Optimized anomaly detection for soil health monitoring using a cosine sculptor efficient pyramid network

Soil health monitoring and management support farmers in optimizing soil enrichment and irrigation, and reducing the impacts on the environment. Nevertheless, long-term agricultural practices impact soil health, yet recognizing assessable soil features that truly reflect soil remains a challenge. Therefore, an advanced technique called Cosine Sculptor Optimization Algorithm Efficient Pyramid-Network (Cosine-SOA_EPy-Net) is designed to monitor soil health in the Internet of Things (IoT). Initially, the IoT simulation is carried out; later, the routing and Cluster Head (CH) selection are done by Cosine-SOA to transfer the soil data to the base station. Following this, at the base station, the monitoring of soil health is executed by considering the soil data. Then, the input data is passed to outlier detection and removal using weighted holoentropy, and thereafter, data normalization is done by median normalization. At last, the soil health classification is done by the EPy-Net. Moreover, the classification performance of EPy-Net is boosted by optimally adjusting the hyperparameters using Cosine-SOA. The experimental results demonstrates that the Cosine-SOA_EPy-Net achieved True Negative Rate (TNR), True Positive Rate (TPR), and accuracy of 96.569%, 94.999%, and 95.578% on the SoilHealthDB dataset. Additionally, the Cosine-SOA attained a value of delay, energy, and distance of 0.371 s, 94.105 J, and 46.554 m using the OSMO soil health dataset.