Digital System and Computing

Farel Abid Yasser Prayanto, Rifki Indra Perwira

While universities heavily rely on digital information systems, static websites and manual administrative communication often limit accessibility and responsiveness for students seeking academic information. To address this, this study developed and evaluated an academic chatbot using the LLaMA 3.1 large language model integrated with a Retrieval-Augmented Generation (RAG) framework for Informatics students at Universitas Pembangunan Nasional “Veteran” Yogyakarta. Employing a Rapid Application Development approach, 263 institutional document chunks were processed to construct a knowledge base for a hybrid retrieval pipeline that combines BM25 lexical search and semantic vector similarity. The proposed system was comprehensively benchmarked against standalone lexical-only and semantic-only baselines using both RAG-specific and natural language generation (NLG) metrics. Experimental results demonstrated that the hybrid strategy achieved the highest answer faithfulness (0.712) and context recall (0.895), representing a 29.5% and 32.8% improvement in faithfulness over the respective standalone baselines, thereby ensuring superior factual consistency. Furthermore, the hybrid system recorded a Token F1 Score of 0.499, a BLEU score of 0.233, and a faster average response time of 7.64 seconds due to parallel query execution and context-size optimization. Finally, exploratory user evaluation yielded high satisfaction with an overall score of 4.46 out of 5.00, confirming its viability for real-world academic assistance.

Digital System and Computing

Galang Yanu Achmad Ramadan, Eko Sulistya

Flood disasters frequently cause significant socio-economic losses in developing countries, while many existing early warning systems remain costly, complex, or insufficiently accessible for real-time community use. This study proposes a low-cost IoT-based flood early warning system using a conductive K-0135 water level sensor integrated with a NodeMCU ESP8266 microcontroller and HTTP-based communication architecture. The novelty of this work lies in the use of a conductive sensor with systematic threshold characterization under both static and dynamic conditions to reduce false alarms while maintaining reliable detection performance. The methodology involved sensor characterization through controlled laboratory experiments, including static testing with 0.5 cm depth increments and dynamic testing simulating rainfall splashes. The results show a non-linear increase in sensor output with depth, ranging from 16.3 at 0.0 cm to 565.3 at 4.0 cm. Dynamic testing produced an average maximum output of 424.7, leading to an optimal detection threshold of 425. The integrated system achieved a communication success rate of 100% in delivering real-time alerts via HTTP requests to a web server and Telegram platform. An HTTP error code −11 was observed, corresponding to a timeout condition caused by network latency; however, this did not affect successful alert transmission. The findings are limited to controlled laboratory-scale testing and have not yet been validated under real environmental conditions. Overall, the proposed system demonstrates the feasibility of a low-cost, threshold-based IoT solution for real-time flood early warning applications and highlights its potential for improving community-level disaster preparedness.

Digital System and Computing

Adelia Clarissa, Anugrah Adiwilaga, Deden Pradeka

Saving behavior plays an important role in supporting early financial literacy and responsible financial management. However, conventional piggy banks still have limitations in transaction transparency, security, and real-time monitoring capabilities. This study aimed to develop and conduct prototype-level validation of an Internet of Things (IoT)-based smart piggy bank integrated with computer vision technology for automatic currency denomination detection. The study employed a Design and Development (D& D) approach combined with the Artificial Intelligence Life Cycle (AILC) for object detection model development and the Software Development Life Cycle (SDLC) for web-based monitoring system development. The dataset consisted of 9, 000 annotated images of Indonesian Rupiah banknotes and coins categorized into nine denomination classes under varying lighting and object orientation conditions. During training and validation, the YOLOv8 nano model achieved an mAP50 value of 0.995 under controlled conditions. However, real-world prototype testing produced an average operational accuracy of 62.2%. This performance degradation is primarily attributed to specular reflections on coin surfaces and edge-hardware-induced latency which compromised real-world inference stability. The main technical contribution of this work lies in quantifying this model-to-edge hardware performance gap, providing an empirical baseline for lightweight computer vision deployment on low-resource microcomputing nodes. These findings indicate a performance gap between controlled and real-world conditions. Overall, the findings demonstrate the feasibility of integrating lightweight computer vision and IoT technologies into an embedded smart saving prototype, although further optimization is required to improve operational robustness and detection stability in real-world environments.

Digital System and Computing

Anjelina Daima, Andi Rosano, Trisna Fajar Prasetyo

Open-field tomato cultivation is vulnerable to environmental fluctuations, yet manual monitoring often leads to inaccurate decisions. This study aims to design and validate a low-cost Internet of Things (IoT)-based soil quality monitoring prototype, addressing the gap in accessible real-time alert systems for small-scale farming. The system integrates a NodeMCU ESP8266 with soil moisture, temperature (DS18B20), air humidity (DHT11), and pH sensors, visualized via Blynk and Telegram. Conducted over 14 d with nine experimental units across three soil media (clay, sandy, and humus), the study focused on technical prototype validation. Results indicate the system monitored moisture levels (43–47%) and temperatures (≈30.3 °C) with high reliability. Automated irrigation activated at ≈60% and deactivated at 80% moisture, maintaining an uptime of ≥95%. Sensor verification showed temperature deviations below ±1 °C. Unlike existing greenhouse-centric models, this work implements a multi-parameter sensing framework tailored for open-field conditions using affordable hardware. While this study successfully validates the prototype's operational stability and data synchronization, it is primarily a technical verification; further research is required to evaluate agronomic impacts such as yield and water-use efficiency.

Digital System and Computing

Yurimasanti Rachman, Deden Ardiansyah

Gate security systems in residential areas often rely on manual controls or simple remotes, which limit flexibility, real-time monitoring, and secure access. This study developed an IoT-based automated gate control system using RFID authentication and a web-based interface. The system integrated Arduino Uno, an ESP8266 Wi-Fi module, an RC522 RFID reader, a relay, and a DC motor to automate gate operation, while user activity was monitored through a local web server built on Apache and MySQL. Testing was conducted under controlled indoor conditions with an average Wi-Fi signal strength of -62 dBm, an ambient temperature of 27°C, and a local network latency of 8 to 12 ms. Across 50 trials using five different RFID cards, the system achieved 100 percent reading accuracy, an average response time of 1.42 seconds from tag detection to motor activation, and stable communication with no packet loss. Mechanical implementation using 8.25 kg of galvanized steel and a dual-rail support system ensured stable and smooth gate movement. These results confirm that the system provides secure, contactless, and remotely accessible automated gate control. Future improvements should focus on cloud-based integration and enhanced network stability for real-world deployment.

Digital System and Computing