Keywords: Big Data Analytics, High Performance Computing

Keywords: Business Intelligence, Crowd Sourcing & Social Intelligence

Keywords: Biomedical and Health Informatics

Keywords: Machine Learning, Data Mining, Neural Networks and Deep Learning
Abstract: The devastation caused by fires is a significant threat to human life. There are traditional fire extinguishing methods but can have negative impacts on the environment. This study utilized data from a system that uses sound waves to extinguish fires without requiring water and chemicals. This paper created machine learning models that can predict whether a fire can be extinguished by the sound waves given the features like the size, fuel, distance, decibel, airflow, and frequency. The researchers used Python programming to create different machine learning models and determined the most accurate model using the classification accuracy and F1 score as performance metrics. The XGBoost model was identified as the most effective in classifying the sound wave flame extinction with accuracy scores of 98.31% and 98.62% for the model with default and optimized parameters, respectively.

Keywords: Machine Learning, Neural Networks and Deep Learning
Abstract: Classifying and mapping water bodies is more significant and essential to human life. Identifying water availability, degradation and disappearance with respect to various seasonal variations and climate variations using change detection techniques and water indices are current research challenges due to temporal variability of spectral, temporal and spatial characteristics of different reflectance bands of Landsat dataset. Especially water bodies such as sea, ocean, lakes, river and glaciers have different spectral and spatial reflectance values. Further, presence of the thematic classes in the imagery leads to misinterpretation error and it is highly complex to obtain the ground-truth data for the change in the multispectral images on the same topographical zone. In order to manage these challenges, an effective deep learning architecture is designed. In this work, a Dense Convolutional Neural Network for change detection and novel water indices is presented. Proposed model is capable of classifying the water bodies in the multispectral images in addition to detecting and quantifying the site-specific changes due to climate and seasons on the basis of spectral and spatial reflectance values. Initially endmembers considered as multivariate components are extracted using sparse principal component analysis (PCA). PCA is capable of handling of nonchanging pixels and continuous narrow bands in the multispectral satellite data for various water bodies.

Keywords: Machine Learning
Abstract: Dissolved copper in water is considered before it can be used for any purpose, especially in the case of Marinduque, Philippines where mine tailing containing high concentrations of the metal was spilled to various water resources. This study aimed to use an optical fiber amplifier to record the color reaction being produced by soluble copper and develop a machine-learning model that determines the level of soluble metals in water. The research utilized a prototype and AI development framework. Preparation of the image processing device and fiber optic sensor (BV501s), the preliminary gathering of baseline data, system modeling & development, and System Evaluation was conducted. The BV501s optical fiber sensor was used as it provides RGB equivalents of tannic reaction to soluble copper. There was a total of 33 samples each with its segmented RGB equivalents. The model for identification of the amount of soluble metal was developed using two (2) models such as multiple regression analysis and support vector regression. Using the applied methodology, copper amounts can be determined in water using regression models. The amount of soluble copper can be identified more accurately using the support vector regression since it yields a higher r2 score compared to the multiple linear regression model using k-fold cross-validation. Integration of the developed model to hardware is also suggested as and comparison of the developed model. Develop a classification model for the potabilit

Keywords: Neural Networks and Deep Learning, Machine Learning, Crowd Sourcing & Social Intelligence
Abstract: The integration of data science and marine science into a single platform has led to a revolution in the understanding of oceanographic processes. Sea Surface Temperature (SST) prediction plays a vital role in various fields, namely marine ecology, climate change studies, and environmental forecasting. This paper delves into the most recent advancements in SST prediction techniques and their impact on oceanographic exploration. Moreover, it presents a novel model aimed at addressing the limitations of previous methodologies. The utilisation of advanced Deep Learning and Machine Learning architectures has significantly improved the accuracy of the SST forecasts, surpassing the less accurate results previously obtained through numerical models. Modern techniques can capture spatial correlations and temporal dependencies in SST data. This enables predicting SST values more reliably and accurately. These cutting-edge discoveries provide valuable insights into oceanographic phenomena, aiding in the enhanced understanding of the ocean and bolstering our capacity to predict and comprehend significant and captivating climate events. This study underscores the importance of leveraging the critical role of harnessing the vast advancements in SST prediction to advance marine science and facilitate informed decision-making across diverse sectors related to the marine realm.

Keywords: Neural Networks and Deep Learning, Machine Learning
Abstract: Problems posed by solid waste management have been raising alarming threats today. The increasing number of garbage clogging the drainage systems and the limited space for waste disposal are some of the vivid indications of the waste crisis. One of the solutions for this problem is an intelligent system for classification and segregation of non-biodegradable wastes is implemented with the aid of Convolutional Neural Networks. The system is trained with an initial dataset coming from the images of the waste categories such as plastic bottles, plastic wrappers, plastic cups, and metal canneries. Through VGG16 deep learning architecture, the system can identify the garbage input and classify them accurately. The significance of this study is regarding automation of the garbage segregation process in building an image classifier model that can be deployed into the Materials Recovery Facilities. This is where they sort and market recyclable wastes for the user-end manufacturers. In this manner, it lessens the production of synthetic materials and magnifies the recycling process. The results are shown through graphical representations of the total accuracy of the system against the images subjected in the testing. Although indirect, this research serves as a solution to present the capability of CNN to solve real-world situations.

Keywords: Machine Learning, Neural Networks and Deep Learning, Data Mining
Abstract: In the context of increasing digital society with pressing privacy concerns, our research investigates the effectiveness of privacy-preserving artificial intelligence solutions like Federated Learning (FL). This study focuses on three main areas: Federated and Centralized Learning (CL) applications, the influence of data heterogeneity on client data accuracy, and the evaluation of contemporary federated algorithms in scenarios of extreme heterogeneity. FL, using the FedAvg algorithm, demonstrated superior testing accuracy (88.54%) over CL (87.98%) on the non-heterogeneous CIFAR-10 dataset, indicating its potential as an efficient, privacy-preserving solution for various machine learning applications. Additionally, our findings highlight an inverse relationship between data heterogeneity and FL model accuracy, underscoring the need for strategies to mitigate this challenge and boost model performance. Upon evaluating several FL algorithms under high data heterogeneity, SCAFFOLD and FedOpt outperformed FedAvg and FedProx, demonstrating the significance of algorithm design in addressing data heterogeneity. SCAFFOLD and FedOpt showcased greater communication efficiency, attributed to their faster convergence and fewer required communication rounds. This study offers invaluable insights into addressing data heterogeneity, improving communication efficiency, and enhancing FL's performance and applicability in real-world scenarios, thereby furthering privacy-preserving AI research.

Keywords: Machine Learning, Neural Networks and Deep Learning, Human Computer Interface
Abstract: This research paper presents an automated system developed using Jetson Nano and the YOLOv5n6 model for efficient and real-time detection and classification of pavement damage. The system offers a promising solution for transportation agencies in countries with extensive road networks, such as the Philippines, by reducing the need for manual inspections and streamlining maintenance efforts. By leveraging deep learning techniques, the proposed system demonstrates high accuracy in identifying various types of pavement damage, including cracks, alligator cracks, and potholes. The system's deployment on Jetson Nano provides efficient processing capabilities, enabling real-time analysis of video feeds from road cameras or mobile devices. The results of comprehensive evaluations indicate the system's adaptability to varying environmental conditions and its potential for large-scale implementation. The automated system contributes to cost savings, improved road safety, and enhanced management of pavement quality.

Keywords: Machine Learning, Neural Networks and Deep Learning
Abstract: The detection and interpretation of human emotions using a variety of sensory cues, including visual, auditory, body language, and physiological indicators, is commonly referred to as emotional recognition. This technology has a broad range of applications, including medical diagnosis, customer service satisfaction, and intelligent call center routing. While facial expressions represent one of the most conspicuous emotional cues utilized in emotional recognition, the integration of additional expressions may enhance its efficacy. To this end, this study proposes an integration of facial and speech expressions as emotional features for the prediction of emotions in machine learning models. The emotional predictions from facial and speech recognition models are subsequently combined to generate the final emotional prediction. Several facial and speech benchmark datasets are employed to evaluate the proposed methodology. The experimental results demonstrate significant promise and the continued development of emotional recognition as a supplementary system for assisting doctors in clinical and research studies appears feasible.

Keywords: Machine Learning, Neural Networks and Deep Learning
Abstract: Software testing is a crucial component of software development. With the increasing complexity of software systems, traditional manual testing methods are becoming less feasible. Artificial Intelligence (AI) has emerged as a promising approach to software testing in recent years. This review paper aims to provide an in-depth understanding of the current state of software testing using AI. The review will examine the various approaches, techniques, and tools used in this area and assess their effectiveness. The selected articles for this study have been extracted from different research databases using the advanced search string strategy. Initially, 40 articles have been extracted from different research libraries. After gradual filtering finally, 20 articles have been selected for the study. After studying all the selected papers, we find that various testing tasks can be automated successfully using AI (Machine Learning and Deep Learning) such as Test Case Generation, Defect Prediction, Test Case Prioritization Metamorphic Testing, Android Testing, Test Case Validation, and White Box Testing. This study also finds that the integration of AI in software testing is making software testing activities easier along with better performance. This literature review paper provides a thorough analysis of the impact AI can have on the software testing process.

Keywords: Machine Learning
Abstract: Federated optimization or federated learning (FL) involves optimization of the global model or the server model by minimizing the global loss function which is weighted average of all the local loss functions. The optimization of the global model requires faster convergence to reduce the number of communication rounds or global iterations which is one of the major challenge in federated optimization. This paper propose FONN which handles this communication overhead in federated optimization by utilizing Nys-Newton, while updating local models. As compared to existing state-of-the-art FL algorithms, SCAFFOLD, GIANT and DONE, utilization of Nys-Newton leads to better convergence and reduction in communication rounds or global iterations while achieving a desired performance from the global model which may be observed from the experimental results on various heterogeneously partitioned datasets.

Keywords: Modulation and Coding Techniques, Wireless Communications and Networks
Abstract: This work focuses on the FPGA implementation of the 3GPP complaint polar encoder for 5G-new radio (NR). We use Xilinx's polar intellectual property (IP) aka polar IP for the polar encoding functionality. We design a polar configuration IP by generating the bit allocation (BA) table, configuring register values and passing the right control parameters. Next, a parameter handler is developed which is a memory mapped module and provides an advanced extensible interface (AXI)-lite interface to the polar IP. To test the functionality of the polar IP, a Verilog test bench is developed. It provides configuration parameters, payload and BA table in compliance with 3GPP standards to the polar configuration IP. We verify the output of our implementation with the Matlab 5G-NR built-in library on the Xilinx evaluation board. Our results show the efficiency of the implementation in terms of latency and resource utilization on the evaluation board.

Keywords: Modulation and Coding Techniques
Abstract: 5G new radio (NR) distinguishes itself by offering remarkable features such as significantly higher throughput and lower latency compared to 4G. In 5G, low density parity check (LDPC) encoding is employed as the channel coding scheme for the transmission of data bits. Subsequent to the LDPC encoding stage, bit selection and bit interleaving operations are performed. Rate matching plays an important role in selecting specific encoded bits for transmission, utilizing techniques like shortening and repetition to support hybrid automatic repeat request functionality. To address the issue of latency encountered in processing large transport blocks, we propose parallel algorithms for the rate matching. These algorithms are implemented on field programmable gate arrays, and then the performance of the optimized algorithms is compared with the existing algorithms specified in the 3rd generation partnership project standards.

