Real-time Extraction Research Articles

Ultra Wide Band communication for condition-based monitoring, a bridge between edge and cloud computing

Ultra Wide Band (UWB) wireless communications offer a radically different approach to wireless communication compared to conventional narrow band systems in an industrial context. A lot of industrial applications do not require real-time acquisition and uses low data rates, but some scenarios, such as the condition monitoring systems and the control processes require higher data rates (i.e. tens of kHz). General purpose wireless solutions do not compliant industrial standards. This paper discusses the potential of UWB technology for realizing use cases in industry 4.0 and shows the implementation of an UWB sensor network for machine vibration monitoring that is able to acquire and transmit (6.8Mbit/s data rate) accelerometric measures sampled up to 32KHz with a maximum Packet Loss Rate (PLR) of 2.44%. It also shows as UWB could be a valid instrument to connect cloud computing and edge computing even for real time extraction of diagnostic and prognostic features from high-frequency signals. Different experiments have been performed in an industrial and laboratory environment to evaluate the performance of UWB radio for communication in a wireless sensor network.

Pattern Matching for Real-Time Extraction of Fast and Slow Spectral Components from sEMG Signals.

Previous studies have demonstrated the potential of surface electromyography (sEMG) spectral decomposition in evaluating muscle performance, motor learning, and early diagnosis of muscle conditions. However, decomposition techniques require large data sets and are computationally demanding, making their implementation in real-life scenarios challenging. Based on the hypothesis that spectral components will present low inter-subject variability, the present paper proposes the foundational principles for developing a real-time system for their extraction by utilizing a pre-defined library of components derived from an extensive data set to match new measurements. The model library was tailored to fulfill specific requirements for real-time system application and the challenges encountered during implementation are discussed in the paper. For system validation, four distinct data sets comprising isotonic and isometric muscle activations were utilized. The extracted during validation showed low inter-subject variability, suggesting that a wide range of physiological variations can be described with them. The adoption of the proposed system for muscle analysis could provide a deeper understanding of the underlying mechanisms governing different motor conditions and neuromuscular disorders, as it allows for the measurement of these components in various daily-life scenarios.

A 16-Channel Low-Power Neural Connectivity Extraction and Phase-Locked Deep Brain Stimulation SoC.

Growing evidence suggests that phase-locked deep brain stimulation (DBS) can effectively regulate abnormal brain connectivity in neurological and psychiatric disorders. This letter therefore presents a low-power SoC with both neural connectivity extraction and phase-locked DBS capabilities. A 16-channel low-noise analog front-end (AFE) records local field potentials (LFPs) from multiple brain regions with precise gain matching. A novel low-complexity phase estimator and neural connectivity processor subsequently enable energy-efficient, yet accurate measurement of the instantaneous phase and cross-regional synchrony measures. Through flexible combination of neural biomarkers such as phase synchrony and spectral energy, a four-channel charge-balanced neurostimulator is triggered to treat various pathological brain conditions. Fabricated in 65-nm CMOS, the SoC occupies a silicon area of 2.24 mm2 and consumes 60 μW, achieving over 60% power saving in neural connectivity extraction compared to the state-of-the-art. Extensive in-vivo measurements demonstrate multi-channel LFP recording, real-time extraction of phase and neural connectivity measures, and phase-locked stimulation in rats.

基于机器视觉的羊只目标提取算法设计与实现

In order to improve the quality of sheep foreground object segmentation, images are segmented using the watershed algorithm in combination with a growing region algorithm, and the pixel-by-pixel comparison of segmentation is optimized to reduce the processing time. Compared with other algorithms, the optimized watershed algorithm can achieve more complete target extraction, and its processing time is improved by over 50% compared with the other six algorithms. Moreover, the optimized watershed algorithm has the optimal overall image quality indicators. This algorithm can provide a reference for the real-time extraction of the activity state of sheep.

