Bulletin of Electrical Engineering and Informatics
Not a member yet
2885 research outputs found
Sort by
Intrusion detection system in lightweight devices: issues and challenges
Full text linkIntrusion detection system (IDS) is a crucial component in ensuring the security of computer networks. It helps in identifying and responding to unauthorized access attempts or malicious activities within a network. The focus of this systematic review is on IDS specifically designed for lightweight devices. This systematic review aims to provide an abstract understanding of the current state of IDSs for lightweight devices. It involves a comprehensive analysis of existing research papers, evaluating the methodologies, techniques, and performance metrics used in these IDS solutions. The goal of the systematic review is to provide a critical assessment and analysis of the literature on IDS in lightweight devices, closing the research gap in this field. The review analyzed and evaluated 55 studies out of 678 initially identified. The findings of the study are presented in the paper, which includes insights into the state-of-the-art proposals in the field, challenges and limitations of existing solutions, and recommendations for future research directions. The outcome of this paper can help the advancement of IDS for lightweight devices
Electromagnetic interference risk from electrostatic discharge in infant incubators
Full text linkThis paper proposes an improved electromagnetic compatibility (EMC) risk analysis approach for medical equipment related to the effect of electrostatic discharge (ESD). This approach not only focuses on the risk of ESD from the susceptibility aspect but also investigates its conducted electromagnetic interference (EMI) characteristics. This study combines the standardized ESD test and conducted emission (CE) measurement simultaneously, applying it to the infant incubator and analyzing the spectrum of ESD current in the phase line in the time and frequency domain. The result shows that an ESD exposure caused current spikes with an average level of 13.8 A. Moreover, it also causes a broad spectral CE noise on the phase line of the infant incubator. Furthermore, the CE noise in the low-frequency range was also detected on the phase line during ESD exposure, indicating the risk of interference with other sensitive medical equipment connected to the same power network. The approach of proposed risk analysis in this study can be used to identify the risks of EMI due to ESD events in implementing the latest IEC 60601-1-2
Machine learning-based and synthetic aperture radar time-series data for rice classification over Sentinel-1 imagery
Full text linkRice extraction is critical in remote sensing, especially in Suphan Buri province, Thailand, using Sentinel-1 synthetic aperture radar (SAR) time-series data and advanced machine learning algorithms. Given the challenges of varied terrains and diverse crop types, the research employs different polarization modes (vertical transmit and vertical receive (VV), vertical transmit and horizontal receive (VH), and VV+VH) to enhance classification accuracy. The study evaluates the performance of three machine learning algorithms: random forest, extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM). The results demonstrate that combined VV+VH polarization outperforms VV and VH alone, providing better accuracy due to its ability to capture more detailed object features. LightGBM emerged as the most effective among the algorithms, particularly when dealing with large datasets. After hyperparameter tuning (n_estimators: 820, max_depth: 10, and learning_rate: 0.01), LightGBM achieved the highest accuracy. The rice class showed exceptional precision, recall, and F1-score, surpassing other land-use classes (agriculture/forest and urban areas). However, these classes still pose challenges, highlighting the need for future studies to integrate multi-sensor data and explore more sophisticated machine-learning models. This research offers a promising approach to enhancing rice monitoring and management in diverse agricultural landscapes, contributing to more accurate and efficient farming practices
Wireless sensor network using nRF24L01+ for precision agriculture
Full text linkPrecision agriculture is a strategy for varying inputs and cultivation methods to suit varying soil conditions and agricultural crops. In order to optimize precision agriculture, wireless sensor network (WSN) is suitable to be integrated. In this research, network devices that communicate using nRF24L01+ based WSN was proposed. As a prototype, four sensor nodes were employed to measure the parameters of air temperature and humidity, soil moisture, and power supply voltage. While, a sink node serves to store measurement data locally. The data are sent to the sink node with a mesh network topology and saved in a comma-separated values (CSV) file and local database. Experimental results show that each sensor node can measure all parameters and successfully send data to the sink node every 1 minute without losing the data. The mesh topology can route data transfer automatically. Round trip time (RTT) of each sensor node depends on the distance from each node. Average power consumption of all sensor nodes in send mode is between 84 mW and 90 mW. Meanwhile, in sleep mode, the sensor nodes 1 and 2 consumed around 21-22 mW and the sensor nodes 3 and 4 consumed around 30 mW which are lower than the send mode
Shallot disease classification system based on deep learning
Full text linkShallot is one of the important horticultural commodities for society and has high economic value. The problem with shallot cultivation is disease attacks on plants, one of which is Fusarium wilt. With the condition that the shallot commodity at the farmer level has a high failure rate, it is hoped that this research can assist farmers in providing information about shallot plants that have diseased plant characteristics using deep learning system convolutional neural network (CNN) method by utilizing leaf images on shallot plants. This research was conducted using the ResNet-18 architecture, with a total of 400 data in the dataset divided into 2 categories, namely healthy and diseased Fusarium wilt. The device used to carry out the classification process in this research is a Jetson Nano 2 GB. The ratio used to form a model from the dataset is 80-20 (80% training data and 20% validation data). The accuracy results for the classification of shallot plant diseases using real-time leaf images during the day have an average accuracy value of 68% on healthy plants and 62% on Fusarium wilt plants, while at night it has an average accuracy value of 53% on healthy plants and 47% on Fusarium wilt plants