Keywords: Wireless Communications and Networks, Vehicular Networks, Modulation and Coding Techniques
Abstract: Abstract— In this paper, we propose a strategy to enhance the performance of task assignment and path planning in applications of distributed multiple unmanned aerial vehicles (multi-UAV). Multi-UAVs are made up of many small UAVs with limited mission resources. They can operate autonomously, appropriately, and universally. UAV swarm task coordination and resource allocation can be achieved with a reasonable allocation of UAV tasks and resources according to the UAV region and its own performance. Based on the in-depth research of the traditional auction algorithms, this work proposes a method that can improve multi-UAV task allocation efficiency, namely the Integer Linear Programming (ILP) algorithm. We are proposing ILP formulation for numerous drones and tasks in an environment that enables the completion of tasks. We also measure the computational time requirements of the proposed ILP approach with various numbers of drones and tasks, so that the limit of this approach can be identified for practical application. This research overcomes the difficulties between task allocation and path planning. Compared with auction-based algorithms this technique can better complete allocation results and reduce resource consumption. It’s expected that the simulation results will show that the algorithm is effective in computing and task execution efficiency

Keywords: Wireless Communications and Networks, Modulation and Coding Techniques
Abstract: Integrating space-time block coding (STBC) with orthogonal frequency division multiplexing (OFDM) has become a promising wireless communications solution that provides high data transmission rates and improved signal quality. Conventional (orthogonal and non-orthogonal) STBCs had a tradeoff between transmission rates and maximum diversity when applied in STBC-OFDM systems with more than two transmit antenna chains. The Fourier transform-based OFDM also has a high peak-to-average power ratio (PAPR) that deteriorates system performance. This work addresses the mentioned issues using wavelet-transform-based OFDM (WOFDM) and STBCs based on maximum-rank-distance (MRD) codes, i.e., MRD-STBCs. In the proposed MRD-STBC-WOFDM system, MRD-STBCs provide rate-1 transmissions with maximum diversity, while WOFDM enhances bit-error-rate (BER) and PAPR performance. Compared with conventional STBC-OFDM, the proposed system with three transmit antennas has significant BER and PAPR improvement. Moreover, the wavelet transform application in MRD-STBC-OFDM reduces the OFDM's complexity.

Keywords: Wireless Communications and Networks
Abstract: The deployment of fog computing paradigms for securing IoT networks is associated with several advantages, including reduced bandwidth, latency, storage, and computational overhead at cloud servers. However, the fog layer has additional security requirements, such as the establishment of secure channels for key distribution and the overhead of repetitive device authentications. In this paper, we have addressed these issues using a permissioned blockchain-based fog network that validates the edge devices through smart contracts by establishing a mechanism for secure storage and exchange of credentials. The communication between edge devices is carried out through the MQTT protocol, and device registration is performed using a smart contract. The key pairs are generated from the secp256k1 elliptic curve to ensure faster trust-based identity management and authentication of edge devices and gateways. The proposed frameworks ensures that the access to Blockchain network is given exclusively for authenticated devices The implementation of the proposed scheme is performed on private Ethereum 2.0, and the performance is evaluated in terms of node registration time, authentication time, key generation time, and throughput. The implementation results are available on GitHub (https://github.com/Aswani08/tenconresults.git)

Keywords: Electronic devices, materials and fabrication process, Analog and mixed signal ICs
Abstract: As Field Programmable Gate Arrays (FPGAs) become large integration, the area reduction of a Basic Logic Element (BLE), which is a circuit for a logical definition in FPGAs, is required. We have studied a Variable Logic Circuit (νVLC) using neuron CMOS inverters (νCMOSs) and have proposed a 4-input Variable Logic Circuit (4-νVLC) as an alternative circuit against a 4-input Look-Up Table circuit (4- LUT) in the ordinary BLE. In this paper, we describe the 4- νVLC configuration and operations, and verify that the results of the HSPICE simulations and the theoretical performances are in general agreement. In addition, we design a layout of BLEs using 4-νVLC to reduce the BLE area and compare the 4- LUT and 4-νVLC.

Keywords: Electronic devices, materials and fabrication process, Device modeling & characterization, Materials and Structures
Abstract: In this study, we analyzed the measurement results of the frequency dependence of the current collapse in AlGaN/GaN high-electron-mobility transistors and confirmed the gate voltage dependence and temperature dependence in the high frequency region. An oscillation phenomenon was confirmed in the frequency dependence of the drain current. It was suggested that the oscillation phenomenon is caused by electrons moving back and forth between the crystal defect and the two-dimensional electron gas (2DEG) following the frequency. In the high-frequency region (0.8 GHz to 5 GHz), we calculated the inflection point from the oscillation waveform of the drain current, and succeeded in separating the response performance owing to crystal defects that enable high-frequency response. As a method for evaluating 2DEG carrier behavior in high-frequency bands, area value evaluation of areas with frequency dependence was investigated. The area value increases as the temperature rises (22°C to 80°C); however, it is confirmed that the area tends to decrease at higher temperatures (100°C or higher). The calculation result of the area value is linked with 2DEG behavior in GaN crystal.

Keywords: Electronic devices, materials and fabrication process, Device modeling & characterization, Materials and Structures
Abstract: In this study, we experimentally observed the behavior of carriers generated by the photoelectric effect in the GaN crystal layer (channel layer) from the viewpoint of high-frequency characteristics to clarify the carrier transport phenomenon. It was found that the accumulation of holes generated by the photoelectric effect near the source electrode causes a new parasitic capacitance component that depends on the gate voltage.

Keywords: Electronic devices, materials and fabrication process, Device modeling & characterization, Materials and Structures
Abstract: In this study, the current–voltage (IV) and Radio Frequency (RF) characteristics were measured in a low-temperature environment, and the transient response waveform was analyzed by the simplified Isothermal Capacitance Transient Spectroscopy (ICTS) method. According to the measurement results of the IV characteristics, the current value decreased rapidly as the temperature decreased, and the linear region could not be confirmed. From the RF characteristics, it was confirmed that the gain varies with temperature in the high-frequency band. Furthermore, the results suggested that substantial structural changes occurred in the device under low-temperature environments. In addition, it was confirmed by simple ICTS analysis that the capture cross-section and crystal defect concentration increased with decreasing temperature. At room temperature, the thermal energy imparted to the crystal layer induces phonon scattering at crystal lattice points. This implies that the process of trapping electrons in crystal defects is suppressed. In a high-frequency region, carriers trapped in crystal defects are released and the crystal defects become conspicuous. As a result, an increase in the capacitance component of crystal defects was confirmed, and it is thought that both the capture cross-section and the crystal defect concentration increased compared to those at room temperature.

Keywords: Antennas, Propagation and Computational EM, RF/Millimeter-wave Circuits and Systems, THz, mmWave and RF Systems for Communications
Abstract: This paper proposes a cost-effective design approach for metasurface-enabled microstrip antennas optimized for the dual-band of mmWave 5G frequencies, specifically 26 GHz (n258) and 28 GHz (n257). The proposed antennas are based on metasurface and are manufactured using standard printed circuit board (PCB) processes on a low-cost FR-4 substrate, making them suitable for mass production at an affordable price. The dimensions of the proposed antenna are 25x25x7.6 mm³. It demonstrates a |S11| <= -10 dB bandwidth of 31.9% (23.2-32.0 GHz) and offers a higher gain of 7.2 dB compared to a non-metasurface patch antenna with a gain of 5.12 dBi. The proposed antenna achieves a peak gain of 13.8 dBi and maintains a good front-to-back ratio exceeding 15 dB across the entire operating band. Moreover, the use of low-cost FR-4 substrate and PCB fabrication processes ensures cost-effectiveness, making these antennas highly suitable for cost-efficient deployment in 5G applications.

Keywords: Antennas, Propagation and Computational EM, RF/Millimeter-wave Circuits and Systems, THz, mmWave and RF Systems for Communications
Abstract: A low-profile triple band antenna is designed, investigated and developed to use as the indoor mounted ceiling antenna for 4G-LTE and 5G-NR distributed antenna system. The proposed antenna is implemented on 0.8-mm thickness PCB substrate and obtains the compact size of 120 ×135 mm2, which composes of oval patch trimmed by three slots and microstrip feed line on chamfered ground plane with tuning slot. A prototype antenna is measured impedance bandwidth which covers three operating frequency bands of 690 – 960 MHz, 1710 – 2700 MHz and 3300 – 4000 MHz. Also, the average antenna gains are 1.8, 3.2, and 2.6 dBi at each operating band, respectively. The proposed antenna obtains omnidirectional pattern over the entire frequency ranging from 700 until 2500 MHz and bidirectional pattern for the frequency above 2500 MHz. According to comparisons, the measured results are in good agreement with the simulated one.

Keywords: Antennas, Propagation and Computational EM, RF/Millimeter-wave Circuits and Systems
Abstract: A novel fifth-generation (5G) antenna design featuring a single element has been proposed to enhance antenna performance of bandwidth, cross-polarization, and gain parameters. This advanced antenna design incorporates a meander-line radiator and a defective ground plane (DGS) as its core structure. Several techniques have been employed to improve its capabilities. Adding neutralization lines to the antenna radiator enhanced the bandwidth. Additionally, the cross-polarization is reduced, and the antenna gain is enhanced by using an antipodal-Vivaldi DGS (AVDGS). These design elements work together to improve the overall performance of the antenna. The suggested single-element antenna possesses a multi-wavelength structure with dimensions of Wtimes Ltimes H = 28times 7times 1.6 mm^3, making it suitable for millimetre-wave applications. The antenna operates at a central frequency of 28 GHz. It exhibits a -10 dB fractional bandwidth of 12.78%, a gain of 7 dBi, and a cross-polarization isolation of 17 dB in the E and H planes with the main lobe of the antenna positioned at pm30^circ.

Keywords: Control system modeling, Industrial automation
Abstract: According to the complex dynamic of an overhead crane, developing a payload path planning is a labored task for inexperienced operators. This paper shows the developing tool for 3D payload path planning of industrial overhead cranes dominated by speed control mode. The velocity trajectory generator program based on user interface (UI) was developed using the App Designer in MATLAB. The trajectories are based on the cubic polynomial trajectory and the adaptive command smoother. The cubic polynomial trajectory is used for generating the smoothed path of each waypoint while the payload sway suppression is handled by the smoother. By this concept, the crane moves through the sequentially given waypoints according to given waypoint constraints such as velocity and time. In the program, the user must define the desired waypoint coordinates while the velocity points and time points can be automatically obtained by the program. Furthermore, the program can visualize the workspace and the payload path prediction in 2D and 3D as well as the plot of velocity trajectories in the three axes. The travelling path information is also provided to evaluate the desired path before the velocity trajectories are exported.

Keywords: Industrial automation, Intelligent control, Neuro-control, Fuzzy control and their applications
Abstract: The sorting operations of cherry are conducted manually by cherry farmers; however, they require skillful workers to examine each fruit, and evaluation is subject to variation. This study analyzed the performance of an automatic cherry sorting system. The key to sorting cherries is to sort the fruits accurately without producing damage, along with quick size estimation and grade identification. An image analysis system using the hue saturation lightness (HSL) color model to identify the grade of cherries was implemented. Size was estimated by the equator diameter of the fruit, and grade was determined by the percentage of colored area on the surface of the fruit. A prototype system for automatic cherry sorting into three grades was developed. A second version of the prototype was developed to include size estimation. In both prototypes, an air eject mechanism with individual collection bins were prepared for a non-invasive handling of the fruits. A stable accuracy of 72.4% for size estimation and 82.5% for grade identification were achieved. The findings favor the use of automatic systems for the sorting of cherries according to fruit grade in combination with fruit size.

