Smart Aquaponics with a Digital Twin
System design for an IoT- and AI-assisted aquaponics farm running on renewable power.
- Role
- Co-author (3rd of 4)
- Team
- 4 authors — IUB CSE
- Timeline
- Mar 2025 – Apr 2025
- Organisation
- Independent University, Bangladesh
- Code
- Private repository
- Reading time
- 1 min
A course research paper (3rd of 4 authors) designing ESP32 sensor nodes for water quality, a Firebase telemetry backbone, a React monitoring and control dashboard, a digital twin for scenario testing and LightGBM-based anomaly prediction.
Overview
Aquaponics couples fish farming with hydroponic plants, but water chemistry drifts quickly and failures are costly. The paper designs a system that watches the water continuously and warns before something goes wrong.
The problem
- pH, temperature and dissolved oxygen must stay in narrow ranges.
- Farms are often off-grid or on unreliable power.
Context & constraints
A system-design paper with use-case, sequence and deployment diagrams; headline performance figures in the paper cite prior work rather than our own measurements.
Approach
- Sensing — ESP32 nodes read pH, temperature and dissolved oxygen every five minutes.
- Backbone — telemetry lands in Firebase.
- Control — a React dashboard for monitoring and actuator control (pumps, aerators).
- Prediction — LightGBM models trained on historical sensor data flag anomalies and suggest maintenance through an assistant.
- Digital twin — simulates operating scenarios before changes reach the real tanks.
- Power — a solar-hybrid supply.
Key capabilities
- Five-minute water-quality telemetry
- Remote actuator control
- Anomaly prediction
- Digital-twin scenario testing
- Solar-hybrid power design
Architecture
Sensing & actuation
Cloud
Apps
Sensors to cloud to dashboard, with a model and a twin reading the same telemetry. Select a component to trace its connections.
Outcome
A complete system design submitted as course research; a natural next step is a physical prototype to measure the design against real tanks.
Lessons
- A digital twin is most useful as a safe place to be wrong.
Building something similar?
Tell me what you’re working on — I reply to every serious message.
Related work
Gesture-Based Messaging for Assistive Care
Two ESP32 units over Bluetooth Classic: a wearable sender with an MPU9250 motion sensor and a receiver with an OLED and buzzer. Reliable at 10–20 metres.
- C / C++
- ESP32
Physics-Guided ML for Fabric GSM
Research paper accepted at IEEE ICACT 2026. On 5,648 production samples, an Extra Trees model reached a test RMSE of 4.10 g/m² (R² 0.952, MAPE under 0.5%) after adding a physics-derived tightness factor; SHAP and LIME confirmed the model learned textile physics.
- Python
- scikit-learn
- Explainable AI (SHAP)
- pandas & NumPy