Cloud-Centric IoT-Based Green Framework for Smart Drought Prediction

Abstract

Drought is a catastrophic natural disaster with significant impact on financial stability, hydrological budget, public health, and agricultural productivity. Numerous drought indices have been introduced to quantify the severity of droughts, but the majority of them are incapable of demonstrating the changes in crucial drought causing elements. Internet of Things (IoT) is appropriate to monitor time-critical environmental parameters. This article proposes an energy-efficient cloud-centric system to assess the drought for the current situation and predict for the future time frame. The architecture determines the active and sleep interval of IoT sensors based on the analysis of data variability using the Bartlett test. The dimensionality of the data about drought causing elements is reduced using kernel principal component analysis (KPCA) at the fog layer. The intensity of drought is determined at the cloud layer using naive-Bayes classifier, and drought severity for different time periods is predicted using the seasonal autoregressive integrated moving average model. Experimentation and performance analysis prove the efficiency of the proposed system in assessing and predicting the drought with better correlations with drought-causing attributes. Furthermore, it shows significant energy savings as compared to other schemes.