Keywords: Industrial automation
Abstract: Edge-enabled Deep Learning (DL) solutions for Predictive Quality Inspection (PQI) of products in Industry 4.0 are mostly designed for static manufacturing environments. In general, modern manufacturing processes are dynamic in nature. In this context, continual learning-based model retraining accommodates the dynamism for PQI of multiple processes (tasks) using a single DL model. However, the impact of the task ordering in sequentially arriving tasks and solution to reduce this impact on the overall PQI is yet to be solved. To this end, a novel mechanism using a light-weight similarity analysis module is introduced in the quality prediction system at the resource-limited edge. Sequential training of tasks above a similarity threshold (γ) is preferred, and dissimilar tasks are overlooked to train a separate model. This enables a PQI system to hover over training efficiency and model sustainability. The experimental results validate the impact of task order and the effectiveness of the proposed similarity-based analysis to reduce this impact by 70% on the model’s overall performance in the real-world use case of plastic bricks.

Keywords: Instrumentation systems, Industrial automation
Abstract: Data Acquisition (DAQ) is a vital process used in various fields including engineering, medicine, and scientific research. In the automotive industry, DAQ systems are essential for vehicle testing and validation. This paper addresses the limitations of traditional DAQs by developing a low-cost, robust DAQ system using LabVIEW, a powerful graphical programming language. The objective of this project is to design and implement a cost-effective and flexible DAQ system that meets specific application needs. The system focuses on voltage probing, Controller Area Network(CAN) communication architecture, load bank configurations, and a LabVIEW interface. The methodology involves selecting appropriate peripherals, establishing Microcontroller Unit(MCU)-LabVIEW communication, designing the voltage probe mechanism, setting up CAN communication, incorporating load bank control, and integrating all features into a LabVIEW dashboard. The developed custom DAQ system demonstrated satisfactory performance, with minor voltage reading errors (0.22%) and a small percentage of frame loss (1.95%) in CAN communication. The system’s usefulness has been proved, highlighting its potential as a valuable tool for testing and validating applications in the automobile industry.

Keywords: Big Data Analytics, Data, Text, Web Mining, & Visualization, High Performance Computing
Abstract: This paper presents an approach to enhance Hadoop performance by leveraging deep Q-Learning, a form of Reinforcement Learning, to optimize parameter settings. The performance of Hadoop, a widely adopted distributed computing framework, relies heavily on configuring numerous parameters. However, the complex and extensive search space poses challenges in determining the optimal settings. In response, we propose an innovative deep Q-Learning algorithm that iterative discovers and applies the most effective parameter configurations, thereby improving Hadoop’s performance. Our approach involves defining state and action spaces, learning from performance feedback, and identifying the optimal configuration to maximize Hadoop’s efficiency. Our proposed model consistently outperforms existing solutions by benchmarking with popular jobs such as TeraSort, WordCount, Hive Aggregate, and WordCooccur, demonstrating significant reductions in execution times and improved computational efficiency. This research accelerates big data processing in Hadoop and provides a scalable and efficient solution applicable to similar distributed computing frameworks. Furthermore, our paper reinforces the effectiveness and applicability of Reinforcement Learning in addressing complex optimization problems. By harnessing the power of deep Q Learning, we showcase its capability to navigate the intricate parameter space of Hadoop, resulting in enhanced performance and streamlined data processing.

Keywords: High Performance Computing, Data Centric Programming
Abstract: In the era of artificial intelligence, the popularity of portable devices and the development of Computer Vision (CV) have improved the convenience of human productive life. However, with the substantial increase of application data and the problems of traditional Von Neumann Computing architectures in terms of performance bottlenecks and power consumption becoming more and more prominent, there is an urgent need to propose new computing models and related optimization algorithms. Memory-Centric Computing (MCC) is considered as a good hardware option to solve this problem. This paper also presents a Crop-Sum algorithm that can be widely used for computer vision development and analyzes its feasibility for use in applications such as character recognition and pedestrian detection. The work results show that the text recognition application optimized using this algorithm is significantly improved in terms of performance and power consumption, and that the Crop-Sum algorithm is feasible for implementing MCC and computational optimization.

Keywords: High Performance Computing
Abstract: Real-time embedded systems are designed to perform specific functions in real-time, with a micro controller, memory, and input/output devices. The scheduler is a critical component that manages resource allocation and schedules jobs based on priority and available resources. Multiprocessor platforms improve performance, scalability, redundancy, and flexibility, with different approaches to scheduling, such as global, partitioned, and semi-partitioned. Minimizing dynamic energy consumption and processor temperatures is essential for improving battery life and reliability and meeting power and thermal constraints in applications such as mobile devices, aerospace, and defense systems. There are many techniques for energy and temperature management but their effect on each other has not been studied in detail. Hence, we want to employ a few of those techniques and want to observe their impacts. In this work, we first propose a basic semi-partitioned scheduler for heterogeneous multiprocessor systems which supports the execution of real-time jobs. Then, we apply some well-known energy and temperature minimization techniques over the proposed scheduler to study their impact on the system. To conduct our experiments, we use benchmark programs whose characteristics have been extracted using various simulators.

Keywords: High Performance Computing
Abstract: Climate and Weather Modelling is a highly complex and computationally intensive task which consumes substantial amounts of energy. A desire to improve forecast skill demands further advances in these forecasts, such as increased model fidelity and more comprehensive physical representations of the underlying processes. Another driver towards better forecasts is the goal of uncertainty quantification, with ensembles of forecasts a popular technique. But ensembles place much higher demands on the computational workload as N ensemble members require N times the compute cycles. This leads to an even higher energy demand. This study examines one approach to reducing the energy demands of ensembles by taking advantage of a hardware feature in modern NVIDIA GPUs known as Multi Instance GPU (MIG). This feature allows us to run multiple ensemble members on hardware-isolated GPU slices to maximize efficient use of the GPU resources and subsequently reduce the Carbon Footprint for an Ensemble forecast. We examine both small and large test cases across a range of setups to determine the optimal runtime configuration. Our study shows a 2.5-2.8x reduction in CO2 emissions across all cases which translates into a savings of between 141-171 tonnes of carbon emissions annually per GPU.

Keywords: High Performance Computing
Abstract: Scene Text Recognition (STR) is a task in computer vision that is used to read texts in natural scene images. STR currently suffers from data distribution shift due to the lack of large real datasets for training. Data augmentation is a method that has been used in multiple studies to address this issue. However, performing augmentation also introduces computational overhead during training. In this paper, we propose FastSTRAug, a CUDA-based library of 36 augmentation functions specifically designed for STR. When executed through varying image sizes, FastSTRAug is observed to be significantly faster over its serial counterpart in most functions, reaching up to 380× speedup on larger images.

Keywords: Machine Learning, Neural Networks and Deep Learning
Abstract: Rubber is considered an important material for humankind and it is one of the most important products in Southeast Asian countries. However, the production of rubber could harm the environment due to the conventional use of acid and salt. We propose a rubber drying process using heat and constructed a rubber heating tunnel. We also propose a strategy to determine the time it takes to dry rubber so that the rubber is sufficiently dried without overheating at different temperature levels. We found that this strategy could not make use of conventional curve fitting methods based on least squares since it cannot handle discrete or categorical input data very well. We propose a non-linear Machine Learning regression technique based on neural network and found that neural network has the ability to predict the output variable quite well despite the input variables contain discrete or categorical values.

Keywords: Sustainable manufacturing process, Sustainable product design and development
Abstract: Magnetic Abrasive Finishing (MAF) is advanced finishing techniques which can generate surface finishes at the nanoscale for both magnetic and non-magnetic materials. By enhancing or optimizing the important MAF process parameters, the material's surface finish may be greatly enhanced. The current paper looks into the experimental studies of Hastelloy C- 276 for surface finish improvement (%∆Ra), material removal (MR) and forces (Fn & Ft) as well as the parametric optimization of MAF process. The optimum results obtained after the application of Particle swarm optimization (PSO) and Genetic algorithm (GA) were compared with each other for improving the finishing of Hastelloy C- 276 using the MAF process. The experimentation was carried out using MATLAB software. Using GA and PSO algorithms, the regression equation was utilized to determine the optimal influencing factors. Particle swarm optimization was found to be the best optimization method and to have produced the best optimal outcomes when these optimum results were compared.

Keywords: Sustainable product design and development, Sustainable manufacturing process, Technology for Humanitarian Purposes
Abstract: Aquaponics systems allow simultaneous growth between vegetables and fish with the use of aquatic products. This process is due to the presence of beneficial microbial communities such as nitrifying bacteria. Numerous water quality problems can be addressed by improving the conversion efficiency of biological filtration systems. Maintaining parameters including increased ammonia and nitrite concentration, organic matter accumulation, and decreased levels of dissolved oxygen are imperative to maximize system productivity. The pilot-scale model of the aquaponics system features two independent systems: the control (with biofilter), and the experimental (without biofilter). The media in the biofilter is as follows: an aeration system (for oxygen supply), and a centralized sensory chamber (to facilitate automatic monitoring of pH, temperature, dissolved oxygen, and turbidity). This study assessed the performance of an aquaponic setup with a biofiltration system in terms of reducing the concentration of dissolved solid particles in the water. The correlations of ammonia, nitrate, and nitrate concentrations with turbidity were explored. The turbidity was monitored using a turbidity sensor. To test ammonia, nitrite, and nitrate, solutions were dropped into water samples, whose developed color was compared to a color chart. The system must operate in conditions that cater to the collective growth and development of the fish, plants, and nitrifying bacteria. The plant crop utilized in

Keywords: Sustainable SCM, Sustainable manufacturing process, Sustainable product design and development
Abstract: This study aims to classify the subjective perspectives of foreign government policies on sustainable supply chains for achieving Sustainable Development Goals (SDGs), using inputs from experts in Thailand. In this study, three broad policy categories were identified: supply chain digitalization, integration, and nearshoring. Through a literature review, 24 policies were identified. The Q-sort method was applied to classify these policies based on the perspectives of Thai industry experts, aiming to obtain consensus. The result shows that the inter-rater agreement was substantial. Finally, this study develops a hierarchy model for prioritization of policies for enhancing the supply chain performance aligned with the SDGs to convert a complicated problem to a hierarchical system of elements to enhance understanding of supply chain development policies.

Keywords: Industry 4.0 Adoption towards Sustainable Manufacturing
Abstract: The advent of the fourth industrial revolution (4IR) offers promising improvements in operational efficiency and profitability for various industries, and can be a key component to leapfrog the many sectors of manufacturing in the Philippines. This will necessitate the reskilling and upskilling of Filipino automation engineers and instrumentation technicians to commission and maintain smart technologies for production facilities. The further expansion of the country’s manufacturing capacities presents an opportunity to locally develop products and solutions for process automation. This includes sensor technologies and applications that can boost the Filipino manufacturers’ capabilities through locally-sourced automation components. Various sensor technologies and applications can be explored for further improvement through research & development. Through technology foresight, this study looks into the potential of Filipino technology firms to develop sensors that are locally designed and assembled, addressing the needs of growing industries. The use of scenario-building approach allows for the identification and ranking of the key predictable drivers based on the insights of industry professionals. Future scenarios are developed while opportunities and risks are evaluated.

