Abstract: A vast number of IoT devices have been fabricated or adapted into different aspects of smart and precision farming to carry out a variety of tasks one of which is soil nutrients detection. While performing these tasks that are often recurrent, these devices generate different datasets which are stored on local memories or communicated remotely to cloud servers. The analysis of these data is important in order to correctly classify and group such data for device identification and differentiation. This is very important because the productivity of crop yields has greatly reduced due to lack of knowledge of the appropriate nutrients in a particular soil. Our research focuses on Nitrogen, Phosphorus, and Potassium, for the fact that most inorganic fertilizers consists majorly of these. As such, in this paper, soil nutrients values (Nitrogen, Phosphorus, and Potassium) are used as input features into the neural network for the classification of IoT-enabled soil nutrients data. Experimental analysis proved that the classification of these soil samples based on nutrients can achieve good accuracies between the range of 81.33% to 97.13%.

Keywords: IoT; Agriculture; Artificial Neural Networks

 

| DOI: 10.17148/IJIREEICE.2020.8418