# Smart IoT-Biofloc water management system using Decision regression tree

**Authors:** Samsil Arefin Mozumder, A S M Sharifuzzaman Sagar  
**arXiv:** [2112.02577](https://arxiv.org/abs/2112.02577)  
**Date:** 2021-12-05  
**Summary page:** https://www.summarizepaper.com/en/paper/2112.02577v1/

## Key points

- Water contamination
- Temperature instability
- Nutrition issues
- Limited space
- High costs
- Converts leftover food into bacterial biomass
- Utilizes sensors to collect data on parameters like pH levels, water temperature, Ammonia levels, Total Dissolved Solids (TDS), and electrical conductivity
- Data stored in the cloud for real-time monitoring and analysis
- Decision regression tree model used to predict water conditions based on collected data
- Android app for real-time sensor data monitoring and manual actuator control
- Control procedure involves saving sensor data, sending it to IoT cloud for storage and monitoring, transmitting it to user app for visualization and manual actuator control
- Automatic adjustments based on deviations from optimal conditions:
- Actuators adjusted based on sensor readings if deviations detected
- Prototype components of the smart IoT Biofloc water management system:
- Water bulb (water heater)
- Pump (oxygen pump)
- Servo motor (cooling functionality)
- The study demonstrates a comprehensive approach towards improving fish farming practices through integration of IoT technology and machine learning algorithms in Biofloc systems. The proposed system shows promising results with a satisfactory accuracy rate of 79% during experimentation.

## Summary

The conventional fishing industry faces numerous challenges such as water contamination, temperature instability, nutrition issues, limited space, and high costs. To address these challenges, fish farming has adopted Biofloc technology which converts leftover food into bacterial biomass. This study focuses on developing an intelligent IoT Biofloc system to enhance efficiency and production in fish farming. The proposed system utilizes sensors to collect data on parameters like pH levels, water temperature, Ammonia levels, Total Dissolved Solids (TDS), and electrical conductivity. This data is stored in the cloud for real-time monitoring and analysis. A Decision regression tree model is employed to predict water conditions based on the collected data. An android app allows users to monitor sensor data in real-time and manually control actuators. The control procedure involves saving sensor data to relevant variables, sending it to the IoT cloud for storage and monitoring, and transmitting it to the user app for visualization and manual actuator control. Sensor data is also inputted into a classifier to predict the water condition of the Biofloc system. If any deviations from optimal conditions are detected, the system automatically adjusts actuators based on sensor readings. A prototype of the smart IoT Biofloc water management system was developed for demonstration purposes with components such as a water bulb representing a water heater, a pump simulating an oxygen pump, and a servo motor showcasing cooling functionality. A sensor arm equipped with TDS and temperature sensors is controlled by a servo motor for periodic cleaning of probes. Overall,this study presents a comprehensive approach towards improving fish farming practices through the integration of IoT technology and machine learning algorithms in Biofloc systems.The proposed system demonstrates promising results with a satisfactory accuracy rate of 79% during experimentation.