Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Not a member yet
776 research outputs found
Sort by
A Novel Methodology for Container Scheduling and Load Balancing in Distributed Environments
Full text linkDeployment of applications in distributed environments via containers has gained huge popularity lately, specifically with cloud-based ecosystems. Inspired by the quick growth of container usage and deployment in distributed environments, efficient scheduling techniques are of prior significance embedded with load balancing in it for cloud computing tasks. Most of the scheduling strategies adopt conventional methods and fail to execute efficiently in the dynamic cloud or distributed environments where applications around the world depend on them for scalability, efficiency, and availability. Existing applications focus more on performance metrics instead of scheduling efficiency, so often they offer performance that can come at the expense of scheduling. This paper proposes a new algorithm that includes consideration of contention over the network, along with efficient canister planning and load distribution. The algorithm we have designed to achieve the proposed scheduling and load balancing is Contention-aware Greedy Heuristic Scheduling and Load Balancing for Containers (CGHSLBC), which has been extensively evaluated under continuous workload and has outperformed current state-of-the-art algorithms by 20% in load balancing efficiency and 25% in network contention reduction, demonstrating its promise for container scheduling in dynamic distributed environments
A Novel Compact CPW-fed Octagonal-Shaped Slotted Antenna for UWB Applications
Full text linkA size-reduced CPW-fed Ultra-Wideband (UWB) octagonal-shaped patch antenna with a combination of multiple slots designed for UWB applications is proposed here. The proposed low-profile antenna includes three equalsized slots in an octagonal radiating patch. Moreover, better matching is provided by the feedline's U-shaped slots. A new combination of CPW configuration with a slotted octagonal patch increases bandwidth and reduces antenna size. The fabricated prototype of this octagonal-shaped antenna is situated on a basic FR4 substrate with a relative permittivity of εr = 4.3. The suggested antenna's substrate size measures 15 x 21 x 1.6 mm3 and has an 8.8 GHz overall bandwidth, which includes the frequency range of 3 GHz to 11.8 GHz. Variations in gain range from 1.3 to 3.2 dB, with an average overall efficiency above 81 %. This antenna has been fabricated and successfully validated with simulated results. Other features include its compactness, directivity, realized gain, and stable radiation properties across the entire operating band, proving its effectiveness
Sliding Mode Based Nonlinear Control of a Three-Level NPC-Type DC-DC Bidirectional Converter for HV Applications
Full text linkThis paper proposes the design and a comparative study of three sliding modebased controllers—sliding mode control (SMC), fuzzy logic sliding mode control (Fuzzy-SMC), and ANFIS sliding mode control (Anfis-SMC)— applied to a newly designed DC-DC isolated bidirectional converter featuring a three-level neutral point clamped (3L-NPC) topology. Utilizing single phase shift (SPS) modulation enabled by a simplified analytical model, the controllers demonstrate strong robustness against load disturbances and input voltage variations, while ensuring accurate reference voltage tracking in both Boost and Buck modes. Comparative analysis shows that all three controllers outperform the traditional PI controller across most performance metrics. Voltage balancing is effectively maintained through an auxiliary inductive circuit. The investigated control schemes are validated via MATLABSimulink simulations, confirming their suitability for the efficient control of the studied converter
Electrocardiogram Waveforms Diagnosis based on Wavelet Representation and SqueezeNet Model
Full text linkAArrhythmia is an irregular in a person's beating heart that can happen occasionally. Heart rhythm problems can have disastrous results and seriously endanger health. Visually analyzing ECG data might be complex due to its large amount of information. Designing an automated method to assess the massive amount of ECG data is crucial. This research shows continuous wavelet transform (CWT) and deep learning strategies to automate detection and classification processes to examine three different ECG signals: congestive heart failure (CHF), normal sinus rhythm (NSR), and arrhythmia (ARR). CWT converts ECG signals into scalogram images for noise reduction and feature extraction. In deep learning, the modified SqueezeNet is employed to recognize the output of CWT, which is produced by the input of the ECG data. The proposed technique achieved 83.3%, 100%, and 94.7% accuracy in detecting CHF, NSR, and ARR. A comprehensive approach for classifying arrhythmias has been proposed, in which scalogram pictures of ECG waves are trained using the SqueezeNet model. The outcomes are superior to other current techniques and will significantly reduce wrong diagnose