Keywords: Meta heuristic algorithms
Abstract: In graph theory, the dense subgraph problem aims at finding the densest subgraph for a given graph. The subgraph is defined as a subset of a large graph and the density is calculated as the number of edges as per the number of vertices. Real-life graph networks are complicated and large in size; finding the dense subgraph is difficult as the number of subgraphs is numerous. It is an NP-hard problem and in most cases, it is infeasible to find an exact solution. Here we propose a genetic algorithm approach to solve it. The problem is a combinatorial optimization problem with a constraint that all solutions must be a subgraph of the given graph. Here, the crossover and mutation are designed to produce only feasible solutions. The proposed algorithm is able to give close approximation within a reasonable time

Keywords: Neural Networks and Deep Learning, Machine Learning, Meta heuristic algorithms
Abstract: Real time power quality monitoring is important to ensure stable functioning of the electrical appliances especially for the manufacturing sector. Deep-WT-ConvT is proposed to to better characterise and differentiate the minor differences between different types of power quality disturbances. However, the use of deep networks requires longer training time, and poses the risk of getting internal covariant shift issues due to distribution change in layer's input during training phase. This issue can be prevented by proper parameter initialisation and with lower learning rate, which slows down the training process. Batch normalisation (BN) layers are proposed to improve the classification performance of the PQD classifier network WT-ConvT. Results shows significant improvement on Deep-WT-ConvT model with accuracy improvement from 92.95% without BN layers to 94.44% with BN layers on 20dB SNR AWGN noise test.

Keywords: Neural Networks and Deep Learning, Machine Learning, Pattern Recognition and Object Tracking
Abstract: Point cloud object classification has gained attention from many researchers since the emergence of public dataset like ModelNet and ShapeNet, which contains full surface objects. However, in practice, objects captured using LiDAR are only partially covered in the scanned area, making such a task burdensome. Here, we proposed a solution to overcome those problems. It is a novel fuzzy convolutional inference (FuzzConv) incorporated with depthwise over-parameterization (DOConv). Instead of applying raw data, the point clouds are transformed into a 3D voxel. We utilized EfficientNet as our backbone and modified the Mobile inverted Bottleneck Convolution (MBConv) with DOConv. In the last fully connected (FC) layer, we added the FuzzConv layer as an inference before feeding the feature map to the output layer. Consequently, to validate the performance of our model, we undertake an evaluation with multiple classifications in ModelNet10, ModelNet40, and our core dataset, the point cloud of human poses. Accuracy, loss, number of parameters, loss, precision, and F1-scores are employed as performance indicators. As a result, our model achieved top performance regarding the accuracy and loss value for the primary dataset, 83 % and 0.56, for ModelNet10 88.1 % and 0.56, and ModelNet40 74.1 % and 1.15.

Keywords: Neural Networks and Deep Learning, Machine Learning
Abstract: Quantum machine learning is one of the most exciting new avenues in the world of artificial intelligence, especially because of the enormous computational power of quantum computers and the promise of the development of near error-free quantum computers in the not-so-distant future. For quantum algorithms to be used in real-life applications, quantum computers must be able to work with classical data. One of the key steps in quantum algorithms dealing with classical data is the encoding of classical data points to quantum states, which can then be processed by quantum gates. It is known that the type of encoding technique that works best for a particular network is dependent on the dataset being used. In this paper, a new parallel structure is proposed utilizing two encoding techniques, namely amplitude encoding and angle encoding, for effective classical data classification via quantum neural network. The paper further proposes a maximally expressible and entangled ansatz used to design a simple Quantum Convolutional Neural Network (QCNN) with only 32 parameters, that is used in the latter stages of the network and is kept the same across all encoding instances so that a comparison between the different encoding methods is possible. Extensive experimentation is carried out on two publicly available image datasets, namely MNIST and Fashion MNIST. The results show that the proposed method achieves better results than any of the encoding techniques deployed alone for binary cla

Keywords: Neural Networks and Deep Learning
Abstract: Exceptional advancements in various computer vision tasks, such as identifying and categorizing objects, have been realized through the use of deep learning models, with a particular emphasis on convolutional neural networks (CNNs). Yet, while these models deliver outstanding results, they remain vulnerable to adversarial examples, thereby raising questions about their safety and dependability. In this paper, we investigate the influence of the image characteristics on the efficacy of adversarial patch attack against an image classifier model. We analyzed such characteristics in the frequency domain, where the frequencies represent the varying levels of detail and structural components that contribute to the efficacy of adversarial patches. Our results showed that low-frequency components had significant contribution to the effectiveness of adversarial patch attacks.

Keywords: Wireless Communications and Networks, Cellular Networks, Modulation and Coding Techniques
Abstract: Multi-carrier Non Orthogonal Multiple Access system is considered a promising technique for future wireless communication systems. MC-NOMA system divides transmission bandwidth into sub-bands and multiple users in each sub-band are served based on power-domain NOMA. This work attempts to find the best clustering scheme that can provide the maximum sum rate for NOMA with unequal bandwidth allocation. We try to improve the system performance by tuning the bandwidth allocated to different clusters to find the best clustering scheme. The analytical results compared and validated using the golden search method reveal that pairing up the best channel gain user with the next best channel gain user with optimal bandwidth allocation gives a better sum rate compared to competitive schemes available in the literature.

Keywords: Wireless Communications and Networks, Cellular Networks, Vehicular Networks
Abstract: In this paper, single-input single-output filter bank multicarrier (FBMC) waveform based on offset quadrature amplitude modulation (OQAM) is investigated in conjunction with intelligent reflecting surface (IRS). A frame structure for IRS assisted FBMC waveform is designed for channel frequency response (CFR) estimation, followed by IRS reflecting coefficients optimization. In particular, an ON/OFF channel estimation technique is proposed by inserting guard symbols between the adjacent training symbols. The guard symbols are utilized to mitigate the inter-symbol-interference between the adjacent training symbols, and to help in calculating the inherent intrinsic interference in FBMC waveform. We next investigate a random phase initialization based successive convex approximation technique to jointly optimize the IRS reflecting coefficients and sub-carriers transmit power allocation using both perfect and imperfect CFRs. Our simulation results demonstrate the accuracy of proposed CFR estimation and reflecting coefficient optimization schemes, and the effect of the guard symbols on their performances.

Keywords: Wireless Communications and Networks, Cellular Networks
Abstract: This paper investigates the performance of an uplink multiuser IoT network in which green self-sustaining IoT users utilize the energy harvested from the downlink signal for uplink signalling using NOMA principles. The uplink user with best link signal-to-noise (SNR) is first chosen. To increase spectral efficiency, another user is also picked for concurrent transmission using non-orthogonal multiple access (NOMA) principles. We demonstrate that the choice of the second user depends on the target rate and number of users, and is not always the user with the second-best SNR. The users are selected using a timer-based mechanism, and no feedback of channel estimates is involved. Considering the time-switching protocol for energy harvesting, we obtain expressions for outage probability and throughput of this scheme. Unlike in traditional uplink NOMA networks, the transmit powers here are random, and this makes the analysis and user selection mechanism interesting. Accuracy of the derived expressions is illustrated by computer simulations.

Keywords: Wireless Communications and Networks, Modulation and Coding Techniques, Cellular Networks
Abstract: Cognitive radio (CR) with adaptive modulation schemes can resolve the spectrum scarcity problem and fulfill the high data-rate requirement for the deployment of the next-generation wireless network. Here, we examine the performance of an energy harvesting cognitive radio network (EH-CRN) based on cooperative prediction-sensing, over Nakagami-m fading channels for higher-order modulation schemes like rectangular quadrature amplitude modulation (RQAM) and hexagonal QAM (HQAM). These schemes provide high data rates with improved power and spectral efficiency. For performance analysis, the exact analytical expressions of energy harvesting, throughput, outage probability, and average symbol error rate (ASER) are derived. The impact of fading severity, threshold data-rate, and constellation size on the system performance is highlighted. The simulation results prove the correctness of the derived exact analytical expressions.

Keywords: Wireless Communications and Networks
Abstract: In this research endeavor, the amalgamation of two cutting-edge technologies, Multiple-Input Multiple-Output (MIMO) and Non-Orthogonal Multiple Access (NOMA), offers a solution to the challenges posed by the 5G cellular system and its futuristic counterparts. The study focuses on wireless networks operating in Rayleigh fading channels, which are challenging conditions. The main objectives of the research are to derive closed-form expressions for Bit Error Rate (BER) and the outage probability equation for Downlink (DL) NOMA. Additionally, the investigation extends beyond 5G into the unexplored territory of 6G wireless technology, where the impact of dynamic bandwidth variations is explored. Furthermore, the study evaluates the system's performance by examining the BER, shedding light on the capabilities of 5G and its evolutionary successors. The findings from this research could significantly contribute to advancements in wireless communication technologies.

Keywords: Electronic devices, materials and fabrication process
Abstract: Conventional countermeasures against Power Analysis Attacks ignore area overhead for implementation since it only pursuit perfect tamper-resistance. While the countermeasures achieve high tamper-resistance, unacceptable area overhead is required. Thus, the design of the countermeasures cannot be applied for IoT edge devices, which are provided as cheaper products. In this paper, first, we show the unacceptable area overhead by the conventional countermeasures, then propose a lightweight and fundamental VLSI design method against Power Analysis Attack. Masked Wave Flip-Flop (MW-FF), proposed design method achieves higher tamper-resistance than Wave Dynamic Differential Logic (WDDL) that is one of the conventional countermeasures with small and acceptable area overhead. Our evaluations show that MW-FF requires only 12.51% and 5.89% of are overhead in FPGA and ASIC respectively, which is 114.03% and 21.82% saved comparing with WDDL.

Keywords: Electronic devices, materials and fabrication process
Abstract: The increasing complexity and scale of modern computing systems have led to the emergence of System-on-Chip architectures (SoC), with Network-on-Chip (NoC) architectures serving as a communication infrastructure within these systems. Due to rigid time-to-market constraints, recent SoC designs involve using third-party IPs. This outsourcing can lead to security vulnerabilities in SoC, especially in NoC packet transmission. Considering security attacks, this paper addresses the need for cost-effective secure packet transmission in NoC. We propose a hybrid encryption framework by discussing the challenges associated with symmetric and asymmetric encryption techniques to achieve a balance between security and efficient data transmission in NoC. By combining the strengths of both encryption methods, our approach aims to provide enhanced protection against security threats while minimizing performance overhead.

Keywords: Emerging memory technologies, Analog and mixed signal ICs
Abstract: Space radiation particles cause a malfunction in electric circuits. It is especially susceptible to memory-sensitive storage devices. When it affects data stored in the memory circuit, it causes disruption. Standard 6T SRAM is incapable of mitigating this disruption. Consequently, numerous authors presented various resilience strategies. However, a trade-off exists between memory cell efficiency and soft error probability. This article describes a polar design soft error upset recovery SRAM memory cell (SUR-16T) effectively recovers lost data due to a high-energy particle strike. SUR-16T has superior write stability, lower hold power dissipation, and shorter write access time at PVT variations compared to the mentioned memory cells. Furthermore, SUR-16T has a 0.96x/1.15x/ 1.10x/ 1.18x/ 1.02x/ 1.64x greater critical charge than SEA-14T/ RHBD-13T/ RHMC-12T/ QCCS-12T/ NRHC-14T/ HRRT-13T at 0.8V. In addition, the proposed memory cell demonstrated a higher relative figure of merit than existing memory cells.

Keywords: RF and Microwaves in Medicine and Biology, RF/Millimeter-wave Circuits and Systems, THz, mmWave and RF Systems for Communications
Abstract: This paper presents a passive 5.8GHz RF Energy Harvester (RFEH) implemented in 65nm CMOS technology that allows batteryless operation and reduction in overall size of WSN and IoT sensor nodes, which are limiting more widespread implementation. The RFEH consists of an on-chip matching network and optimized rectifier that are co-designed for maximum efficiency and sensitivity in the RFEH in gathering the required energy for the load circuit. The passive 5.8GHz RFEH prototype demonstrated a sensitivity of around -13dBm in gathering enough energy to produce 0.5V on a 10uF capacitive load, allowing the harvested energy to be reasonably combined with energy from other sources. When used as the only energy source, the ability to gather and provide a continuous supply of 8uW (10uA at 0.8V) is demonstrated, which is enough to power low-power RF circuits such as OOK and FSK receivers without the need of any battery, allowing batteryless operation.

