tap coefficients are used, while other TD-based HT-FIR filter schemes with a BER lower than the SD-FEC threshold require at least 4-SPS upsampling rate.

Generalized Image Reconstruction in Optical Coherence Tomography Using Redundant and Non-Uniformly-Spaced Samples.

In this paper, we use Frame Theory to develop a generalized OCT image reconstruction method using redundant and non-uniformly spaced frequency domain samples that includes using non-redundant and uniformly spaced samples as special cases. We also correct an important theoretical error in the previously reported results related to OCT image reconstruction using the Non-uniform Discrete Fourier Transform (NDFT). Moreover, we describe an efficient method to compute our corrected reconstruction transform, i.e., a scaled NDFT, using the Fast Fourier Transform (FFT). Finally, we demonstrate different advantages of our generalized OCT image reconstruction method by achieving (1) theoretically corrected OCT image reconstruction directly from non-uniformly spaced frequency domain samples; (2) a novel OCT image reconstruction method with a higher signal-to-noise ratio (SNR) using redundant frequency domain samples. Our new image reconstruction method is an improvement of OCT technology, so it could benefit all OCT applications.

Functional Split and Frequency-Domain Processing for Fronthaul Traffic Reduction

This work proposes a compression method developed to transport radio signals over a centralized radio access network using a functional split in which the fronthaul transports frequency-domain samples. The proposed method allows detecting which subcarriers from an LTE/NR OFDM symbol are inactive to avoid sending them over the fronthaul, hence decreasing the required throughput and making it vary with the radio usage. The proposed method was evaluated in different scenarios, and the results were analyzed in terms of user equipment goodput and fronthaul traffic. The simulations show a significant reduction in the fronthaul traffic, especially when the radio usage is low.

Artificial Neural Network-Based System for Location of Structural Faults on Anchor Rods Using Input Impedance Response

Structural faults on anchor rods due to corrosive processes are among the major natural causes worldwide for the collapse of guyed towers. As such towers are widely used on both power transmission lines and antenna systems, their collapse may cause unexpected interruptions of telecommunications and power delivery services. Despite being of utmost importance, underground corrosion monitoring is still a technological challenge, because it is a highly nonlinear problem, influenced by several soil characteristics and its complex iterations with the metallic structure of the rods. Some techniques have already been proposed to locate such faults; however, either they require previous knowledge about the medium or they need the interpretation of the acquired signal by a human specialist. These requirements make it difficult to use such solutions in automatic monitoring as well as to make their integration with Internet of Things (IoT) systems. In this article, a novel technique is proposed to automatically locate structural faults on anchor rods, based on a two-step machine learning (ML) solution. The system is fed by a vector composed of frequency-domain samples of the input impedance electromagnetic (EM) signals measured on the rods through a dedicated designed connector. Experimental results have demonstrated that the proposed combination of EM signal acquisition and ML data processing was able to perform the fault location on anchor rods accurately and with a total independence of any additional human analysis.

FourierCam: a camera for video spectrum acquisition in a single shot

The novel camera architecture facilitates the development of machine vision. Instead of capturing frame sequences in the temporal domain as traditional video cameras, FourierCam directly measures the pixel-wise temporal spectrum of the video in a single shot through optical coding. Compared to the classic video cameras and time-frequency transformation pipeline, this programmable frequency-domain sampling strategy has an attractive combination of characteristics for low detection bandwidth, low computational burden, and low data volume. Based on the various temporal filter kernel designed by FourierCam, we demonstrated a series of exciting machine vision functions, such as video compression, background subtraction, object extraction, and trajectory tracking.

Design of high-order QAM modem algorithm

More and more attention has been paid on the issue of bandwidth utilization. Nowadays, with limited frequency spectrum resources, the capacity of traditional communication system has not been able to meet the demands of users. Quadrature Amplitude Modulation (QAM) has become a critical technical solution for broadband wireless access and wireless video communication due to its advantages of high frequency spectrum utilization and high-power spectrum density. In order to further improve the control information transmission performance of high-order QAM, we propose a high-order QAM modulation and demodulation algorithm, which is designed from the aspects of interleave between channels, leading sequence, constellation modulation, frequency domain sampling and (Orthogonal Frequency Division Multiplexing) OFDM generation, etc. The simulation results demonstrate that the algorithm can ensure the reliable reception of control information.

Adaptive scale model reconstruction for radio synthesis imaging

A sky model from CLEAN deconvolution is a particularly effective high dynamic range reconstruction in radio astronomy, which can effectively model the sky and remove the sidelobes of the point spread function (PSF) caused by incomplete sampling in the spatial frequency domain. Compared to scale-free and multi-scale sky models, adaptive-scale sky modeling, which can model both compact and diffuse features, has been proven to have better sky modeling capabilities in narrowband simulated data, especially for large-scale features in high-sensitivity observations which are exactly one of the challenges of data processing for the Square Kilometre Array (SKA). However, adaptive scale CLEAN algorithms have not been verified by real observation data and allow negative components in the model. In this paper, we propose an adaptive scale model algorithm with non-negative constraint and wideband imaging capacities, and it is applied to simulated SKA data and real observation data from the Karl G. Jansky Very Large Array (JVLA), an SKA precursor. Experiments show that the new algorithm can reconstruct more physical models with rich details. This work is a step forward for future SKA image reconstruction and developing SKA imaging pipelines.

Flexible and Efficient DSP-Assisted Subcarrier Multiplexing for an Analog Mobile Fronthaul

The digital formation of an analog subcarrier multiplex employing in combination both a technique using pre-IFFT frequency-domain samples and one using post-IFFT time-domain samples is proposed and demonstrated. This combined technique enables a compromise for sampling rate requirements, while maintaining low complexity and good performance.