A novel moving window-based power spectrum features for single-channel EEG classification using machine learning

Electroencephalogram (EEG) signal classification is a crucial and very difficult task. Meanwhile, extracting features that are representative and able to discriminate different types of EEG signals is a complex task. Such features are usually fed to machine learning algorithms to classify the EEG signals based on the extracted features. This paper proposed a highly accurate and real-time features extraction method that can be used to help physicians in detecting different types of seizures and states in EEG signals characterized by a set of features extracted from the power spectrum of the EEG window. This is achieved by applying the following four steps. First, the EEG signals dataset contains different classes of EEG signals: Normal Eye Closed, Normal Eye Opened, Focal Seizure, Non-Focal Seizure, and Ictal Seizure activities. Second, each EEG signal has a length of 4097 samples sampled with a sampling frequency of 173.6 Hz which resulted in 23.6 seconds in length, this signal will be truncated into windows (Sub-signals) with a length of 349 samples (Approximately 2 seconds) with a total number of 12 windows for each signal. Afterward, the Fourier Transform (FT) based power spectrum will be computed for each window, then a set of different features are extracted from each window's FT power spectrum, and these features are classified using different Machine Learning (ML) algorithms. The results showed that the proposed methodology yields around 98% accuracy for the five different classification scenarios using different ML algorithms. The suggested method is hence robust, fast, real-time, accurate, and simple.

MobileSP: An FPGA-Based Real-Time Keypoint Extraction Hardware Accelerator for Mobile VSLAM

Keypoint extraction is a key technique for Visual Simultaneous Localization and Mapping (VSLAM). Recently, Convolutional Neural Network (CNN) has been used in the keypoint extraction for improving the accuracy. As one of the state-of-the-art CNN based keypoint extraction techniques, the SuperPoint ranked top in the CVPR2020 image matching challenge. However, the use of complex CNN makes it difficult to meet the real-time performance on a mobile platform with limited resource such as mobile robots and wearable Augmented Reality (AR) devices. In this work, based on the SuperPoint, we proposed an FPGA-based real-time keypoint extraction hardware accelerator through algorithm-hardware co-design for mobile VSLAM applications, which is named as MobileSP. Several algorithm and hardware level design techniques have been proposed to reduce the computation and improve the processing speed while maintaining high accuracy, including a partially shared detection & description encoding architecture, a pre-sorting based Non-Maximum Suppression (NMS) engine and a software-hardware hybrid pipeline computing technique. The design has been implemented and evaluated on a ZCU104 FPGA board. It achieves real-time performance of 42 fps with low Absolute Trajectory Error (ATE) of 1.82 cm simultaneously, outperforming several state-of-the-art designs.

Real-Time Extraction and Annotation of Social Media Contents for Predicting National Consumer Confidence Index

Real-Time Extraction and Annotation of Social Media Contents for Predicting National Consumer Confidence Index

Hybrid endoscope with television and multispectral image processing for the internal organs cancer early diagnostics

Endoscopes are widely used to diagnose various internal diseases. The most advanced endoscopes contain miniature television cameras built into the probe which provide high spatial resolution. However, traditional television cameras significantly distort the spectral composition of the transmitted image because in the RGB system they reproduce only limited width spectral lines. However, the full optical spectrum of an image of an internal organ under examination provides a broader range of information that can be crucial in cancer diagnostics. Thus, the spectral resolution of the image plays an important role. Multispectral processing has been suggested for efficient detection of both spatial and spectral information simultaneously. As a result, a series of monochrome sub-images can be obtained, each of which corresponds to a selected resolved spectral interval. Acousto-optical tunable filters are used as a selective element allowing real-time extraction of sub-images. However, an increase in spectral resolution is accompanied by a decrease in spatial resolution. A new hybrid endoscope was suggested to solve this problem. It implements the method of combined application of a TV camera and a multispectral processing device. It is shown that the suggested method of hybrid endoscopy makes it possible to ensure high resolutions: spatial resolution with a camera and spectral resolution (no less than a hundred of resolvable intervals) due to multispectral processing. The resulting optical images reveal a clear outline of the neoplasm area. The results of multispectral processing diagnose the malignant nature of the neoplasm. A schematic diagram of the proposed endoscope is presented, and the principles of control are shown. The characteristics of the television and multispectral images are determined. An estimation of the linearity characteristics of the acousto-optical filters shows the possibility of transmitting the maximum amount of information due to the linearity of the signal transformation in the filter sections. Implementation of the proposed method in the design of a hybrid endoscope allowed obtaining the most complete spectral information about neoplasms and providing an opportunity of diagnostics of cancer of internal organs at earlier stages of development.

