Journal of Informatics And Telecommunication Engineering
Not a member yet
373 research outputs found
Sort by
Development of a Smart IoT Dashboard for Sustainable River Water Quality Monitoring in Ciujung River
No full textThe water quality of the Ciujung River in Serang has experienced a significant decline due to domestic and industrial waste pollution, directly affecting public health and environmental sustainability. Current monitoring systems remain largely manual and lack responsiveness, resulting in delayed and less data-driven pollution management. This study aims to develop an Internet of Things (IoT)-based water quality monitoring system integrated with an intelligent dashboard to support sustainable environmental programs. The proposed system monitors key water quality parameters, including pH, temperature, turbidity, and total dissolved solids (TDS), in real time. The methodology includes designing a microcontroller-based sensor prototype, integrating data communication modules (LoRa/GSM), processing data via a cloud server, and implementing interactive visualization through a web-based dashboard. Furthermore, the system features an early warning mechanism when water parameters exceed environmental quality thresholds. Field trials are conducted at several strategic points along the Ciujung River to evaluate data acquisition reliability, connectivity stability, and sensor accuracy. The expected outcome is an efficient, responsive, and adaptive monitoring system that supports data-driven decision-making in river water management and reinforces commitments to sustainable development
Combined Barker-M-Sequence Coded LFM for High-Performance Subarray-MIMO Radar Applications
No full textSubarray-Multiple-Input Multiple-Output (SMIMO) radar is an advanced technology that integrates the advantages of phased-array and MIMO radars to enhance target detection resolution. A key challenge in SMIMO implementation lies in improving velocity resolution without compromising spectral efficiency, while maintaining accurate target detection capability under high sidelobe levels and inter-channel interference. This study proposes a novel approach—Combined Barker-M-Sequence Coded LFM—in which the LFM signal is phase-modulated using a hybrid code formed by concatenating a Barker sequence (length 11) and an M-sequence (length 7). Simulation results show that the proposed signal achieves a Peak Sidelobe Ratio (PSLR) of −20.83 dB, significantly outperforming LFM-Barker (−8.45 dB) and LFM-M-sequence (−16.3 dB). It also delivers a velocity resolution of 0.95 m/s and a range resolution of 225 m, representing a 38% improvement over standard LFM. Moreover, under SNR = −5 dB, the system achieves a SINR gain of 4.7 dB relative to LFM-M-sequence. This approach enables more efficient waveform utilization in modern radar applications—such as air surveillance, military defense, and autonomous vehicles—particularly in challenging environments characterized by low SNR, multipath propagation, and high clutter
Anomaly Detection in Cloud Device-Based Information Technology Infrastructure Using Isolation Forest Algorithm
No full textCloud device-based information technology infrastructure generates large volumes of operational data that are dynamic and heterogeneous, increasing the complexity of monitoring and anomaly detection processes. Conventional rule-based approaches and supervised learning methods are often less effective due to limited labeled data and their inability to detect newly emerging anomaly patterns. Therefore, this study aims to apply and evaluate the Isolation Forest algorithm as an anomaly detection method for cloud device-based information technology infrastructure. The research data consist of system and network performance metrics, including CPU usage, memory utilization, disk activity, and network traffic collected from a cloud environment. The research stages include data preprocessing, normalization, and feature selection to improve data quality and model performance. The Isolation Forest algorithm is implemented using an unsupervised learning approach, where anomalies are identified based on the algorithm’s ability to isolate data points that exhibit characteristics deviating from the majority of normal data. Model performance is evaluated using a confusion matrix with accuracy, precision, recall, and F1-score metrics, while parameter optimization is conducted using the grid search method to obtain the best configuration. The results indicate that the Isolation Forest algorithm is able to detect anomalies effectively, achieving high accuracy and a good balance between precision and recall. The model with optimal parameters demonstrates improved performance by reducing detection errors compared to the baseline configuration. Thus, the Isolation Forest algorithm can serve as a reliable and scalable solution to support monitoring activities and enhance the reliability of cloud infrastructure
Design and Engineering of an AI-Enabled Mobile Microlearning Application Integrating Short-Form Video and Learning Analytics for Vocational Soft Skills Development
