CWT-GRU-based Natural Disaster Prediction Framework using Wireless Sensor Networks

In order to identify earthquakes quickly, the disaster prediction model based on Wireless Sensor Networks (WSNs) tracks seismic vibrations as well as ambient parameters using dispersed wireless sensor nodes. However, challenges such as poor communication, high noise levels, energy constraints, as well as the intricacy of real-time information processing hinder accurate prediction as well as efficient disaster response. Hence, this paper performs the WSN-based Earthquake Disaster Prediction (WSN-EDP) model with the help of deep learning framework. The dataset is initially collected from the standard benchmark sources such as Kaggle called Earthquake dataset. The pre-processing of this data is accomplished by the Empirical Mode Decomposition (EMD) approach. The extraction of the features is further done by the Continuous Wavelet Transform (CWT) technique. Finally, the prediction of the WSN-EDP model is performed by the deep learning model known as Gated Recurrent Unit (GRU). The proposed CWT-GRU is 6.48% and 29.06% better than the other considered existing methods with respect to prediction accuracy and MSE respectively for the developed WSN-EDP model.