Multilevel Authentication Employing Watermarking based on QR Codes, Mobile OTP, and Hadamard Transformation

High-speed internet technology has made it possible for users to transact with banks more quickly thanks to online banking services. The increased connectivity also creates numerous online security holes like phishing and pharming. By confirming the user's legitimacy, this mobile banking problem may be resolved. Here, we suggest a hybrid system that makes use of QR codes, One-Time Passwords (OTPs), and digital watermarking—a method of data concealment. The bank generates QR codes using OTP as the key to the watermark sequence. Here, OTP can be utilised for the initial authentication verification. The sequence for the second level of verification is concealed throughout the watermarking process using the Hadamard transformation. Online transactions may be made with maximum security thanks to the OTP and watermark combination. One benefit of this strategy is that the QR code may be created within a limited design window with a suitable real-time extraction method. Performance may be assessed by utilising an android application, including the real-time extraction using a mobile device.

Song Recommendation System Based on Real-Time Facial Expressions

Abstract: Traditional methods of playing music according to a person's mood required human interaction. Migrating to computer vision technology will enable the automation of the such system. This article describes the implementation details of a real-time facial feature extraction and emotion recognition system. One way to do this is to compare selected facial features from an image to a face database. Recognizing emotions from images has become one of the active research topics in image processing and applications based on human-computer interaction. The expression on the face is detected using a convolutional neural network (CNN) for classifying human emotions from dynamic facial expressions in real time. The FER dataset is utilized to train the model which contains approximately 30,000 facial RGB images of different expressions. Expression-based music players aim to scan and interpret data and build playlists accordingly. It will suggest a playlist of songs based on your facial expressions. This is an additional feature of the existing feature of a music player.

Adaptive evolutionary jellyfish search algorithm based optimal photovoltaic array reconfiguration under partial shading condition for maximum power extraction

This paper proposes an adaptive evolutionary jellyfish search algorithm (AEJSA) to optimally reconfigure photovoltaic (PV) array under partial shading condition (PSC) for real-time maximum power extraction. Jellyfish search algorithm (JSA) is selected owing to its effectiveness for real-time optimization. Besides, a series of discrete operations are performed on JSA to solve the discrete optimization problem of PV array reconfiguration. Due to the inherent drawback of JSA that it is easy to trap at the local optimal solution, an adaptive threshold for changing search mechanism is adopted to balance the local exploration and global exploitation. If the number of times that the value of objective function keeps unchanged exceeds this threshold, three operations (exchange, moving, and inver-over) will be implemented on the whole population for a wide global exploitation. In addition, to verify the feasibility of the hardware implementation of AEJSA, a hardware-in-the-loop test on a RTLAB platform is employed. Eleven meta-heuristic algorithms are applied and compared to AEJSA under objective PSC and subjective PSC to evaluate the optimized performance of AEJSA under various shadow conditions. The simulation results show that the mismatched power loss obtained by AEJSA is smallest, which reduced by 7.26% against gravitational search algorithm.

