Jurnal Teknologi Informasi dan Terapan

Fuzzy Sugeno Model for SNR-Based Adaptive Modulation in Underwater Acoustic Communication

2025-12-31T07:39:21+00:00

Underwater communication faces significant challenges due to the dynamic characteristics of the channel and is strongly influenced by the physicochemical parameters of the water. This study proposes channel quality modeling using the Sugeno Fuzzy Inference System (FIS) with input variables of temperature, salinity, dissolved oxygen (DO), and turbidity. The system produces a Signal-to-Noise Ratio (SNR) output that is used as a basis for channel quality mapping, Bit Error Rate (BER) estimation, and the selection of adaptive modulation techniques (BPSK, QPSK, or 16QAM). Simulation results show that the Sugeno fuzzy model is able to follow the theoretical pattern well, where increasing temperature, salinity, and turbidity decrease the SNR value, while DO plays a role in maintaining channel stability. Based on the test results, at high SNR (≥ 15 dB) the system recommends 16QAM, at medium SNR (11–15 dB) QPSK, and at low SNR (≤ 10 dB) BPSK. This approach has proven effective in suppressing BER and increasing the reliability of underwater acoustic communications in fluctuating mangrove water environments.

Clustering Analysis for Green Economy and Citizens-Based Social Forestry Business Development Model

2025-12-31T07:39:21+00:00

This study aims to prove that clustering analysis can optimize the development model of social forestry businesses based on green economy and citizens. Clustering analysis can use machine learning methods. Some of these methods are K-Means and K-Medoids. First, the research data was obtained from the assessment results of forest edge residents. Residents assessed 13 green economy variables. The social forestry business development model based on green economy and citizens requires labeled data. Therefore, this study compares the performance of K-Means and K-Medoids to cluster the assessment data of forest edge residents. To determine its performance, this study uses three variations of k values, namely K = 4, K = 8, and K = 12. Performance testing uses the Davies Bouldin Index (DBI) method and computation time. Based on Davies Bouldin test, K-Means method is better than K-Medoids at K = 4, but K-Medoids method is better than K-Means at K = 8 and K = 12. Based on computation time test, K-Means method is better than K-Medoids. Based on this test, K-Means method is more suitable for big data and fast computing time.

Defense System Design on Internet of Things Servers Using Gemstone Architecture to Mitigate of Denial of Service Attacks

2025-12-31T07:39:21+00:00

One of the problems in the Internet of Things (IoT) system is the Distributed Denial of Service (DDoS) attack on the information technology infrastructure on the Internet network. This is because the IoT device system does not have a portal gateway configuration that is capable of providing the necessary security and privacy protection. In this study, the focus is on the mechanism for reducing the impact of DDoS attacks of the http flood type on the layer framework using PHP Swoole. PHP Swoole utilizes event-driven which provides several features to access the transport layer in onConnect so that it can be used to implement initial security such as access control lists, connection concurrency management, and optimizing server performance. The methodology used is to develop a TCP connection filtering algorithm by implementing a simple queue system by accepting 67% of connections to be forwarded to the next layer and 33% of connections will be queued for a timeout of 2 seconds. The results of this study show that the server can minimize the impact of DDoS and handle traffic specifically for http requests with an average latency of 871.8ms.

Spatial Mapping of Endemic Diseases in Lamongan Regency Using Natural Breaks Clustering

2025-12-31T07:39:21+00:00

Endemic diseases are health problems that require continuous monitoring to prevent their spread. However, the lack of interactive spatial data visualization can make it difficult for health agencies to analyze the distribution of diseases effectively. This study aims to build a spatial mapping system for endemic diseases in Lamongan Regency using the Natural Breaks clustering method. The data used in this study include sub-district data, population, and endemic diseases obtained from the Lamongan Regency health profile website. The system was developed using the Laravel 10 framework and MySQL database, with Metabase as a data visualization tool. The geovisualization method is applied to display the distribution of diseases in the form of interactive maps based on sub-districts. The system allows users to trace the spatial and temporal trends of endemic diseases more easily. The test results show that the system is able to display information accurately and interactively, so that it can help health agencies in monitoring and planning regional health policies more effectively

Performance Comparison of CNN Transfer Learning Models for Coffee Bean Quality Classification

