Get in TouchJuly 15th, 2025
SIGMA is a IoT & web-based closed-house poultry monitoring system made to support poultry owner to monitor their coop especially for chicken farm. This system allows users to monitor parameters inside the coop e.g. ammonia, humidity, and temperature level.
SIGMA is a Capstone project of my team (Me, Farrel, and Rayhan) as our thesis to finish our bachelor degree in Computer Engineering. SIGMA is an IoT based poultry monitoring system built with various modern technologies to help local farmers in Boyolali to monitor their chicken coop online. With modern technologies such as Next.js, Typescript, Tailwind CSS, Django, ESP32, DHT22, and DFRobot MiCS-5124, this system allows users to monitor parameters inside the coop e.g. ammonia, humidity, and temperature level. User also able to manage the chicken data such as chicken amount, mortality rate, and production batch management.
Addition: This project was funded by Faculty of Engineering, Universitas Diponegoro, Indonesia through Strategic Community Engagement Grant 2025.
You can visit the webpage here, though you might not be able to log in since the credentials are required and only available for the farmers.
Modern poultry farms now use closed-house systems to maintain stable air quality and optimal temperatures for the chickens. However, closed-house systems trap ammonia, leading to health issues and reduced productivity. Additionally, temperature and humidity levels require careful monitoring. Therefore, this project aims to develop an integrated monitoring system for ammonia levels, temperature, and humidity using ESP32 technology and a web-based platform using React to help farmers monitor their farms online and accessibly on all devices.
This research was developed using the Agile methodology. The system integrates ESP32 as a microcontroller, DFRobot MiCS-5524 as an ammonia sensor, and DHT22 as a temperature and humidity sensor, with a PostgreSQL database framework using Django. Next.js serves as the web framework, receiving data from the database API and displaying it on the user interface. Black-box testing confirmed that all website features and functions operate correctly without critical errors. Usability testing demonstrated a smooth user experience in navigating pages and accessing features. Performance testing resulted in a good overall score across all web page performance metrics. Overall, this closed-house poultry farm monitoring system effectively assists farmers in managing and monitoring their farms.
The core objective was to enable the collection, transmission, and visualization of key environmental parameters. They are temperature, humidity, and ammonia concentration alongside poultry management data such as mortality rates and batch scheduling. Agile methodology is used to ensure iterative improvements based on continuous user feedback. The cycle began with planning, where system needs were identified through participatory discussion with local farmers, collect data from various scientific papers, and the system requirements. During the design phase, prototypes were co-created with end-users to prioritize intuitive interfaces, later validated through usability testing via SUS surveys and Maze task analyses. The development phase incorporated performance optimizations (e.g., Google Lighthouse and GTmetrix audits) to guarantee speed and accessibility, while review sessions with farmers refined features like poultry data management. The development phase is also done with an iterative cycle, thus not limited to a linear progress. Finally, the system was deployed and launched in phases, allowing for incremental adjustments based on field usage.
To achieve the objective, we conducted several methods for each part of the development based on agile methodology:
The picture below shows the chart of software process using Agile methodology.
The hardware system uses an ESP32 microcontroller with Wi-Fi connectivity, chosen over the ESP8266 for its capabilities, along with a DHT22 sensor for temperature and humidity and a DFRobot MiCS-5524 for ammonia detection, sending data to a backend server every five minutes via HTTP requests. The sensors are tested through accuracy precision testing. The sensors were turned on for a lengthy period of time to detect its surrounding area. This testing is repeated many times in the same place to collect as much data as possible to see the stability of the sensor and its precision in a controlled environment. Based on Desnanjaya et al. (2022) and initial testing, it is observed that DHT22s are quite stable and can measure accurately. Meanwhile, the DFRobot MiCS-5524 sensor readings have some noise in them. To combat that MiCS-5524 reads and prints data every 2.5 seconds, but in that 2.5 seconds, every 0.25 seconds, we read and sample the ammonia values and average the values so that the final data readings every 2.5 seconds are muted.