A power-aware ECG processing node for real-time feature extraction in WBAN

In modern WBAN (Wireless Body Area Network) applications, wearable sensor nodes are advancing towards in-node detection of cardiovascular diseases with continuous observation of ECG signal. However, a robust power-aware architecture with in-node processing for extraction of cardiac events triggers challenges due to constraints in hardware resources in addition to intensive computational complexity. In this work, a dual mode based transmission system depending on battery-level is discussed that reduces bitstream load of raw ECG signal for saving transmission power. For encoding the bitstream, a power-aware ECG processing architecture is proposed that performs a real-time feature extraction for continuous health monitoring. The proposed architecture explores the advantage of low computationally complex operations focusing on a resource constraint sensor node. The architecture deals with two types of cardiac anomalies i.e., irregular heart rate variation (HRV) and 1st degree AV block; detecting R-peak and P-peak through adaptive thresholding technique. The overall sensitivity and positive predictivity for the proposed algorithm reach upto 99% while validated with the MIT-BIH arrhythmia database. The proposed architecture is synthesized in TSMC 90 nm technology node with an estimated power consumption of 2.11 µW, consuming an area overhead of 0.087 mm2. The architecture operates at 1.2 V with an operating frequency of 10 KHz.

A 60-GHz Hybrid FMCW-Doppler Radar for Vibration Detection With a Robust I/Q Calibration Method

In the field of noncontact sensing, both the continuous-wave (CW) Doppler radar and the frequency-modulated CW (FMCW) radar have their own superiorities. A combination of these two radar modes would be a suitable choice to accommodate various application scenarios. This article presents a 60-GHz hybrid FMCW-Doppler radar transceiver fabricated in a 65-nm CMOS process. An on-chip frequency quadrupler driven by a 15-GHz external signal source is utilized for the 60-GHz hybrid waveform generation. Based on the hybrid operating modes, a robust I/Q imbalance calibration method is also proposed. Specifically, in the Doppler radar mode, the imbalance factors are extracted by evaluating the quadrature beat signals during the FMCW period, and then the interferometry signals are calibrated to improve the detection performance. Unlike the trajectory-based ellipse-fitting method, the accuracy of the estimated imbalance factor is independent of the target displacement. Compared with the test-signal-based imbalance factors extraction ahead of the target detection, the real-time extraction during the target detection is more efficient and can also mitigate the estimation errors caused by environmental variations. With the proposed hybrid radar system and the I/Q calibration method, an 80-mm linear displacement is recovered with a 0.016-mm root mean square error (RMSE), and a periodic vibration with a 1- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu \text{m}$ </tex-math></inline-formula> amplitude is detected with a 16.4-dB signal-to-noise ratio (SNR). Multiple-target detection is also performed in the FMCW radar mode, and the vibration trajectories are recovered accurately.

Control Design for the NPC Three-Level Grid-Connected Converter with Non-Ideal Grid System

Background: The large-scale distributed new energy integrations bring about various non-ideal grid situations, which could deteriorate the control performance of grid-connected converters. Objective: This study aims to improve the control performance of the grid-connected converters when the grid is non-ideal. Methods: Firstly, a neutral point clamped (NPC) three-level grid-connected converter was studied, and the characteristics of the non-ideal grid were analyzed in detail. Then, a real-time signal extraction method called generalized sliding discrete Fourier transform (GSDFT) has been proposed to extract the non-ideal signals, e.g., harmonic components and negative sequence components, based on which a double closed-loop control strategy was designed to improve the control performance of the grid-connected converters when the grid is non-ideal. Results: The experimental results show that the converter double closed-loop control strategy in the non-ideal grid environment is suitable for a variety of non-ideal grid environments, and the gridconnected effect is good. Conclusion: Firstly, this paper analyzes the characteristics of non-ideal grid and establishes the mathematical model of NPC three-level grid-connected converter in non-ideal grid environment. Secondly, aiming at the problem of characteristic harmonics and negative sequence components in non-ideal power grid, a real-time and fast extraction method of non-ideal signals based on generalized sliding discrete Fourier transform is proposed. Finally, taking the three-phase grid current of grid-connected converter as the control target, the closed-loop control strategy of non-ideal signal extraction based on GSDFT is designed, and a good control effect is achieved.

