IoT Based Automated Solution for Carp Fish Farming
using Sensor Networks and Biochemical Parameters

In Sri Lanka, people use fish as their main protein food resource. Sri Lanka has a total of 1700 kilometers of coastline, with an estimated 121,000 hectares of lagoons and estuaries [1] But we can see that saltwater fish prices are very high these days. So, lots of people in Sri Lanka didn’t buy fish products because of the high cost and they couldn’t afford to spend that much of money because the economic crisis. They didn’t have to way to get protein in their day-to-day meals. The government and NAQDA now have an eye on developing inland aquaculture. Previously in the inland fish industry farmers sold the wild-caught that they caught from natural ponds. So only the village-based families got a chance to eat those freshwater fish species. But now the government and NAQDA have built inland aquaculture fish farms all around the country so that lots of people can get freshwater fish species as their food resource. Carp fish species are the main species that Sri Lankan farmers use to breed on their farms. We identified that one of the major problems in those farms is managing and detecting water quality after fish feeding. In NAQDA farms they use larger ponds, medium ponds, small ponds, and tanks for fish breeding. One of the major challenges of this fish framing is maintaining the water quality to have healthy fish and get maximum output of fish-based products to local and foreign markets[3].

Min, Wei, Showkat and Nen-Fu [7] built an IoT and AI-based surrogate model-based smart aquaculture management system. Sensors built into the system is managed by an Arduino Mega2560. They link the various feeding system parameters using a deep learning model to forecast Californian bass fish development. To remotely monitor and manage the system, a mobile application has been created. To determine the input and output variables with the greatest influence on the model's ability to forecast the future, a correlation study was carried out.

Research Problem

How to manage Carp fish farm water conditions and prevent diseseses using IoT based System?

One of the major challenges of this fish framing is maintain the water quality to have healthy fish and get maximum output of fish based products to local and foreign markets.[2] As per the data we gathered from the NAQDA farmers the main reason for water quality problems is providing fish pallets and other feeding items more than the fish needed. Simply overfeeding and fish wastage. So, because of overfeeding and fish wastage the water parameters of fish tank/pond Container become unlivable for Carp fish. Decreased oxygen levels can lead to stress and death of fish. Toxicity can damage to their health and fish may not grow well in those condition and also can create an environment that is conductive to the growth disease. Causing pathogens which can leads to outbreaks of disease to carp fish. Argulus, Tricodina, Aerominia, columnaris, gnteric, redmouth disease carp edma virus, spring viraernia of card are some fish diseases that spread because of bad water quality conditions. Still the farmers did not have a way to identify these diseases early. So, they can start took necessary actions when the fish started to show the symptoms. So, at that time it would be too late and they can lose their entire fish aqua culture.[3]

The proposed IoT system and mobile application consists of 5 main components. They are;

1. Sensor data gathering (pH, temperature, turbidity)
2. Predict current water quality parameters and future water quality paramete forecasting.
3. Identify Chemicals that need to bring water pH value to optimal level and chemical calculation
4. Early disease risk identification, disease risk forecasting and disease data visualization
5. Identify carp fish feeding behaviours

The proposed solution provides a smart approach for fish farmers, researchers, and industry specialists to detect water quality conditions and identify the risk of diseases. As shown in Fig. 3, the device contains an ESP32 module, pH sensor, and turbidity sensor. The temperature of the water controls every aspect of the fishpond's operation. Along with regulating the growth and development of the pond's vegetation and other animals, it also regulates the amount of oxygen in the water. The optimal temperature for river-dwelling fish, such as trout, is around 14°C. With a 2°C allowed range, 25°C is the best temperature for tropical fish. 20 to 25°C is the ideal water temperature for carp fish growth and reproduction [12]. We used an ESP32 Microcontroller as the main board in the system we designed. It is a Wi-Fi and Bluetooth-integrated low-cost and low-power chip microcontroller [15]. We used the DS18B20 Temperature probe to keep track of the water's temperature. It uses a single wire bus for communication and requires a power supply ranging from 3.0V to 5.5V. operational temperature range is between -10°C and 85°C with an accuracy of +/-0.5 °C.[13]. The turbidity sensor need 5V DC to operate. The response time is roughly 500ms, and the operating current is 40mA. It provides both digital output and analogue output between 0 and 4.5V. The adapter's dimensions are 38mm*28mm*10mm/1.5inches *1.1inches*0.4inches, and it can function between 5°C and 90°C [14]. To measure the pH value used a pH probe sensor. The module power is 5.00V, size 43mmx32mm, measuring range 0-14ph and temperature range is 0-60°C accuracy is ± 0.1pH (25 ℃) and the response time is ≤ 1min. To measure the pH value used a pH probe sensor. The module power is 5.00V, size 43mmx32mm, measuring range 0-14ph and temperature range is 0-60°C accuracy is ± 0.1pH (25 ℃) and the response time is ≤ 1min.

Three sensors temperature, turbidity and pH sensor were used to get the real-time data. We've also created a mobile app for React Native that allows farmers to keep an eye on the relevant water quality metrics detected by smart ponds or tank sensors. The monitored data is directly stored in a cloud database and the app reads the data and shows the real-time data on a dashboard. The main purpose of the system is to predict future water conditions, diseases and future diseases and notify the carp fish farmers. The results of the developed machine learning models learning process highlight its prediction accuracy. Temperature, turbidity, and pH sensor were used to get the real-time data. We've also created a mobile app for React Native that allows farmers to monitor the relevant water quality metrics detected by smart ponds or tank sensors. The program reads the monitored data from a cloud database and displays the real-time information on a dashboard. Utilizing real-time farm information enables you to identify problems before they materialize because the projection is based on the most recent data. The evolution of one parameter and the evolution of external factors might be related. We have utilized the ML model to estimate the major variables influencing the development of the smart fishpond based on the real-time observations of a variety of data. Numerous studies have shown that water temperature is one of the most crucial factors affecting water dissolved oxygen levels. Several factors, including the kind of aquaculture, have a significant impact on the dissolved oxygen levels in aquaculture systems. Depending on the nature of production, these factors' impact varies. The majority, however, may be measured and taken into account by an Internet of Things-based prediction model. Some of these parameters include the feeds' chemical composition and nutritional content. Additionally, it's important to take into account the length, feed level, and biomass of sh as well as algal biomass in fish systems.