Keywords: RF/Millimeter-wave Circuits and Systems, RF and Microwaves in Medicine and Biology, Microwave Metrology
Abstract: A dual-mode stepped-impedance resonator with noise elimination is presented in this paper. The dual-mode resonator is designed as a meander input/output coupling port and still has a compact circuit structure. This circuit structure can also have a good passband in the operating frequency and eliminate interference from the required band. The operating response of the resonator circuit at a central frequency is 2.1 GHz. The passband of the proposed resonator has a loss of insertion at 0.52 dB and a loss of reflection at 30 dB. The frequency range of 2.5 to 5.5 GHz is suppressed by increasing the folded input/output feed, resulting in a 20 dB rejection.

Keywords: RF/Millimeter-wave Circuits and Systems
Abstract: This paper presents a design of a hybrid multistage differential rectifier dedicated for indoor light energy harvesters. For the first stage, a fully cross- coupled rectifier with two additional PMOS switches is used and conventional fully cross-coupled rectifiers are cascaded for the succeeding six stages. The proposed rectifier is designed using TSMC 65nm CMOS process for a smaller chip area. With an input voltage of 0.5 V operating at 10 MHz, the circuit obtained an output voltage of 2.08 V and output current of 10.4 mA. It allows improving the output current to as much as twice the value compared to that of a conventional multistage differential rectifier structure. The maximum power conversion efficiency obtained is 43.08%. The total core chip area is 0.052 um². All the design underwent multiple thorough verifications and simulations using Synopsys Custom Design Tool.

Keywords: Control system modeling, Intelligent control, Neuro-control, Fuzzy control and their applications
Abstract: This paper proposes a model reference output feedback discrete-time terminal sliding mode (DT-TSMC) approach for SISO systems. The approach is based on a chatterfree equivalent control design that relies on a delay disturbance observer to handle exogenous disturbances. The stability of the proposed is shown via rigorous analysis and it is demonstrated that the term with the fractional power improves the steady state phase of the reference tracking and that an error of O(T^2) can be achieved. The paper concludes with a simulation example that shows a comparison of the proposed DT-TSMC with a classical discrete-time sliding mode control (DT-SMC) approach. The results show that for a similar transient response, DT-TSMC produces better steady state performance.

Keywords: Control system modeling, Space and underwater robots, Intelligent control, Neuro-control, Fuzzy control and their applications
Abstract: During satellite mission planning and operation, the main function of the satellite's attitude determination and control subsystem (ADCS) is to gather information about the satellite's orientation relative to the inertial reference frame. Additionally, this subsystem generates control actions that produce the required torques for adjusting the satellite's orientation, particularly in the context of the Low-Earth Orbit (LEO) regime. This paper focuses on the satellite three-axis attitude control problem for a de-tumbling mode of spacecraft using only magnetorquers as actuators under the presence of noise and investigates their performance through Hardware-in-the-Loop simulation (HiLs) tests, which consisted of a relative Earth's magnetic field generator along with the SGP-4-based satellite orbital propagator high-level control software. The design, development, and verification of proposed satellite attitude control system (ACS) strategies are presented. In detail, the B-dot algorithm is used for the de-tumbling mode to stabilize and reduce the angular rate, along with the pointing algorithm for orienting to the desired attitude. Then, a cascade PID is implemented to generate enough torque through the three-axis magnetorquers on the frictionless air-bearing platform to verify the performance of the controller using an onboard computer. Finally, the effectiveness of the co-simulation tested as the primary experiment was confirmed through the integrated simulation process.

Keywords: Intelligent control, Neuro-control, Fuzzy control and their applications, Control system modeling
Abstract: In this paper, an optimal fuzzy-PID controller is proposed for a knee exoskeleton model for patients with knee problems caused by strokes, post-polio, osteoarthritis, etc. For this purpose, a fuzzy control logic based proportional integral derivative (Fuzzy-PID) controller is considered. The fuzzy controller in the proposed control structure is used to adjust the PID controller parameters for the knee exoskeleton model. The error between the targeted angle and actual angle of knee exoskeleton system and the change of this error are taken as the input variables of the fuzzy controller. The output variables are chosen as the parameters of PID controller. The membership functions of the fuzzy controller are optimized with a well-known optimization algorithm called class topper optimization. The performance of the proposed controller is examined under different scenarios and plotted on a graph.

Keywords: Intelligent control, Neuro-control, Fuzzy control and their applications, Networked control systems, Instrumentation systems
Abstract: During the Taal Volcano eruption in January 2020, which inflicted damage in its 17 km radius danger zone as well as neighboring areas, many infrastructures were destroyed. Ash spread even to Manila, resulting in soil deformation, sulfur dioxide emissions, and other issues. Exposure to extremely high quantities of sulfur dioxide can be fatal, and its effects can harm the eyes, mucous membranes, and respiratory tract. In severe cases, it has proven to be deadly. With that in mind, the researchers proposed a system: a detection, monitoring, and early warning system for sulfur dioxide emissions from volcanic activity. The system consists of a sensor node and utilizes a cloud for storage and data display. The sensor node comprises a sensor module connected to a microcontroller and a Wi-Fi module for sending the data through the cloud.

Keywords: Intelligent control, Neuro-control, Fuzzy control and their applications
Abstract: A key challenge in automating vehicle driving maneuvers is to address uncertainty, mainly related to the interactions between the ego vehicle and all the surrounding traffic as well as the road infrastructure. In the case of an overtaking maneuver, the neighboring vehicles include the vehicle(s) moving in front of the ego vehicle - in particular, the vehicle(s) to be overtaken - and possibly other vehicles moving in the adjacent lane. In that case, the underlying research question is: how a driving automation system of an ego vehicle shall deal with occurring uncertainties to safely perform overtaking maneuvers, knowing that each of its neighboring vehicles is managerially and operationally independent, their disposition has been evolutionarily formed, and by their interactions, they jointly raise emergent behaviors. In fact, together, these vehicles compose a System-of-Systems (SoS). Recently, a novel formal language, called Fuzzy SosADL, has been specially conceived for modeling opportunistic SoS, while mastering uncertainty and interactions under uncertainty. This paper presents a case study of Fuzzy SosADL, in terms of Fuzzy Mediating Control Systems for safely supporting vehicle overtaking maneuvers in two-way roads, one of the hardest applications of driving automation systems.

Keywords: Image / Video / Multimedia Signal Processing, Pattern Recognition and Object Tracking
Abstract: Due to the advancement of photo editing techniques, it has become easier to create fake photos that look incredibly realistic and are edited in a way that leaves no visible signs of manipulation, making them ideal for synthesis. However, Instagram, WeChat, and TikTok are some of the popular social media platforms where the images have been lossy compressed before uploading them. As a result, learning to spot forged images in their compressed form is crucial. As part of this, some forensic detection techniques have made great strides in uncompressed scenarios, but there is still much to learn about the forensics of lossy compressed images. Therefore, this research proposes a hybrid deep learning framework by dissecting compressed and manipulated images at the preprocessing and feature extraction levels. The suggested noise stream progressively prunes the texture information to prevent the model from fitting the compression noise. Hence, a noise stream is employed to extract temporal correlation characteristics to address the potential problem of ignoring temporal consistency in lossy compressed images. Further, residuals from two streams are fed to custom ResNet blocks to enhance the clues of manipulation and pooled to concatenate the enhanced fingerprints. Finally, the proposed method outperforms state-of-the-art techniques in identifying manipulation in lossy compressed images.

Keywords: Image / Video / Multimedia Signal Processing, Pattern Recognition and Object Tracking
Abstract: Although generic object-detection methods have achieved a great deal, it needs more exploration for underwater images. Underwater object detection (UOD) is associated with challenges such as degraded image quality, low visibility, low contrast, colour shift, and limited computational capacity availability on the deployment environment. Moreover, most previous studies on deep learning-based underwater object detection have generally concentrated on increasing detection accuracy by utilizing huge networks. This work proposes a two-stage method where in the first stage, underwater images are enhanced based on PCA-fusion method. This step involves mutiple image enhancement steps such as color correction process, followed by the White Patch Retinex Algorithm for white balancing. On the colour-corrected image, we apply global histogram equalization, unsharp masking, and median smoothing separately to improve the contrast, sharpen the image, and reduce the white patch noise. Then, we create a single enhanced image by combining the results of the three methods using the principal component analysis (PCA) based fusion method. Finally, in the second stage, the enhanced image is passed as an input to the attention-fused lightweight single-stage object detection model for localization and classification. Experimental results show that the proposed method outperforms the state-of-the-art algorithms based on various image enhancement and object detection evaluation metrics on the URPC2019

Keywords: Image / Video / Multimedia Signal Processing
Abstract: The introduction of quad-tree with nested multi-type tree (QTMT) brings a significant reduction in bitrate to Versatile Video Coding (VVC) because QTMT enables a more flexible coding unit (CU) partition based on image content. However, the extensive rate distortion optimization (RDO) process induces a massive increase in encoding time. To mitigate this burden, a perceptual-based fast CU partition algorithm for VVC intra coding is proposed in this paper. The just noticeable difference model (JND) is adopted to simulate the human visual system, and JND variance is employed for early partition termination and division mode selection by reflecting the perceptual texture consistency. Experimental results on VTM17.0 show that the proposed method can achieve 31.13% time complexity reduction, with 1.32% Bjontegaard delta bit rate (BDBR) increase.

Keywords: Image / Video / Multimedia Signal Processing
Abstract: The image file format used in iris recognition systems has been proven to affect the performance of iris authentication. PNG and JPEG, the image file formats that are among the most common, will be compared in this study by converting a single iris image to the two image file formats, which will then be evaluated using total and average processing time FRR and FAR. The two file formats differ in their used image compression algorithms. JPEG achieves a lossy image compression, which results in irreversible data loss after every compression, using DCT (Discrete Cosine Transform). On the other hand, PNG achieves lossless image compression using the DEFLATE algorithm. Both algorithms' motive is to compress data. The latter, however, executes this without losing data. The image file formats are compared on a Raspberry Pi 4B that runs the iris recognition system utilizing Daugman's algorithm. After conducting 80 trials of iris recognition in the system, the JPEG database resulted in a total and average processing time of 70.8301s and 1.7708s respectively, and a FRR of 0.05 and a FAR of 0.05. Meanwhile, the PNG database yielded a total and average processing time of 72.2576s and 1.8064s respectively, and a FRR of 0.10 and a FAR of 0.05. This comparative study identified that the extracted iris features from JPEG produced better recognition performance results in the implemented iris recognition system over PNG.

Keywords: Image / Video / Multimedia Signal Processing
Abstract: Image compression is a topic of significant interest as it reduces file sizes in stored data. In this paper, we propose a model that achieves multiple levels of compression, thereby minimizing the storage space required for images, which typically consume substantial amounts of data due to their size and resolution. We combine an image downscaling and upscaling model with an image compression model. By leveraging convolutional techniques to identify image features, we can effectively reduce the size of the image through downscaling and subsequently upscaling it. Additionally, we employ entropy image compression and arithmetic encoding to compress and reconstruct the image while preserving its lossless data. Through experimentation with the Kodak dataset, we observed that our proposed model achieved a compression rate of 96.92%, significantly reducing the data needed for file storage. Moreover, our reconstructed images attained a standardized measure with a signal-to-noise ratio of 33.10 dB and a structural similarity of 0.9219. Notably, the perceptual quality of the images, including intricate details, remained intact to the human eye.