No full textThe rapid growth of mobile technologies has reshaped how learning systems are designed, deployed, and evaluated, particularly in vocational education contexts. From a Mobile Software Engineering perspective, learning platforms must address constraints such as short interaction cycles, heterogeneous devices, scalability, and real-time analytics. This study focuses on the design and engineering of an AI-enabled mobile microlearning application that integrates short-form video, learning analytics, and LMS services to support vocational students’ soft-skills development. The proposed system is engineered as a mobile-first application with modular micro-content (60–180 seconds), rule-based personalization, and event-driven analytics to capture user interaction patterns. A Research and Development approach using the ADDIE framework is adopted, with emphasis on the software design, architecture, and prototyping stages. Validation involves expert review of system usability, content–software alignment, and limited pilot testing with end users. The results demonstrate that a mobile-engineered microlearning system can achieve high completion rates, acceptable latency under concurrent access, and effective analytics-driven feedback loops. The study contributes a practical mobile software engineering artefact and design insights for AI-enabled learning applications in vocational education
Hybrid CNN-LSTM for Indonesian Cyberbullying Detection on Social Media X
No full textCyberbullying on social media platform X has become a critical digital threat and requires automatic detection mechanisms to mitigate psychological impacts on victims. This study proposes a hybrid deep learning architecture that combines Convolutional Neural Network (CNN) for local feature extraction and Long Short-Term Memory (LSTM) for sequential context understanding in classifying Indonesian language cyberbullying comments. This study evaluates model performance using a dataset of 13,677 comments from social media X through a series of systematic testing scenarios, including the impact of regularization, utilization of FastText embeddings, and comparative studies against state-of-the-art models. Experimental results demonstrate that the Early Stopping mechanism is a critical factor in this architecture, where without this mechanism the model experiences accuracy degradation of up to 32%. The proposed CNN-LSTM model achieves 88.38% accuracy and 88.00% F1-Score, improving to 0.9559 AUC with FastText integration. This model achieves over 97% of IndoBERTweet's performance with 22 times lower computational complexity (4.97 million versus 110.88 million parameters) and outperforms machine learning methods such as SVM with an accuracy margin of more than 10 percentage points. This study concludes that the CNN-LSTM architecture offers a robust and efficient solution for cyberbullying detection, particularly for resource-constrained environment
Integrating Automatic Stock Monitoring and Digital Inventory Systems for MSMEs A Mobile Application Approach (Case Study in Serang City, Indonesia)
No full textMicro, Small, and Medium Enterprises (MSMEs) in Indonesia continue to face inefficiencies in inventory management due to manual stock recording, data inconsistency, and delays in operational decision-making. In Serang City, these challenges often lead to stockouts, excess inventory, and limited business scalability. This study aims to develop and evaluate a mobile-based automated inventory management system that supports real-time stock monitoring and decision-making for MSMEs. The research employs a Research and Development (R&D) approach integrated with the Agile-Scrum methodology, encompassing problem identification, user requirement analysis, system design, prototype development, functional testing, and usability evaluation. Functional validation was conducted using black box testing, while system usability was assessed using the System Usability Scale (SUS) involving 15 MSME users. The results indicate that all core system functions achieved a 100% success rate, including automated stock recording, cloud-based data synchronization, real-time notifications, and dashboard analytics. The usability evaluation produced an average SUS score of 82.5, classified as Excellent, indicating high user acceptance and ease of use. These findings demonstrate that the proposed system effectively improves inventory accuracy, operational efficiency, and decision-making quality, contributing to MSME digital transformation in developing regions
Classification of Bougainvillea Plant Types Using Convolutional Neural Network Algorithm
No full textBougainvillea is one of the most popular ornamental plants, featuring a variety of types with morphological characteristics that often appear very similar. This resemblance frequently complicates the conventional identification process, particularly for sellers and buyers at Rabiku Florist. This study aims to develop an Android application capable of automatically classifying different bougainvillea types using a Convolutional Neural Network (CNN) algorithm. The system is developed using the Rapid Application Development (RAD) methodology, leveraging the MobileNetV2 architecture and integrating it with the TensorFlow Lite framework to ensure compatibility with mobile devices. The application is designed to identify five types of bougainvillea using digital images captured via the device’s camera or selected from the user’s gallery. Based on implementation results, the system demonstrates strong classification performance and delivers accurate information to users. This application is intended to serve as a practical and user-friendly tool for both the general public and businesses in accurately identifying bougainvillea species.Keywords: Image Classification, Bougainvillea, Convolutional Neural Network, MobileNetV2, Android