Simulation-Based Evaluation of Dense Convolutional Neural Networks for Skin Cancer Detection
Full text linkSkin cancer, particularly melanoma, poses significant challenges to public health, with early detection being critical for effective treatment. Traditional diagnostic methods often fall short, particularly in resource-limited settings. In response, artificial intelligence (AI) techniques, especially deep learning models, have emerged as promising tools for automated skin cancer detection. This study evaluates the performance of Dense Convolutional Neural Networks (DCNNs) in classifying and detecting skin lesions, leveraging simulation-based approaches to assess the effectiveness of various AI models. Utilizing datasets such as HAM10000 and ISIC2017, which contain a wide variety of skin types and lesion stages, the models were trained and tested using key performance metrics such as accuracy, precision, recall, and F1-score. The results shows that DCNNs outperformed traditional machine learning techniques like Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Decision Trees (DT), demonstrating superior accuracy, generalization ability, and efficiency in handling large, imbalanced datasets. The simulation-based approach provided insights into the ability of DCNN models to manage dataset inconsistencies and class imbalances, showcasing their potential as robust tools for skin cancer detection. These findings highlight the ability of AI in advancing dermatological diagnostics, offering more timely and accurate detection, and potentially improving patient outcome
Methodological Approach to Automated Recognition of Atrial Fibrillation and Subsequent Classification
Full text linkThis study considers topical issues aimed at improving the methodology of early recognition of atrial fibrillation and monitoring its treatment against the background of other heart rhythm disorders. The task set in this study is an essential component of the search for solutions, whose purpose is to increase the efficiency of information systems for cardiac diagnostics and monitoring within the framework of complex research to improve the means of heart rhythm analysis and arrhythmia recognition. In the context of this study, a linear discriminant analysis approach based on the concept of K-entropy was proposed as a means of automating the procedure for the recognition of AF against the background of other rhythm disorders using a limited data sample. With regard to the classification of atrial fibrillation samples, the use of decisive rules and arrhythmia types, based on the analysis of scatterograms, is put forth as a solution. The results of the proposed methods for recognizing the presence of atrial fibrillation and its classification demonstrated superior performance when compared to existing methods. The proposed method exhibited a specificity of 98.5% and a sensitivity of 98%. The proposed method for determining the presence of atrial fibrillation demonstrates suboptimal accuracy when applied to a limited sample size. Further development of the method should be concentrated in this area
Leveraging Ensemble Learning Models for Human Activity Recognition
Full text linkThis paper presents a novel method for categorizing human activities by processing sensor data obtained from IoT devices, focusing on improving accuracy. The proposed approach leverages an ensemble learning framework with majority voting, integrating hyperparameter-optimized classifiers to enhance predictive performance.The ensemble approach minimizes individual biases and errors, effectively handling the variability inherent in sensor data. Adequate preprocessing techniques refine data quality before feeding it into the model. A diverse set of base classifiers, such as KNN, Decision Tree, and Random Forests, are considered for classification. Hyperparameter-optimized KNN Grid Search, Gradient-Boosted Decision Trees, and Random Forests with Optimal Trees are ensembled. Extensive experiments were conducted on Human Activity Recognition datasets, WISDM, HAPT, HAR, and KU-HAR.The model performance was rigorously evaluated using classification metrics such as accuracy, precision, recall, and F1-score. Empirical results demonstrate that the proposed ensemble method significantly enhances classification accuracy. Future research will investigate applying deep learning techniques to capture complex feature interactions within sensor data better
A Message Transmission Scheme for IoT Inspired by The Shamir Scheme and Based on The Hidden Number Problem (HNP)
