Rider Water Cycle Optimization-Based Hierarchical Attention Network and Spark Architecture-Aware IoT System for Crop Yield Prediction

The loss of crop yield is a common problem that still exists among farmers. The crop yield can be maximized by the right selection of crops and this can be done by considering the meteorological factors and doing soil analysis. The main reason for low crop production is the lack of knowledge about crop selection and soil fertility. Recently, the modern agriculture industry has been smarter, more precise and more data-centered than ever. The “smart agriculture” system is popular due to the advanced growth of Internet-of-Things (IoT) and this provides an innovative and profitable farming system by increasing crop production thereby reducing the wastage of irrigation. This research intended to make an accurate and efficient system with the help of IoT and Deep learning algorithms for the selection of the right crop for obtaining maximum crop yield. Hence, this research developed an efficient scheme called the proposed Rider Water Cycle Optimization-based Hierarchical Attention Network (RWCO-based HAN) for predicting crop production. Moreover, the Water Cycle Algorithm (WCA) is integrated with the Rider Optimization Algorithm (ROA) for devising the proposed RWCO. Here, the prediction process is achieved with the spark architecture through the process of Cluster Head (CH) selection and routing. The nodes carry the crop data from the distributed network domain and transfer the information to the sink node or Base Station (BS) through optimal CH such that CH is chosen with the RWCO algorithm. The metrics Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), throughput, energy and delay are used for evaluating the performance of the devised method and obtained values of 0.107, 0.327, 12.6, 0.5435, 0.5431[Formula: see text]J and 0.4597[Formula: see text]s, respectively.