Multi-Detection System Using Faster R-CNN for Fish Species Classification and Quality Assessment on Android
No full textSpecies identification and quality assessment of fish in trade still rely on manual visual observation, which is subjective and requires specialized expertise. This method's limitations make it difficult for consumers to distinguish species with similar morphology and accurately assess fish quality, which can lead to inappropriate purchasing decisions. This research develops a multi-detection system based on Faster R-CNN with VGG16 backbone for fish species classification and quality assessment simultaneously on Android platform. The system uses convolutional layers to extract visual features from input images, Region Proposal Network for fish object detection and localization, and fully connected layers for simultaneous classification of species and quality levels. The research dataset consists of 3,000 images of five fish species (gourami, tilapia, nile tilapia, snapper, and pomfret) with four quality levels, divided into 2,400 training images and 600 testing images. The trained model is converted to TensorFlow Lite format for implementation on Android devices. Test results show the multi-detection system achieves 92% accuracy in fish species classification and quality assessment, demonstrating the effectiveness of the Faster R-CNN approach for multi-detection applications in the Android-based fisheries sector
Enhancing Oil Palm Leaf Disease Classification using a Pruned SqueezeNet Architecture
No full textThe SqueezeNet architecture is known to be effective but possesses a considerable number of parameters, which can be optimized using pruning—a compression technique that significantly reduces model parameters without sacrificing accuracy. This research aims to apply the L2-Norm based pruning method to the SqueezeNet architecture and compare its performance (accuracy and efficiency) against the default SqueezeNet model for classifying four classes of oil palm leaf diseases. The study used a primary dataset of 4,000 images, divided into training (70%), validation (20%), and testing (10%) sets. The SqueezeNet architecture was pruned using L2-Norm structured pruning with a uniform distribution at rates from 10% to 50%, followed by fine-tuning. The results show that the default SqueezeNet achieved 97.50% accuracy with 724,548 parameters. Significantly, a 10% pruning rate actually increased the accuracy to a high of 99.25% while simultaneously reducing the parameters to 579,036. Overly aggressive pruning, such as 40%, drastically decreased accuracy to 93.25%. It is concluded that the 10% pruning rate is the most optimal, proving that this method not only makes SqueezeNet lighter but also more effective. This 10% pruned model is highly suitable for application implementation due to its enhanced efficiency. Future research is recommended to validate these findings using a more diverse dataset and to test the model on actual edge devices
Integrating Blockchain-Based Smart Contracts for Digital Certification: A Micro-Credentials Model for Vocational Higher Education
No full textThe rapid advancement of the digital industry requires vocational education in Indonesia to undergo transformation, particularly in providing competency validation systems that are efficient, adaptive, and trustworthy. In reality, however, competency certification processes in many vocational institutions are still conducted manually and tend to be bureaucratic, limiting their ability to respond to the dynamic needs of industry. This condition may reduce graduates’ competitiveness and widen the skills gap between vocational education and the labor market. Micro-credentials have emerged as an innovative approach to recognizing competencies in a modular, flexible, and industry-oriented manner. Nevertheless, their implementation still faces significant challenges, especially in terms of validation speed, reliability, and transparency. To address these challenges, this study develops a micro-credential–based competency validation model integrated with blockchain technology through the use of smart contracts at Politeknik LP3I Medan. This research adopts a Research and Development (R&D) approach based on the Borg and Gall model, including needs analysis, learning module design, system development, limited trials, expert validation, and effectiveness evaluation. Alpha testing involving 15 students demonstrates a system success rate of 95%, with an average verification time of 14.4 seconds. Usability evaluation indicates that the system is user-friendly and well accepted