Keywords: Biomedical Imaging and Video analytics, Computational Systems, Modeling and Simulation in Medicine, Biomedical Signal Processing and Instrumentation
Abstract: This study uses ultrasound videos and optical flow analysis to estimate heart failure automatically. Feature vectors of left ventricular motion are employed in a Support Vector Machine to classify heart condition based on ejection fraction. The system achieved 70% accuracy, demonstrating the importance of cardiac motion direction in diagnosis

Keywords: Image / Video / Multimedia Signal Processing, Pattern Recognition and Object Tracking, Signal Processing Algorithms and Architectures
Abstract: Every day, there is an upsurge in the number of terrorist attacks carried out by drones. As a result, drone detection has become mandatory. Real-time detection of drones is a very challenging task due to their small size, lightning conditions, and relative viewing angles. In this article, a new UAV dataset is presented to perform drone detection tasks using two deep learning techniques, YOLOv5 and YOLOv8, along with the existing Det-Fly dataset. Implementing the YOLOv5 technique, the mean average precision (mAP) for drone detection on both the Det-Fly and UAV datasets is 97.2% and 94.1%, respectively. Similarly, the corresponding values for the YOLOv8 algorithm are 99.5% and 95.0%, respectively.

Keywords: Neural Networks and Deep Learning
Abstract: Checking the large number of x-ray results can cause the doctors to be exhausted. Moreover, those fatigued may induce misdiagnosis. In this research, Artificial intelligence will be used for creating applications that will support the doctors in Spondylolisthesis Classification.

Keywords: Wireless Communications and Networks, Optical Communications and Networks
Abstract: Visible light positioning(VLP) has gained prominence as a highly accurate indoor positioning technique. Few techniques consider the practical limitations of implementing VLP systems for indoor positioning. These limitations range from having a single LED in the field of view(FoV) of the image sensor to not having enough images for training deep learning techniques. Practical implementation of indoor positioning techniques needs to leverage the ubiquity of smartphones, which is the case with VLP using complementary metal oxide semiconductor(CMOS) sensors. Images for VLP can be gathered only after the lights in question have been installed making it a cumbersome process. These limitations are addressed in the proposed technique, which uses simulated data of a single LED to train machine learning models and test them on actual images captured from a similar experimental setup. Such testing produced mean three dimensional(3D) positioning error of 2.88 centimeters while training with real images achieves accuracy of less than one centimeter compared to 6.26 centimeters of the closest competitor.

Keywords: Control system modeling, Intelligent control, Neuro-control, Fuzzy control and their applications, Mobile robotics
Abstract: This study developed a resonance model for safety analysis of the spiral lift on autonomous driving platform. Because of difficulties in modeling a spiral shape, an unlocking model was applied to design the spiral lift. The unlocking model was based on the shear force and was useful in predicting the response of the spiral spring. The safety usage of the spiral lift was investigated in consideration of vibration from passengers.

Keywords: Instrumentation systems, Mobile robotics, Control system modeling
Abstract: In this study, a conceptual measurement model for a targeting global coordinate from an original points was investigated using a laser sensor and AHRS. The target coordinate was estimated from the current global coordinate, distance and Euler angle. The global model is generally nonlinear, thus equal search was applied for the estimation. The method will be convenient to obtain positions of buildings, stadiums and land.

Keywords: Machine Learning, Neural Networks and Deep Learning, Data Mining
Abstract: Knowledge Graphs play an important role in various AI applications, such as question-answering systems. However, most Knowledge Graph construction approaches do not consider the temporal aspect. Temporal information is crucial in various real-world scenarios, where relationships between entities and facts evolve over time. In this paper, we therefore propose an end-to-end system for temporal knowledge graph construction.

Keywords: Renewable Energy Sources and Technology, Energy Storage System, Power Generation, Transmission and Distribution
Abstract: With the increasing demand for convenience life, various types of wearable devices that use information and communication technology (ICT) have been launched. Furthermore, wearable energy harvesting technology is used to facilitate the semi-permanent use of such devices. In this study, a self-powered system is designed using a solar panel to facilitate the semi-permanent use of a beacon. As a result, 1-hour self-powered charging produced sufficient power for operating the device for more than 24 h.

Keywords: Control system modeling, Mobile robotics
Abstract: GNSS (Global Navigation Satellite System) is currently used in a variety of applications, and its positioning accuracy requirements become more and more demanding. We focus on an automatic strawberry pollination robot (under development in Kobe City College of Technology) and study a method of successive moving baseline vector estimation. In this paper, we propose a method to estimate accurate successive moving baseline vectors based on the difference of the single GNSS receiver observables obtained at two successive observation time (epoch). Furthermore, we also propose a method to estimate the bias error included in the solution of the moving baseline vector. Throughout the experiment, the proposed method and the relative positioning method are compared, and the results show that the proposed method can provide the baseline vector with the accuracy of 1.06 cm.

Keywords: Instrumentation systems, Industrial automation
Abstract: Image processing was applied for the real-time evaluation of ultrasound imaging quality. Simultaneous observation of the ultrasound probe’s contacting force, tilting orientation, and operational trajectory in spatial coordinates can provide useful feedback for supporting ultrasonography training. This paper presents a 6-DOF passive robotic arm enabling the complete tracking of the probe position and orientation. The custom designed handle located at the end effector is assembled from the inner and outer shells, which allows installation of a tiny three-axis force sensor for measuring the probe contact force. The forward kinematics of the robot arm is derived for mapping the joint variables to the position and orientation of the tip. The real-time measurement of the joint angles is achieved from the non-contact magnetic encoders for the first to fourth revolute joints and the rotary potentiometers for the 2-DOF ball joint. The three-dimensional testbed consisting of the radial arrays of hexagonal slots is used for evaluating the position and orientation accuracy of the passive arm. The results of this study can be used as a guideline for further development of passive robotic arms to achieve a higher level of accuracy on tracking the probe trajectory.

Keywords: Humanoid robots, Control system modeling, Mobile robotics
Abstract: Legged robots have high mobility for application with uneven terrain. Bipedal robots require the lower number of actuators with less complication of control system, but their walking stability is the main challenge. This paper presents a design of the bipedal robot having two actuated joints for the hip and the knee and one passive joint for the ankle of each leg. To minimize the inertia of the moving legs, the four brushless motors with integrated controller and planetary gearbox are concentrically located at the hip axis while the knee joints are driven through the parallel linkages. The zero-moment point (ZMP) along the foot support contacting the floor was derived based on the table-cart model. During the stance phase of stable walking, the desired ZMP must be located within the foot boundary. The ZMP generator for shifting the center of mass (COM) forward according to the stride was simulated along with kinematics of the legs. The hip and the knee joint trajectories were implemented on the bipedal robot prototype. The desired leg kinematics was validated by the experiment. As the load supported by the feet, the effect of gravity on the leg joint positions and torques was also studied. Additional sensing and the control of joint stiffnesses will be applied for achieving the dynamic walking of the bipedal robot.

Keywords: Humanoid robots, Control system modeling, Space and underwater robots
Abstract: Soft pneumatic actuators providing infinite passive degrees of freedom are widely applied for safe physical human–robot interaction and fulfilling the limitation of conventional rigid structures. This paper presents a design of the soft actuator with pneumatic chamber enabling the control of bending and contact force. The actuator is solely fabricated from the thermoplastic polyurethane (TPU 95A) by using the fused deposition modeling (FDM) 3D printing. Several printing setups were performed for trial and error to prevent the air leak. The experimental setup was prepared for measuring the contact force in both vertical and horizontal directions. The grasping force varying against the input pressure ranged from 0 to 58 psi was studied at three contacting points along the longitudinal axis of the actuator with three different bending conditions. The stiffness of the actuator is related to the input pressure by considering the increase of total contacting force at different bending deformations. The results of this study can be applied to the stiffness control of soft pneumatic actuators.

Keywords: Biomedical Imaging and Video analytics, Biomedical Signal Processing and Instrumentation
Abstract: Brain-computer interface (BCI) systems rely on capturing characteristics of human brain activity from the electroencephalography (EEG) signals, especially for the reliable classification of motor imagery tasks. For multi-channel EEG signals, it is crucial to precisely capture the spatio-temporal variation along with the frequency characteristics. Hence, instead of directly operating on raw EEG data, in this paper, discrete wavelet transform (DWT) is first applied to the motor-imagery multi-channel EEG data and then a deep learning architecture is designed incorporating spatial-temporal operations, which operates on the DWT-transformed EEG signal. In the proposed architecture, temporal convolution followed by spatial convolu- tion is performed on the DWT-operated MI-EEG signal, and this part is termed as SAT-net. Next, by considering all channels together convolutional operation is performed to reduce the number of channels and this part is termed as SOC-net. Finally, a fully connected layer is used to classify the MI-EEG data from the derived feature vector. Extensive experimentation is performed on multiple subjects taken from the MI-based EEG dataset BCI Competition IV 2a. It is found that the proposed model offers a classification accuracy of 84.65%, consistently providing better classification performance than that obtained by some state-of- the-art methods.

Keywords: Biomedical Signal Processing and Instrumentation, Bioinformatics
Abstract: Feature extraction is an essential process for removing the unwanted part and interference of the Electromyography (EMG) signal, and to extract the useful information hidden in it. Inorder to obtain high performance of Myoelectric Control (MEC), the choice of features plays an important role. The studies carried out earlier to overcome force level variation have used features which are redundant, affecting the robustness and the classification performance. This study's main objective is to assess a database's performance consisting of nine upper limb amputee subjects with EMG data recorded at three different force levels when six motions were classified using twenty different time domain features that are frequently found in the literature. Training is carried out at one force level, and the other two unknown force levels are used for testing. Out of the twenty features, the one that is the most stable is displayed for each force level. The results show that root mean square (RMS) feature outperformed other features for training at low and medium force levels, and Wilson amplitude (WAMP) feature for training at a high force level, when compared with the most widely used linear discriminant analysis (LDA) classifier. The average classification accuracy for the nine amputee subjects trained with the RMS feature at low and medium force levels was 42% and 51.78% percent, respectively. For high force level, when trained using WAMP feature, an accuracy of 46.78% is obtained.

Keywords: Biomedical Signal Processing and Instrumentation, Biomedical and Health Informatics, Bioinformatics
Abstract: Vasovagal syncope (VVS) is the commonest cause of short-term loss of consciousness, which negatively impacts quality of life. To gather diagnostic information, medical professions often perform a head-up tilt test (HUTT) during direct observation. During this test, subjects may experience common symptoms such as nausea, pallor, sweating, palpitations, near faint and syncope. The purpose of the study was to develop an algorithm that uses electrocardiography (ECG) and blood pressure (BP) recordings from HUTT to predict VVS before its onset. In this study, the calculated cumulative risk based on the analysis of the three specific sets of features was compared to a pre-established VVS risk threshold. The purpose of this comparison was to determine if the cumulative risk was above or below the threshold and whether an alert should be generated. An alert would only be triggered when the cumulative risk exceeded the threshold. The prediction time was defined as the duration between the first alert and the actual syncope episode. A total of 137 subjects were recruited in this study. Our proposed model accurately predicted syncope onset in 87 out of 120 subjects. The model’s sensitivity was 81.6% while its specificity was 66.2%. The precision was determined to be 62.5%, the F1 score was 70.8%. Additionally, the model was able to predict syncope before its onset with a median prediction time of 221.45 seconds (Interquartile range: 180.0 - 294.0 s).