On our hardware system, to evaluate our sensors measurement we use the accuracy precision testing method. To do this we have 2 sets of sensors, each set consist of MiCS-5524 and DHT22. Those 2 sets of sensors are turned on in the same controlled area. Those collected data are then cleaned and processed to a graph that shows the sensor readings and will be displayed on the graph section. Alongside that precision test we also need to know if the hardware is capable of its functionality.
The ESP32 transmitter equipped with a DHT22 sensor is capable of sending data when it is powered on. Similarly, the ESP32 receiver with the Dfrobot MiCS-5524 sensor can receive data when active and is also capable of sending all the collected data to the API. The sensors on both the transmitter and receiver are able to read and measure the required data effectively. The transmitter’s ESP32 can display its data output on the serial monitor, while both ESP32 modules can connect to the internet and use an API token for authentication. Furthermore, the ESP32 transmitter and receiver can successfully communicate with each other, and the API is capable of receiving data from the receiver ESP32.
The backend, which is REST API-based and built with Django and PostgreSQL, stores sensor data, calculates scores based on thresholds, and manages poultry data like mortality and scheduling, accessible via a web interface. Before deployment, the system will be load-tested using Postman’s Collection Runner tool based on a similar study conducted by Adrianto and Suyatno (2024). The metrics will be evaluated are throughoutput, total request, average response time, and error rate. In this study, we load test the system with 5 virtual users accessing the backend without delay for 5 minutes with 5 repeated runs, then we average every metric with this formula.
On our backend system, we evaluate the performance of the system by conducting a load test using the postman performance collection tool.
Based on the test results, a total of 13,302 requests were sent to API. Throughput was measured at a rate of 20.402 requests/second, average recorded response time was 104.2 ms. The average error rate is 13.07%.
The web application, developed with Next.js for the frontend and Django for the backend, fetches data through API calls, providing a user-friendly interface for data visualization and management, improving upon previous systems developed by Chigwada et al. (2002) that used PHP, WampServer, MySQL, or Flutter. Multiple testing approaches were employed to ensure the system’s quality and reliability, including black-box testing, usability testing, unit testing, integration testing, performance testing, and accessibility evaluation. This monitoring system aims to provide poultry farmers with a practical and effective tool to enhance their farm management practices, ultimately improving broiler chicken productivity and welfare.
Performance metrics play a critical role in assessing the usability and accessibility of web-based systems. Based on Dulaimi et al. (2022), which utilized GTmetrix for comparative validation, our study adopts a dual-method approach with combining GTmetrix and Google Lighthouse to evaluate key performance indicators (KPIs) such as Largest Contentful Paint (LCP), First Input Delay (FID), and Cumulative Layout Shift (CLS).
Web pages contain components to display the parameters and coop data retrieved from API. Pages need a good performance for enhancing accessibility and usability for the users.
We evaluated dynamic pages critical to end-users (e.g., dashboard, data visualization, hardware controls) using Google Lighthouse and GTmetrix. Results demonstrate good performance, with most pages scoring >90% overall, underscoring the system’s efficiency and accessibility.
Google Lighthouse Total Result:
GTmetrix Total Result:
Black box testing results summarized in picture below, indicate that all tested features functioned according to expectations. This included successful role-based login, proper data display by coop floor, accurate device status monitoring, seamless notification delivery, and the correct calculation and display of chicken mortality and performance metrics. Moreover, the ability to download data in accessible formats (e.g., .xlsx files) supports transparency and offline record to keep an important factor for communities with inconsistent internet access.Black box testing results summarized in Table 5, indicate that all tested features functioned according to expectations. This included successful role-based login, proper data display by coop floor, accurate device status monitoring, seamless notification delivery, and the correct calculation and display of chicken mortality and performance metrics. Moreover, the ability to download data in accessible formats (e.g., .xlsx files) supports transparency and offline record to keep an important factor for communities with inconsistent internet access.
