An efficient disaster management system based on deep learning in bio-inspired wireless sensor network

Abstract

The risk rates, such as death, injuries, and distress, are increased in natural and artificial disasters. So, timely control and management of such disasters is a necessary process. The Wireless Sensor Network (WSN) plays a vital role in this process. However, in such techniques, prediction and communication sometimes fail. Therefore, a novel Jellyfish-based Deep Neural Disaster Management (JbDNDM) framework was developed for disaster management phenomena to address these limitations. Initially, the required nodes and network are initialized in the monitoring regions. The proposed JbDNDM system is activated at the sensing module of the network to sense the building fire and the affected zone based on parameters such as smoke, temperature, and gas mixture using the fitness function of the jellyfish and estimating the affected people. The sensed information is then sent to the management to provide emergency favors. Furthermore, the proposed JbDNDM approach was implemented in the MATLAB tool with several performance measurements such as delay, throughput, network lifetime, sensing accuracy, and Packet Delivery Ratio (PDR). The utilization of multi-relay communication increased the network performance, and the jellyfish function increased the sensing accuracy. The network efficiency results were compared with the existing techniques, such as OECF, HMRN, and WWO. The Proposed network obtained 0.01ms delay, 80.01Kbps throughput, 99.7% PDR, and 75h network lifetime. The sensing accuracy of the model is 99%.