Keywords: Biomedical Signal Processing and Instrumentation, Biomedical and Health Informatics
Abstract: The current advancements in machine learning research pertaining to speech and health are highly interesting. One aspect of speech-processing research that is gaining popularity is the use of computational paralinguistic analysis to evaluate a variety of health conditions. In this study, we have used the Hilbert-Huang transform (HHT) for the time-frequency analysis of speech signals for the identification of the common cold. The HHT is a time-frequency transform that is adaptive and ideal for non-linear and non-stationary signals. The HHT is the combination of empirical mode decomposition (EMD) and the Hilbert transform (HT). The HHT gives the time-frequency representation (TFR) matrix of the speech signal. Then, the entropy of each frequency component in TFR is computed and used as a distinguishing feature between cold and healthy speech. The efficacy of the proposed methodology is evaluated on the URTIC dataset utilizing a deep neural network. The results of the study demonstrate that the time-frequency entropy features extracted using the HHT are effective in distinguishing between cold and healthy speech.

Keywords: Instrumentation systems, Industrial automation
Abstract: This study aimed to improve the production process of canned pineapple by introducing an algorithm that utilizes machine vision and image processing techniques to detect and locate pineapple eyes accurately after the peeling process. The result can be subsequently transmitted the coordinates to the automatic pineapple eyes removal machine. The algorithm's primary objective is to improve productivity, especially by reducing waste due to human error. Therefore, the prototype has been made. Including the image acquisition process and the rotating system to capture pineapple surface features, the system was operated by a stepper motor along with an inspection camera for data collection. The algorithm was created based on thresholding method, with the calibration, pineapple eyes can be located in term of position(x) and angle (𝜑) of rotation axis. In summary, from our design and experiment. Our algorithm results in a maximum error of position (x) at 4.871 mm and a maximum error of angle (𝜑) at 3.415 degree at the processing time of 0.68 millisecond, with the entire process taking 20.92 seconds.

Keywords: Mobile robotics
Abstract: This paper describes a simple target detection system for autonomous vehicles to perform an emergency stop. We especially focus on the system only with a digital camera as the perception unit for the situation. The detecting performance of the imaging sensors, including the digital cameras, often suffer from “haze” due to atmospheric obscuration. In this paper, we review the image “dehazing” algorithms aiming for the in-vehicle usage, and we propose a system configuration for the safety measure with applying a selected algorithm. A series of experimental results depicted the feasibility of the system configuration and its applicability to real-time processing.

Keywords: Space and underwater robots, Mobile robotics, Control system modeling
Abstract: This paper concerns the development and improvement of the underwater navigation system for Autonomous Underwater Vehicles (AUVs) when faced with velocity-aiding sensor failures. The study addresses this challenge of sensor fusion by applying the Error-State Kalman Filter (ESKF), a form of indirect state filtering. Specifically, the ESKF targets the limitations of velocity measurements encountered during near-bottom operations. The proposed method was applied to the Xplorer-Mini AUV and evaluated using the Gazebo physics engine simulator in ROS 2. To assess its performance in handling the loss of velocity signals, a series of simulation-based experiments were conducted and compared against the traditional Inertial Navigation System (INS) and Extended Kalman Filter (EKF) algorithms. The results of the experiments demonstrate that the ESKF outperforms traditional INS and EKF algorithms, offering valuable insights into state estimation techniques for developing autonomous underwater vehicles in challenging environments.

Keywords: Biomedical Signal Processing and Instrumentation, Computational Systems, Modeling and Simulation in Medicine, Biomedical and Health Informatics
Abstract: This study investigates the discrimination between cold speech and healthy speech using features based on empirical mode decomposition (EMD). The EMD is employed to break down the signal into several intrinsic mode functions (IMFs). From each IMF, various statistical values like minimum, maximum, mean, standard deviation, first, second, and third quartiles, skewness, kurtosis, and energy of each IMF are extracted and used as a feature for distinguishing cold and healthy speech. The T-test examines the importance of EMD-based features for classifying cold speech. EMD-based feature performance is assessed using the deep neural network (DNN) classifier. The findings show that EMD-based features effectively discriminate between cold and healthy speech classes. Combining Mel-Frequency Cepstral Coefficients (MFCC) characteristics with EMD-based features improves the performance for identifying healthy and cold speech classes. On the URTIC database, the combination of MFCC and EMD-based features achieve a UAR of 66.92%.

Keywords: Biomedical Signal Processing and Instrumentation
Abstract: In frequency domain, wavelet coefficient is one of the characteristics of ERG signal which is hidden in frequencydomain. This paper aims to develop a tool for analyzing the constituent frequencies and time of interest in the dog’s ERG signal using wavelet analysis. We design a lowpass filter to reduce noise during the measurement of ERG response. ERG dataset was obtained for the light intensity series and their constituent frequencies are analyzed by using the continuous wavelet transform (CWT). The results show that, in timedomain, as light intensity increases, both amplitudes of a-wave and b-wave are higher. We can visualize the noise interference in the ERG signals. Then a lowpass filter is designed to attenuate the noise effect. The filtered signals have lower a-, and b-waves’ amplitude which affect the important characteristics in time domain. Therefore, we choose to filter signal from 26ms onward, the main components are recovered, and the filtered signal becomes smoother. In frequency-domain, the peak magnitude of wavelet coefficients is increased as light intensity increases.

Keywords: Translational Engineering in Healthcare, Biomedical and Health Informatics, Wearable Sensors for Health care monitoring
Abstract: Cardiovascular diseases (CVDs) are a key global health concern, accounting for a significant proportion of deaths worldwide. Early detection and precise diagnosis are required for effective CVD treatment and control. One non-invasive method that has shown promise for detecting CVDs through hypertension is the analysis of photoplethysmography(PPG) signals. It measures the difference in the amount of blood and oxygen saturation in the underlying venule. PPG signals are non-invasive, cost-effective and easy to acquire, making them an ideal candidate for hypertension screening. This study proposes intelligent machine learning models to detect hypertension with the PPG signal’s frequency-domain and time-domain features. The performance of the classification models is evaluated using various performance metrics such as accuracy, specificity, recall and F1-score. The results show that PPG signals can be used to accurately detect hypertension and Possibly facilitate the early diagnosis of cardiovascular diseases. We saw that random forest classification had the highest accuracy of 93% for detecting hypertension.

Keywords: anomalous detection in digital signal and network, pattern recognition and data analysis, abnormal event localization
Abstract: To build a reliable system, anomaly detection is the principal task for ensuring the system's security. However, the complexity of systems and software has increased over time. As a result, the likelihood of system failures and vulnerabilities has also grown. For this reason, employing manual anomaly detection approaches is impractical. This work proposes the use of a Convolutional Neural Network for log-based anomaly detection and enhances a log parsing method through parameter entity labeling. We have chosen the ThunderBird and BlueGene/L datasets for our experiments, employing a down-sampling technique to address data imbalance issues and reduce model training time. The results show that when comparing the detection outcomes of models trained with the down-sampled training dataset and models trained with the full training dataset (without using down-sampling), the models trained with the full training dataset exhibit higher recall, while their precision and specificity remain comparable. Additionally, the results indicate that our approach demonstrates slightly better detection performance than the previous log parsing method. Precision, recall, and specificity reach 0.9999, 0.9933, and 0.9914, respectively, when experimenting with the ThunderBird dataset.

Keywords: image and video anomaly detection
Abstract: Stereo Imaging technology integration into medical diagnostics and surgeries brings a great revolution in the field of medical sciences. Now, surgeons and physicians have better insight into the anatomy of patients' organs. Like other technologies, stereo cameras have limitations, e.g., low resolution (LR) and blurry output images. Currently, most of the proposed techniques for super-resolution focus on developing complex blocks and complicated loss functions, which cause high system complexity. We proposed a combined channel and spatial attention block to extract features incorporated with a specific but very strong parallax attention module (PAM) for endoscopic image super-resolution. The proposed model is trained using the da Vinci dataset on scales 2 and 4. Our proposed model has improved PSNR up to 2.12 dB for scale 2 and 1.29 dB for scale 4, while SSIM is improved by 0.03 for scale 2 and 0.0008 for scale 4. By incorporating this method, diagnosis and treatment for endoscopic images can be more accurate and effective.

Keywords: image and video anomaly detection, pattern recognition and data analysis, abnormal event localization
Abstract: Golf is widely recognized as one of the most popular sports globally. However, one drawback of playing golf is the relatively high cost of equipment and coaching. While numerous training programs are available to assist players in their practice, there is currently no swing analysis program developed by Thai professionals. In this project, advanced deep learning models were employed: SwingNet, capable of predicting the sequence of eight golf swing events in videos and determining the confidence level of each swing, and MoveNet, designed to identify joint positions on the body and represent them as skeletons. These models were integrated into a customized template-matching algorithm that utilized angle-based measurements to analyze the sequence of golf swings. This analysis assessed the similarity score, represented as a percentage, between two individuals for each golf swing event. Furthermore, various techniques were implemented to enhance the efficiency of SwingNet. Through performance evaluation, it was observed that the efficiency of SwingNet surpassed by one percent compared to the pre-trained model.

Keywords: pattern recognition and data analysis, anomalous detection in digital signal and network, human-computer interaction
Abstract: Toilet quality assurance is a crucial process in ensuring lavatories meet rigorous performance, durability, and hygiene standards. The current standard of Maximum Performance (MaP) testing faces challenges, leading researchers to explore innovative approaches such as sound source classification for quality assurance. This approach involves collecting a diverse dataset of lavatory sounds and extracting relevant acoustic features. Deep learning models, particularly convolutional neural networks (CNNs), are trained on these features to accurately classify sound sources. The trained models were evaluated and compared by considering metrics such as classification accuracy, computational complexity, and model parameters. This paper performs these tests and chooses the most effective model to enhance the quality assurance process for toilets. Incorporating sound source classification techniques has several benefits, including the optimization of testing processes, non-intrusive performance assessment, and efficient resource utilization through targeted testing and troubleshooting. By improving the standards of lavatory quality, this approach ensures enhanced performance, durability, and hygiene of lavatories.

Keywords: pattern recognition and data analysis
Abstract: Intelligent Transportation Systems (ITS) offer a means to increase efficiency in road management, safety, and traffic enforcement. The Philippines, particularly Metro Manila, is notorious for its high levels of traffic congestion result in significant economic loss. It is possible to accumulate large amounts of traffic video data by installing traffic cameras but the manual preparation of such custom datasets for ITS applications is taxing and laborious. In this paper, we develop an algorithm for vehicle detection, classification, and flow rate measurement algorithm for ITS applications using YOLOv7 and a tracker trained on a custom dataset provided by the UP National Center for Transportation Studies (NCTS) augmented with active learning algorithm. Its detection, classification, and tracking performance were compared to that of a model trained without using active learning. The results indicate that using uncertainty-based active learning algorithm is effective in improving the model's tracking capability. The best results from the active learning models with the tracker was able to achieve a higher HOTA value of 67.423 vs 67.355 (+0.068%) for the first evaluation, and 71.652 vs 70.614 (1.038%) for the second evaluation on the NCTS tracking evaluation sets. For a specific sequence in DETRAC, the improvement is 69.842 vs 69.84 (+0.002%). At the third cycle of training, the active learning model counts better with a total count of 98 vs 100 from a true count of 91.

