Fog-assisted Energy Efficient Cyber Physical System for Panic-based Evacuation during Disasters

Abstract

Abstract Disasters around the world have adversely affected every aspect of life and panic-health of stranded persons is one such category. An effective and on-time evacuation from disaster-affected areas can avoid any panic-related health problems of the stranded persons. Although the nature of disasters differ in terms of how they occur, the evacuation of stranded persons faces approximately same set of issues related to the communication, time-sensitive computation and energy efficiency of the devices operated in the disaster-affected areas. In this paper, a cyber physical system (CPS) is proposed that takes into account various challenges of the disaster evacuation, so an efficient on-time and orderly evacuation of stranded panicked persons could be realized. The system employs fog-assisted mobile and UAV devices for time-sensitive computation services, data relaying and energy-aware computation. The system uses a fog-assisted two-factor energy-aware computation approach using data reduction, which enables the energy-efficient data reception and transmission (DRecTrans) operations at the fog nodes and compensates to extend the period for other functionalities. The data reduction at fog devices employs Novel Events Identification (NEI) and Principal Component Analysis (PCA) for detecting consecutive duplicate traffic and data summarization of high dimensional data, respectively. The proposed system operates in two spaces: physical and cyber. Physical space facilitates real-world data acquisition and information sharing with the concerned stakeholders (stranded persons, evacuation teams and medical professionals). The cyber space houses various data-analytics layers and comprises of two subspaces: fog and cloud. The fog space helps in providing real-time panic-health diagnostic and alert services and enables the optimized energy consumption of devices operate in disaster-affected areas, whereas the cloud space facilitates the monitoring and prediction of panic severity of the stranded persons, using a conditional probabilistic model and seasonal auto regression integrated moving average (SARIMA), respectively. Cloud space also facilitates the disaster mapping for converging the evacuation map to the actual situation of the disaster-affected area, and geographical population analysis (GPA) for the identification of the panic severity-based critical regions. The performance evaluation of the proposed CPS acknowledges its Logistic Regression-based panic-well being determination and real-time alert generation efficiency. The simulated implementation of NEI and PCA depicts the fog-assisted energy efficiency of the DRecTrans operations of the fog nodes. The performance evaluation of the proposed CPS also acknowledges the prediction efficiency of the SARIMA and disaster mapping accuracy through GPA. The proposed system also discusses a case study related to the pandemic disaster of coronavirus disease 2019 (COVID-19), where the system can help in panic-based selective testing of the persons, and preventing panic due to distressing period of COVID-19 outbreak.