ReTSoSPA: IoT-based Real Time Soil Sensing for Precision Agriculture

In conventional agriculture, farmers sense their soil based on experience, past production and economic records, cost of production such as fertilizer and market demand. But all these factors might vary based on soil health of different locations, nutrient levels after each crop cycle, weather conditions, change in temperature and rainfall. Also the traditional way of farming lacks precision and sustainability resulting in over or under application of fertilizers. This research aims to improve agricultural practices by continuously monitoring soil nutrients, moisture level, soil temperature, pH, air humidity and temperature in real time. This system ensures cost and resource effectiveness, economical development, environmental sustainability and data-driven decision making. In this paper, we have developed a real time dataset from field level by establishing an IoT soil sensing hardware system with 5 sensors. The soil data from sensors are sent from a transmitter node to a receiver node and stored in the Cloud server through WiFi. We have trained Machine Learning algorithms for prediction of crops potato, tomato, and three classes of rice that consist of the 61 types of rice produced in Bangladesh at different times of the year. Based on the predicted crop, the optimal amount of fertilizer like Urea, TSP and MoP is recommended. Finally, we have built a Web application to help farmers monitor soil conditions remotely for real time prediction.