Keywords: pattern recognition and data analysis, abnormal event localization, anomalous detection in digital signal and network
Abstract: Abstract—This study aimed to assess the generalization performance of a Metal Health Monitoring system, which is crucial for practical applications. Previous research has not thoroughly examined this aspect of performance. To enhance the system's performance, we conducted experiments using 90 metal pieces, anticipating improved results with increased sample size. The pieces were divided into nine classes, representing undamaged and damaged conditions at eight different positions. Vibration waveforms were obtained by attaching piezoelectric sensors to the pieces. The waveforms were then split into training and evaluation datasets, and a neural network (NN) was trained on the former to classify the latter. The findings revealed that the NN achieved a remarkable accuracy of up to 80.6% in classifying the damage positions, even for metal pieces not included in the training set. These results indicate a high level of generalization performance in the Metal Health Monitoring system.

Keywords: Biomedical and Health Informatics, Biomedical Signal Processing and Instrumentation
Abstract: The ability to differentiate sensor-induced physiological signals between healthy and diseased subjects is useful for developing an e-health system. Patients with neurodegenerative disorders are among those who can benefit from the use of e-health. Entropy methods have been utilized to quantify the complexity of such physiological signals for pattern classification. To date, these methods have been applied individually. In this retrospective study, several entropy methods are examined and used as feature extraction methods for machine learning to classify gait patterns in neurodegenerative diseases. Experimental results show that the combination of entropy methods and standard statistical measures performed much better than the individual measures for physiological pattern differentiation. Several machine learning models were also evaluated for learning on these features. This study also found that the one-dimensional convolutional network model trained with the combined features provided the most favorable results, where the best entropy measures depend on certain values for time delays and embedding dimensions.

Keywords: Biomedical and Health Informatics, Wearable Sensors for Health care monitoring, Biomedical Imaging and Video analytics
Abstract: The Exoskeleton leg is built to enable amputee patients to walk, ascend stairs, sit, and do fundamental movements. For any basic movement, patients can control the robotic Exo-leg by the movements of their residual limbs concerning the position of lower limb amputation. Prosthetic Exo-leg has been constructed to support patients with transfemoral amputation cost-efficiently. Moreover, the machine consists of multi-functional features under prosthetic exoskeleton legs that will initially support the patients of lower limb amputation, more specifically transfemoral amputation. Enabling the E-leg to a wide range of users and a variety of applications requires the socket to be adaptable through kinematics and actuation. The performance test shows the change of major movement in standing and walking that is expected to improve through further patient training and rehabilitation.

Keywords: Biomedical Imaging and Video analytics, Computational Systems, Modeling and Simulation in Medicine
Abstract: Bridging exercise is a widely applied training for stroke rehabilitation to improve balancing ability on weight-bearing activities. Aiming to reduce the workload of physical therapists and enable the systematic recording of motion data, this paper presents an affordable rehabilitation monitoring system using an RGB camera. For predicting the correctness of the bridge posture, the MediaPipe framework is applied for detecting the human body segments which are used as the input data of the decision tree classifier instead of using a complex neural network. The model was trained using the data collected from five healthy participants performing the correct and Wide Knee postures when the knees are separated laterally. The experimental results show that nearly 100 percent accuracy can be achieved in confirming the correct posture and identifying the Wide Knee posture. The time performance of the decision tree classifier trained by the different number of frames is also evaluated. This system is very promising to help therapists monitor patients and provide feedback for improving the effectiveness of the rehabilitation.

Keywords: Biomedical Signal Processing and Instrumentation, Bioinformatics
Abstract: Surface Electromyography has gained tremendous significance in the recent years due to its suitability and reliability in a wide range of applications like automatic prosthetic control, diagnosis of neuromuscular disorders, in robotics and many such fields. Considering such applications, identification of various muscular movements is necessary and hence, EMG pattern recognition is needed. This paper focusses on a generalised EMG pattern recognition of various hand movements. The data from Ninapro Database - 4 has been used for pattern recognition. The database has Surface Electromyogram (sEMG) data of 52 various hand movements. The data was subjected to pre-processing, feature extraction and classification. An average accuracy of 64.87% was obtained for a combination of seven features in the time (temporal) domain, using Linear Discriminant Analysis (LDA) as the classification model. The obtained classification accuracies are compared and discussed with respect to the state-of-the-art literature.

Keywords: Biomedical Signal Processing and Instrumentation, Biomedical and Health Informatics
Abstract: College students carry their bag to attend class every day, mainly using backpack or sling bag. Heavy load carriage may affect body posture, and prolonged load carriage may cause permanent postural change and gait alteration. This study aims to identify the effect of different load carriage styles on the walking gait. Twelve female college students performed normal walking while carrying 5%, 10% and 15% of load in backpack and sling bag. Paired sample t-test with an alpha level of 0.05 were used to compare the normalized data. The result revealed that load carriage will significantly alter the trunk posture. Backpack carriage will significantly increase the truck inclination angle while sling bag carriage significantly increase the trunk lateral bend angle. Load carriage also caused significant changes in pelvis orientation, sling bag load carriage significantly reduce the pelvis tilt and obliquity, while backpack significantly reduce the pelvis obliquity, even with only 5% of body weight load carriage. Hence, prolonged load carriage shall be avoided to prevent postural alteration.

Keywords: Wearable Sensors for Health care monitoring, Biomedical Signal Processing and Instrumentation
Abstract: Inertial Measurement Unit (IMU) has been widely recognized to be the practical alternative to capture and analyze human gait. However, due to its inherent characteristics, it can only measure the basic kinematics of the body segment it attached to. With the help of the machine learning, IMU can be used to determine the dynamic behavior of the major lower extremity muscle. This paper explores the use of feature-extracted IMU data and a neural network to estimate muscle activity during walking. IMU and Electromyogram (EMG) data were collected from fifty-eight healthy participants. Principal Component Analysis (PCA) and Tsfresh (Time Series FeatuRe Extraction on basis of Scalable Hypothesis tests) were applied to extract the relevant features from the data. These features were then used to train the Feedforward Neural Network (FNN). A combination of Tsfresh and FNN yielded the best results with correlation coefficient (r) of 95.73% and Root Mean Square Error (RMSE) of 11.20%. This research can potentially help reduce the number of sensors needed in gait analysis, allow for portable motion capture, and improve the accuracy and efficiency of the FNN model in estimating muscle activity.

Keywords: Pattern Recognition and Object Tracking, Image / Video / Multimedia Signal Processing, Digital & Multirate Signal Processing
Abstract: In Modern times, automated security for identifying a person is one of the main concerns. There is a significant need for a trustworthy and secure identity verification solution. A reliable way to identify someone can be using a biometric identification system. The finger knuckle pattern offers excellent discriminatory features for biometric identification with indirect touch, including the advantages of long-range visibility. Existing models are failing to handle the depth information in finger knuckles that are highly relevant to understand the identification patterns. Therefore, we elaborate on the significance of utilizing the middle finger knuckle for biometric identification. We propose an ensemble approach that appropriately captures the rich features to identify a person based on their finger knuckle. The proposed model performance is evaluated on a standard dataset (HKPolyU 3D photometric stereo knuckle image dataset). Experimental results illustrate that the proposed model outperforms the existing results. Further, this approach would be advantageous in forensic investigations, security, and surveillance.

Keywords: Pattern Recognition and Object Tracking, Image / Video / Multimedia Signal Processing, Signal Processing Algorithms and Architectures
Abstract: Archery is a precision sport that requires high consistency and accuracy. The score's consistencies are highly correlated with respect to the archer's shooting pattern. With the current advancement of technology, archery scoring remains manual using scope observation in the field. By leveraging computer vision, recent efforts are made to automate the scoring process with the use of cameras. Previously studied computer vision approaches primarily explored processes such as homography estimation, image subtraction, and circle detection. However, these approaches do not accommodate the real target board faces, highly skewed camera angle, and estimated circles that may deviate from actual target rings. In this paper, a two-stage automatic archery scoreboard scoring system is proposed to detect arrows using camera. Firstly, the initiation stage extracts the scoring areas of the target ring by incorporating curve interpolation. Subsequently, the operation stage detects and localizes an arrow in consecutive frames. The proposed method demonstrates accurate arrow extraction from the target ring with less susceptibility to noise. It is able to operate consistently in challenging conditions due to an isolated detection stage.

Keywords: Pattern Recognition and Object Tracking, Image / Video / Multimedia Signal Processing
Abstract: Virtual reality technology has the potential to revolutionize immersive experiences in various applications, including office settings. However, efficient text entry in VR remains a significant challenge. This study addresses this challenge by proposing a machine learning-based solution, the 2S-LSTM typing method, to enhance text entry performance in VR. The 2S-LSTM leverages the back of the hand image. It employs a two-stream residual Long Short-Term Memory (LSTM) network, combined with a Kalman Filter (KF), to improve hand position tracking accuracy and reduce jitter. The results from questionnaire-based evaluations and typing data analysis demonstrate the superiority of the 2S-LSTM solution over existing solutions like Oculus Quest 2 and Leap Motion in terms of typing efficiency, fatigue reduction, accurate hand position replication, and positive user experience. These findings contribute to the advancement of text entry in VR environments and pave the way for immersive work experiences in the office and beyond.

Keywords: Pattern Recognition and Object Tracking, Image / Video / Multimedia Signal Processing
Abstract: Multispectral sensors are used to ensure visibility in various applications. However, when multiple sensors are used for capturing images, a misalignment may occur between the images taken by each sensor unless special care is taken. To correct such misalignments, image registration based on feature matching is conducted. However, the features captured by each sensor differ, thereby complicating the registration process. In this study, we develop an approach to overcome these challenges and to improve the registration accuracy between short-wave infrared and mid-wave infrared (SWIR and MWIR, respectively) images. First, we compare and validate SiLK, a detector-based feature matching method, and LoFTR, a detector-free feature matching method. The results clearly demonstrate the superior accuracy of LoFTR. Moreover, SWIR and MWIR images exhibit a characteristic color inversion according to Kirchhoff s law. Therefore, by inverting the color of a SWIR image and aligning the color tone between image pairs, we can improve the matching accuracy. Furthermore, by diversifying the color tones of the training data through augmentation, we can handle the domain gap between SWIR and MWIR images, thereby further enhancing the matching accuracy.

Keywords: Signal Processing Algorithms and Architectures, Digital & Multirate Signal Processing
Abstract: Variable digital filters are reconfigurable signal processing systems that possess tunable frequency responses. Therefore, a variable digital filter can carry out real-time online tuning during filtering operations. To produce a new response in a timely manner, the transfer-function's coefficients of a variable filter must be the functions of the parameter that is adopted for tuning the filter's response. In this way, the coefficients are changeable. This paper considers designing a variable bandpass filter with a full-band tunable center-frequency (CF). Since the CF can be changed by using the CF parameter, one can represent the filter coefficients as the functions of the CF parameter. Another stiff problem that needs to be carefully solved is the stability issue. The reason is that updating the filter coefficients during online tuning may incur instability. This paper deals with this stability issue by locating the filter's coefficients inside the stability triangle through parameter transformations. To this end, we transform the denominator coefficients into other unconstrained parameters that can take any values without causing instability. After the transformations, a nonlinear optimizer can be used for optimizing the new parameters. A set of fixed-coefficient bandpass filters with full-band tunable CFs are designed for illustrating the achieved stability guarantee together with the achieved considerably accurate approximations.

Keywords: Signal Processing Algorithms and Architectures
Abstract: Hammerstein spline adaptive filtering (HSAF) is presented based on normalised version of least mean fourth (LMF) algorithm. HSAF comprises of a nonlinear memoryless adaptive Lookup Table with the spline interpolation function and linear adaptive filter modified by gradient-based scheme. LMF is determined by properties of the error term of the fourth power mean. Normalised version of LMF (NLMF) is applied on HSAF to get the fast convergence. Experimental results show that can provide the competitive results to the traditional least mean square algorithm.