Full text linkThe Internet of Things (IoT) requires cryptographic mechanisms that are both lightweight and resistant to emerging attacks. Classical public-key protocols such as RSA or ElGamal, as well as many post-quantum lattice-based schemes, consume too many resources for devices with limited memory, processing power, and energy. In this study, we propose a three-pass message transmission protocol that avoids any prior key exchange. Inspired by Shamir’s keyless scheme and relying on the hardness of the Hidden Number Problem (HNP) and the Decisional Diffie–Hellman (DDH) assumption, the protocol operates in a finite field with safe primes and refreshes random masks at each execution, providing strong resistance to brute-force and small-subgroup attacks. We formally prove IND-CPA security and implement the HNP 3-pass scheme, showing that each pass executes in 1.4--1.8 ms on a workstation, with 132-byte public keys and 192--256-byte secret keys. Estimated energy consumption per iteration is 101.562 mJ. Comparative simulations on a workstation and embedded platforms (Arduino Uno and Raspberry Pi) against RSA-512, ECC (secp256r1/secp521r1), and post-quantum Kyber-512 show that our scheme achieves execution times comparable to ECC and Shamir’s 3-pass protocol, is significantly faster than RSA, and consumes less energy than Kyber-512. This combination of low latency, moderate key sizes, and energy efficiency highlights the practicality of the HNP 3-pass protocol for resource-constrained IoT environment
Design of Compact, Low-cost Electrochemical Electronic Reader (Potentiostat) for Chemical Compound Analysis
Full text linkThe development of low-cost electrochemical electronic reader (potentiostat) has been vastly growing recently since it could serve as a rapid chemical compound detection. Nevertheless, the realization of a low-cost potentiostat, having a compact size and providing multi-feature, is challenging. Here, a compact, low-cost potentiostat, supporting multi electrochemical methods was demonstrated. The total dimension was (10.8 x 4.6 x 3.5) cm3 with the weight of only 75.6 g. By using 16-bit analog-to-digital converter (ADC) and 12-bit digital-to-analog converter (DAC) components, the potentiostat facilitated a current input range of ±580 µA with a resolution of 23.5 nA and a voltage sweep range of ±1.5 V with a resolution of 800 µV. For the electrochemical measurement, it supported cyclic voltammetry (CV) and differential pulse voltammetry (DPV) methods. Furthermore, in comparison with a commercial potentiostat (Sensit Smart), the potentiostat showcased a good accuracy performance. In detail, the average relative accuracy of the CV method test was 91% and 95% for the anodic (Ianodic) and cathodic peak currents (Icathodic), respectively. For DPV method, the lowest relative accuracy of the peak current (Ipeak) was still 83%. These results demonstrated that our potentiostat could promisingly be utilized for chemical compound detection in low-cost setting or rural areas
DCDNet: A Deep Learning Framework for Automated Detection and Localization of Dental Caries Using Oral Imagery
Full text linkDental caries is a common oral health condition that requires early diagnosis and identification for effective intervention. Existing deep models, such as Faster R-CNN, YOLOv3, SSD, or RetinaNet, exhibit great effectiveness in generic medical imaging; however, they struggle to precisely and explicitly handle localization in complex dental radiographs. In this paper, we propose DCDNet, a convolutional neural network architecture specifically designed for the detection and segmentation of dental caries in oral X-ray images. However, such deep learning methods currently lack strong generalization due to imbalanced training data, limited lesion-localization ability, and noninterpretable features, which hamper their utility for large-scale clinical evaluation. In addition, most models overlook the severity distinction between classes, which is less ideal for the entire diagnosis and treatment planning process. DCDNet was trained and tested on the UFBA UESC Dental Image Dataset, which comprises over 1,500 labeled grayscale dental radiographic images. The proposed network incorporates multiscale feature extraction, residual connections, and non-maximum suppression (NMS) for more accurate classification and bounding box prediction. Data augmentation techniques were used to increase generalization. The model was evaluated based on accuracy, precision, recall, and F1-score, and compared with ResNet50, VGG16, AlexNet, Faster R-CNN, YOLOv3, SSD, and RetinaNet in terms of accuracy. DCDNet achieved excellent performance in all its performance indices, with precision at 97.23%, recall at 97.02%, F1-score at 97.12%, and overall accuracy at 97.61%. Experiments demonstrate that the proposed DCDNet surpasses all the baselines and state-of-the-art methods by a significant margin. Ablation experiments validated the importance of residual connections, NMS, and data augmentation for performance improvement. DCDNet represents a significant step toward automatic dental diagnosis, having successfully detected and localized carious lesions in X-ray images. Its design overcomes the drawbacks of previous models and is a ready option for integration into clinical routine