2025-12-31T07:39:21+00:00

According to SNI Standard No. 01-2907-2008, accurate sorting of coffee beans is crucial for improving export value. Manual sorting is time-consuming, subjective, and error-prone, especially when visual differences are subtle between roast levels. This study proposes and evaluates an automatic, machine-learning based system to support quality assurance in coffee production. We compare three transfer-learning CNN architectures: Xception, MobileNetV2, and EfficientNet-B1 on a publicly available dataset of 1,600 coffee bean images divided into four classes (dark, medium, light, green). All models were trained with the same preprocessing and hyperparameter settings. EfficientNet-B1 achieved the highest test accuracy (100%), followed by Xception (99.5%) and MobileNetV2 (97%). We discuss trade-offs between accuracy and computational efficiency and recommend EfficientNet-B1 for high-accuracy applications and MobileNetV2 for edge/mobile deployment.

CNN Implementation in Progressive Web App for Automatic Garbage Classification using TensorFlow.js

2025-12-31T07:39:21+00:00

The substantial and continuously increasing volume of global waste has become a critical environmental challenge, exacerbating the inherent inefficiency of conventional manual sorting techniques. This research addresses this problem by developing and evaluating an automated waste classification system using Convolutional Neural Networks (CNN), specifically the VGG16 architecture, integrated into a Progressive Web App (PWA) to enhance accessibility and sorting efficiency. Our primary goal is to deliver an intelligent, lightweight, and cross-platform solution capable of performing client-side inference on diverse devices. The VGG16 model was retrained using transfer learning on a validated public dataset of 10,365 images, comprising two classes (organic and inorganic waste). The trained model was converted to a browser-compatible format, TensorFlow.js, and deployed within the PWA framework which utilizes Service Workers for offline capabilities. Despite the significant challenge posed by the VGG16 model's large size, the system successfully performed client-side inference by prioritizing GPU acceleration and achieved 0.94 overall accuracy on the test dataset². This result, supported by high F1-scores for both waste categories, confirms that deploying high-accuracy CNN models at the edge using PWA and TensorFlow.js is a feasible and promising strategy for practical, technology-based waste management and environmental education.

Lightweight Deep Learning Approach for Sugarcane Leaf Disease Classification Using MobileNetV2

2025-12-31T07:39:21+00:00

Sugarcane is one of Indonesia’s strategic crops, yet its productivity is frequently disrupted by leaf diseases such as yellow leaf, rust, and red rot. Previous studies have shown that deep learning models are promising for plant disease detection, but many of them rely on heavy architectures that limit deployment in real-world agricultural settings. To address this gap, this study applies MobileNetV2, a lightweight Convolutional Neural Network, for the classification of sugarcane leaf diseases. Using the publicly available Kaggle dataset, the model was trained and evaluated on four classes: healthy, yellow leaf, rust, and red rot. The results demonstrate that MobileNetV2 achieved 97.0% test accuracy, with strong precision, recall, and F1-scores across all categories. These findings highlight that efficient deep learning architectures can deliver reliable disease classification while remaining practical for implementation on mobile or edge devices. Compared with previous approaches, this study contributes by demonstrating that lightweight model like MobileNetV2 can provide a balance of accuracy and efficiency, making them suitable for supporting precision agriculture practices in resource-limited environments

Digital Image-Based Chili Quality Detection Using a Web-Based Convolutional Neural Network

2025-12-31T08:02:53+00:00

ABSTRACT Chili is one of the main horticultural commodities in Indonesia, with high economic value and stable market demand. Accurate determination of chili quality levels is an important factor in maintaining quality, selling price, and distribution efficiency. Until now, the process of assessing chili quality has generally been carried out manually through direct visual inspection by experts or field officers. This traditional approach has limitations, such as varying levels of accuracy due to assessor subjectivity and the limited availability of experts. Advancements in digital image processing technology, particularly deep learning, offer opportunities to develop more accurate and consistent automated detection systems. This study proposes a Convolutional Neural Network (CNN) model to classify chili quality levels based on digital images, which is then integrated into a web-based application. This study uses a dataset of 405 chili images from 11 varietal categories, each labeled with quality (good, pest-infested, or unknown), which undergoes preprocessing stages including resizing, normalization, and data augmentation. The CNN model was designed with convolutional layers, max-pooling, dense layers, and a Softmax activation function, and was trained using the Adam optimizer and Categorical Cross-Entropy Loss. The web application implementation was carried out using the Flask framework, allowing users to upload images and obtain prediction results in real time. The testing results showed that the developed CNN model achieved an accuracy of 1.000 on the test data, with reliable detection performance under variations in lighting and image backgrounds. This research contributes to the development of smart agriculture technology by providing an accurate, fast, and easily accessible solution for chili quality detection