Quick Generation of SSD Performance Models Using Machine Learning

Increasing usage of <i>Solid-State Drives</i> (SSDs) has greatly boosted the performance of storage backends. SSDs perform many internal processes such as out-of-place writes, wear-leveling, and garbage collection. These operations are complex and not well documented which make it difficult to create accurate SSD simulators. Our survey indicates that aside from complex configuration, available SSD simulators do not support both <i>sync</i> and <i>discard</i> requests. Past performance models also ignore the long term effect of I/O requests on SSD performance, which has been demonstrated to be significant. In this article, we utilize a methodology based on machine learning that extracts history-aware features at low cost to train SSD performance models that predict request response times. A key goal of our work is to achieve real-time or near-real time feature extraction and to achieve practical training times so our work can be considered as part of solutions that perform online or periodical characterization such as adaptive storage algorithms. Thus, we extract features from individual read, write, <i>sync</i> , and <i>discard</i> I/O requests and use structures such as exponentially decaying counters to track past activity using <inline-formula><tex-math notation="LaTeX">$O(1)$</tex-math></inline-formula> memory and processing cost. To make our methodology accessible and usable in real-world online scenarios, we focus on machine learning models that can be trained quickly on a single machine. To massively reduce processing and memory cost, we utilize feature selection to reduce feature count by up to 63%, allowing a feature extraction rate of 313,000 requests per second using a single thread. Our dataset contains 580M requests taken from 35 workloads. We experiment with three families of machine learning models, a) decision trees, b) ensemble methods utilizing decision trees, and c) <i>Feedforward Neural Networks</i> (FNN). Based on these experiments, FNN achieves an average <inline-formula><tex-math notation="LaTeX">$R^2$</tex-math></inline-formula> score of 0.72 compared to 0.61 and 0.45 for the Random Forest and Bagging, respectively, where <inline-formula><tex-math notation="LaTeX">$R^2 \in (-\inf, 1)$</tex-math></inline-formula> of 1 indicates a perfect fit. However, while the random forest model has lower accuracy, it uses general processing hardware and can be trained much faster, making it viable for use in online scenarios.

Real-time determination of sandy soil stiffness during vibratory compaction incorporating machine learning method for intelligent compaction

An emerging real-time ground compaction and quality control, known as intelligent compaction (IC), has been applied for efficiently optimising the full-area compaction. Although IC technology can provide real-time assessment of uniformity of the compacted area, accurate determination of the soil stiffness required for quality control and design remains challenging. In this paper, a novel and advanced numerical model simulating the interaction of vibratory drum and soil beneath is developed. The model is capable of evaluating the nonlinear behaviour of underlying soil subjected to dynamic loading by capturing the variations of damping with the cyclic shear strains and degradation of soil modulus. The interaction of the drum and the soil is simulated via the finite element method to develop a comprehensive dataset capturing the dynamic responses of the drum and the soil. Indeed, more than a thousand three-dimensional (3D) numerical models covering various soil characteristics, roller weights, vibration amplitudes and frequencies were adopted. The developed dataset is then used to train the inverse solver using an innovative machine learning approach, i.e. the extended support vector regression, to simulate the stiffness of the compacted soil by adopting drum acceleration records. Furthermore, the impacts of the amplitude and frequency of the vibration on the level of underlying soil compaction are discussed. The proposed machine learning approach is promising for real-time extraction of actual soil stiffness during compaction. Results of the study can be employed by practising engineers to interpret roller drum acceleration data to estimate the level of compaction and ground stiffness during compaction.

Predicting Successful Throwing Technique in Judo from Factors of Kumite Posture Based on a Machine-Learning Approach

Identifying the key points of a movement performed by an expert is required for beginners who want to acquire a motor skill. By repeating a learning cycle, the beginner tries the movement, focusing on the key points. We can find many guiding methods for adopting motor skills in the fields of coaching and training for sports. However, the methods strongly depend on the experience of trainers and coaches, who need to select the appropriate methods for different types of athletes. Although methods based on objective information obtained from videos and sensors applicable to individual movements have been proposed in order to overcome the subjectivity of these approaches, we cannot apply those to movements that include external factors, such as pushing and/or attacks from an opponent, as seen in combat sports. Furthermore, such sports require fast feedback of the analysis to the athletes in order to find the key factors of offensive/defensive techniques at the training site. Focusing on judo throwing techniques, this paper proposes a novel real-time prediction method called RT-XSM (Real-Time Extraction method for Successful Movements) that predicts which throwing technique is most likely to be successful based on Kumite posture just before the throw. The RT-XSM uses logistic regression to analyze datasets consisting of the factors of Kumite posture (a standing posture when both players grip each other) and throwing technique classification. To validate the proposed method, this paper also demonstrates experiments of the RT-XSM using datasets acquired from video scenes of the World Judo Championships.