Usability testing is a critical process in assessing an application’s effectiveness, efficiency, and user satisfaction by engaging participants in task-based evaluations (Weichbroth, 2024). Building on this, Jannah et al. (2022) employed the WebUse (Website Usability Evaluation) method, which relies on user feedback through structured questionnaires to measure user experience. WebUse uses learnability matrix which is considered as qualitative approach while system performance needs numerical matrix or quantitative approach to evaluate each unit in detail.
To ensure the usability of the system and enhance user experience, we adopted the System Usability Scale (SUS), a rapid and user-friendly survey tool that captures usability quality from an end-user standpoint (Aziz et al., 2021). The usability test was done by having the developers explain the system and its overall features to the respondents, then allowing them to freely use the website for 30 minutes. Google Form was given to respondents afterwards.
Participants interacted with the system for 30 minutes, after which they completed a Likert-scale (1–5) survey. Odd-numbered statements contributed a score of (response – 1), while even-numbered statements used (5 – response). The total score was multiplied by 2.5 to generate a final SUS score (0–100). Ten respondents participated, yielding an average SUS score of 89.5, classified as "Excellent" on the acceptability scale (Vlachogianni et al., 2022) This outcome reflects strong community-endorsed usability, with users affirming the system’s intuitiveness and feature adequacy. Picture below shows the calculated total score of SUS questions given.
In the detailed results, the calculated total scores across the ten questions ranged from 34 to 38, resulting in a combined total of 358. After applying the SUS multiplier, the overall score reached 895, leading to an average of 89.5. Based on the results, the interpretation can be made using the grade acceptability range, grade scale, and usability level to provide clearer insights. In the case of moderated testing, the average SUS score achieved was 89.5, which falls within the acceptable range. According to the grade scale, this score is classified as an A, indicating an excellent usability level. The high SUS score underscores the value of inclusive design practices and direct user feedback in developing accessible tools. Looking ahead, future work will focus on scaling participatory evaluations to broader community groups, ensuring that the technology continues to align with real-world needs.
The SIGMA system was introduced to address inefficiencies in traditional poultry farming, where manual monitoring of coop conditions (temperature, humidity, livestock health) often led to delays, errors, and economic losses. Through tutoring, a few farmers were trained to use SIGMA’s dashboard, data visualization tools, and automated alerts, marking a shift from guesswork to data-driven decision-making. Post-implementation surveys revealed that most of users found the system intuitive, with an average usability score of 89.5 (rated “Excellent” on the System Usability Scale), underscoring its accessibility even for non-technical users. Farmers particularly valued real-time parameter monitoring, which allowed them to track coop conditions remotely and reduce physical checks, repurposing saved time for feed optimization and market negotiations. Automated data logging eliminated manual record-keeping errors, while alert notifications enabled proactive responses to equipment failures or environmental risks (e.g., overheating).
The project’s outcomes align with broader studies on technology adoption in agriculture, where user-centric design and iterative feedback determine long-term viability. For instance, Maze task metrics showed a 92% success rate in data management after interface refinements, reflecting the importance of adapting tools to local contexts. Moving forward, scaling SIGMA’s impact will require addressing connectivity gaps through offline modes, securing cooperative funding for hardware maintenance, and establishing peer-mentorship programs to sustain knowledge transfer.
The community empowerment initiative achieved significant success by leveraging participatory usability testing and performance optimization to address real-world challenges faced by farmers. The System Usability Scale (SUS) results (average score: 89.5) confirmed the system’s ease of use, while task-based evaluations using Maze demonstrated high success rates (92–100%) for critical functions like poultry data management. Performance audits via Google Lighthouse further validated the system’s accessibility, with most pages scoring above 90%, ensuring fast load times and stability even in low-connectivity areas. These outcomes highlight how user-centered design and community feedback can bridge digital literacy gaps and enhance practical usability.
The project’s impact extends beyond technical metrics, fostering confidence and efficiency among users while aligning with local needs. However, sustaining this empowerment requires ongoing engagement. Future efforts should prioritize localized training, iterative co-design workshops, and infrastructure improvements to ensure long-term adoption. By maintaining a participatory approach, this initiative can serve as a scalable model for technology-driven community development in similar agricultural contexts.