Optimizing the best student selection: hybrid K-Means approach and entropy-grey relational analysis
Full text linkThe selection of the best students is an important process in recognizing students' achievements and dedication in various fields. Through careful and fair selection, students who stand out in both academic and non-academic terms can be identified and assigned. The purpose of the research on the use of hybrid entropy-grey relational analysis (GRA) and K-Means clustering in the selection of the best students is to develop a more objective, accurate, and comprehensive assessment system. The silhouette score results show that 2 clusters have a value of 0.5733, so in this study 2 clusters are used with the best cluster at cluster 0. Data from cluster 0 will be used in determining the best students using hybrid entropy-GRA. The results of the best student ranking using the hybrid entropy-GRA method, for the first best student with a final score of 0.25 were obtained by Mareta Amelia. The hybrid approach of K-Means and entropy-GRA offers a powerful tool to improve decision-making in the student selection process. The hybrid approach of K-Means grouping and entropy-GRA presents a powerful solution, improving the decision-making process and ensuring that high-achieving students are accurately recognized and rewarded
Environmental odor detection and classification with electronic nose system
Full text linkA prototype of an electronic nose (e-nose) system integrating a set of general-purpose gas sensors, an electronic module, and signal processing and classification methods has been designed and implemented to detect certain environmental odors that might pose a risk to human health. The proposed device explores the filter diagonalization method (FDM), an advanced signal processing technique for accurate spectral estimation, to detect the presence of odors together with random forest (RF), a popular machine learning algorithm, to classify the features of such spectra. Experimental results show that the proposed FDM-RF approach can recognize the targeted odors with an accuracy of 96.4%
Glioma segmentation using hybrid filter and modified African vulture optimization
Full text linkAccurate brain tumor segmentation is essential for managing gliomas, which arise from brain and spinal cord support cells. Traditional image processing and machine learning methods have improved tumor segmentation but are often limited by accuracy and noise handling. Recent advances in deep learning, particularly using U-Net and its variants, have achieved significant progress but still face challenges with heterogeneous data and real-time processing. This study introduces a hybrid bilateral mean filter for noise reduction coupled with an ensemble deep learning model that integrates U-Net, InceptionV2, InceptionResNetV2, and W-Net to enhance segmentation accuracy and efficiency. Additionally, we propose a novel modified African vulture optimization algorithm (MAVOA) to further refine segmentation performance. Evaluated on the BraTS 2020 dataset, our model achieved a loss of 0.023 with strong performance metrics: 98.2% accuracy, 97.2% mean intersection over union (IOU), and 99.1% precision. It effectively segmented glioma subregions with dice scores of 0.96 for necrotic areas, 0.97 for edema, and 0.91 for enhancing regions. On the BraTS 2021 dataset, the model maintained high accuracy 96.4%, mean IOU 95.9%, and dice coefficients of 0.91 for necrotic areas, 0.95 for edema, and 0.92 for enhancing regions
Enhancing data integrity in internet of things-based healthcare applications: a visualization approach for duplicate detection
Full text linkThis study addresses the critical issue of data duplication in healthcare-related internet of things (IoT) datasets, which can compromise the reliability of analyses and patient outcomes. A Python-based visualization framework using Pandas and Matplotlib was developed to detect and represent duplicate records. The methodology was applied to six cancer-related datasets sourced from Kaggle, ranging from 300 to 55,000 records, encompassing numerical, textual, and categorical data types. The visualization technique provided clear insights into duplication patterns, identifying specific counts such as 7 duplicates in the wearable device dataset, 19 in the thyroid recurrence dataset, and 534 in the synthetic healthcare electronic health record (EHR) dataset. Compared to traditional detection methods, the visualization tool facilitated faster and more intuitive initial data assessment, demonstrating its effectiveness for rapid quality checks in healthcare datasets. However, scalability limitations were observed in larger datasets, where visual clarity declined. These findings highlight the value of visualization as a preliminary data quality assessment tool and suggest future integration with advanced detection algorithms to enhance robustness and scalability
Improving COVID-19 chest X-ray classification via attention-based learning and fuzzy-augmented data diversity
Full text linkThis paper presents a hybrid deep learning (DL) framework that combines model-level and data-level enhancements to improve classification performance without compromising clinical relevance. The proposed framework consisted of an EfficientNetB0 model with a hybrid attention module, which focused attention both spatially and channel-wise, and a VGG-16 model that was trained on training data augmented using a fuzzy-logic-based contrast and brightness enhancement. The attention module focused the model by recalibrating the features in an adaptive manner. The fuzzy-logic augmentation increased data diversity while maintaining the anatomical fidelity of the medical image domain. In addition, an uncertainty-aware ensemble approach was utilized to combine both models' predictions, which considered model confidence and entropy of the predictions, to enhance the reliability of the predictions. The proposed framework achieves a classification accuracy of 99.6%, outperforming several existing approaches