Real-time snowy weather detection based on machine vision and vehicle kinematics: A non-parametric data fusion analysis protocol

Introduction: This study introduces a new analysis protocol for detecting real-time snowy weather conditions on freeways by utilizing trajectory-level data extracted from the Second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study (NDS) dataset. The data include parameters reduced from a real-time image feature extraction technique, time series data collected from external sensors, and CANbus data collected by the NDS ego-vehicles. To provide flexibility in winter maintenance, two segmentation types of one-minute and one-mile segments were used to sample snowy trips and their matched clear weather trips. Method: In this study, four non-parametric models were developed using six data assemblies to detect snowy weather on freeways. The data assemblies are arranged based on three data sources, including image database extracted from an in-vehicle video camera, sensors, and CANbus data, to examine the effectiveness of snow detection models for different data types considering real-time availability of data. Results: Overall, the developed models successfully detected snowy weather on freeways with an accuracy ranging between 76% to 89%. Results indicated that high accuracy of estimating snowy weather can be accomplished using the data fusion between external sensors data and texture parameters of images, without accessing to CANbus data. Practical Applications: Practical applications can be driven with respect to the time or distance coordinates, using different data fusion assemblies, and data availability. The study proves the importance of employing vehicles as weather sensors in the Connected Vehicles (CV) applications and Variable Speed Limit (VSL) to improve traffic safety on freeways.

Deep-learning informed Kalman filtering for priori-free and real-time hemodynamics extraction in functional near-infrared spectroscopy.

Separation of the physiological interferences and the neural hemodynamics has been a vitally important task in the realistic implementation of functional near-infrared spectroscopy (fNIRS). Although many efforts have been devoted, the established solutions to this issue additionally rely on priori information on the interferences and activation responses, such as time-frequency characteristics and spatial patterns, etc., also hindering the realization of real-time. To tackle the adversity, we herein propose a novel priori-free scheme for real-time physiological interference suppression. This method combines the robustness of deep-leaning-based interference characterization and adaptivity of Kalman filtering: a long short-term memory (LSTM) network is trained with the time-courses of the absorption perturbation baseline for interferences profiling, and successively, a Kalman filtering process is applied with reference to the noise prediction for real-time activation extraction. The proposed method is validated using both simulated dynamic data and in-vivo experiments, showing the comprehensively improved performance and promisingly appended superiority achieved in the purely data-driven way.

Characteristic Analysis of a Chipless RFID Sensor Based on Multi-Parameter Sensing and an Intelligent Detection Method.

Chipless radio frequency identification (RFID) technology has been widely used in the field of structural health monitoring (SHM), but most of the current research mainly focuses on the detection of mechanical properties and there are few studies on the multi-physical parameters (for example, temperature and humidity) in the climatic environment around the structure. Thus, it is necessary to design a small and compact sensor for multi-parameter detection. This paper proposes a multi-parameter chipless RFID sensor based on microstrip coupling, which supports 4-bit ID code and integrates two detection functions of temperature and humidity. Through linear normalization fitting, the sensitivity of the sensor is about 2.18 MHz/RH in the ambient relative humidity test and the sensitivity of the sensor is about 898.63 KHz/°C in the experimental test of water bath heating from 24.6 °C to 75 °C. In addition, this paper proposes an engineering application detection method, designs a lightweight dynamic spectrum detection and wireless transmission platform based on a lightweight vector network analyzer (VNA) and realizes the real-time extraction and transmission of RFID spectrum sensing data. The means are more flexible and economical than traditional experimental scenarios.