Hybrid Routing Identification in Wireless Sensor Networks using Convolutional Neural Networks and Random Forest

Numerous societal domains have benefited from the extensive deployment of Wireless Sensor Networks (WSNs), such as educational institutions, agriculture, manufacturing facilities, environmental monitoring, and many more. Unfortunately, because the sensor nodes are wireless, it is not possible to change their batteries while are positioned in an unattended or isolated area. Consequently, a great deal of research has been documented with the aim of increasing the node's longevity. While cluster-based routing has solved this issue to a large extent, it may be even better if key considerations were taken into account while selecting the cluster head (CH). Preprocessing, feature selection, and model training make up the suggested system. In the first of two phases of data preprocessing, and check the data for errors and completeness; if any are present, then fill in the gaps. Repairing out-of-the-ordinary data should follow the second step of locating the incorrect data. With principal component analysis (PCA), we can minimize the dataset's dimensionality while preserving the majority of the original variability; this is achieved during feature selection. It used the CRFB model for training, and the resulting model had an impressive accuracy of 92.32%.