Child Stunting Risk Analysis through Machine Learning Models using XGBoost Algorithm

2025-12-31T07:39:21+00:00

Stunting is a chronic nutritional disorder that significantly affects child growth, development, and the overall quality of future human resources. According to the 2024 Indonesian Nutritional Status Survey (SSGI), the prevalence of stunting remains high at 19.8%, equivalent to approximately 4.48 million children under five. Early detection of stunting risk is essential for timely and data-driven interventions. This study employed the CRISP-DM methodology, encompassing business understanding, data collection, preparation, modeling, and evaluation phases. The dataset was processed through cleaning, variable encoding, and stunting status classification based on WHO standards. An XGBoost-based predictive model was developed and evaluated using accuracy, precision, recall, and F1-score metrics. The model achieved 98% accuracy in predicting stunting risk. Feature importance analysis revealed that height is the most influential variable determining stunting risk.

Face Tracker Audio for Saronen Music Using Augmented Reality on Social Media

2025-12-31T07:39:21+00:00

Saronen music is a traditional Madurese music commonly played at cultural and traditional events. Saronen music typically combines traditional instruments such as gamelan, trumpet, kenong, korca, large drums, and small drums, producing a unique sound characteristic of Madura. This research aims to preserve saronen music through the development of an interactive filter on Instagram and Facebook using augmented reality. The face tracker feature allows users to interact with each instrument using head movements or facial expressions. The development method, MDLC encompasses concept, design, material collection, design, testing, and distribution. Functional research results indicate the filter can run well on Android and iOS operating systems. Testing on 31 social media users yielded positive feedback from the majority, stating that the filter has clear instructions, a low error rate, is easy to use, satisfying, and attractive. This innovation is expected to be a means of preserving saronen music for the younger generation through social media.

PaletteStream: A Promotional, and Community Web-Based Platform for Visual Artists with Gamification Implementation

2025-12-31T07:39:21+00:00

In the digital era, social media has become a primary medium for visual artists to promote their work and engage with audiences. However, mainstream platforms such as Instagram and Facebook often fall short of addressing the specific needs of artists due to algorithms that are not tailored to the art domain. While specialized platforms like DeviantArt and ArtStation exist, most have yet to implement effective gamification features that could enhance user engagement and motivation. To address these challenges, PaletteStream is developed as a dedicated web-based platform for visual artists, focusing on collaboration, promotion, and community building. By integrating gamification elements using MVC Architecture, PaletteStream aims to facilitate artistic collaboration, improve the efficiency of art promotion, and foster an active and supportive artistic community. This project also contributes to technological innovation in the arts and advances in application development centered on user experience. The results reveal that the PaletteStream Platform for gamification systems properly awards points and badges based on established rules and provides considerable performance gains with an average API response time of less than 3 seconds.

Optimizing Security with an Iot: A Data-Driven Visitor Identification Framework

2025-12-31T07:39:22+00:00

Security is an important factor that must be considered. The properties, such as computers, sensor devices, teaching aids, and other equipment, must be monitored in the best possible condition to be used whenever needed. If these items are stolen, it will interfere with students' learning abilities so that Learning Outcomes are not met. Currently, visitor identification systems are evolving, initiated by the implementation of IoT devices, face recognition, voice recognition, and so on. When these systems were executed, several obstacles were still found, such as the identification process being slow, requiring large amounts of training data, and the application interface only running on smartphone devices. Therefore, a breakthrough is needed to recognize visitors quickly, easily, and to boost protection. In this research, the author proposes an Identification Information System (IIS) for room visitors using a Data-Driven Modeling method based on the Internet of Things (IoT). This system is equipped with an IoT driver module to interact with the Raspberry Pi and a magnetic lock. It aims to allow a room administrator to control and lock doors online via a computer or smartphone. Based on the experiments that have been carried out, the proposed system is adequate to run optimally from some testing cases that have been designed.