Deep Learning-Infused Hybrid Security Model for Energy Optimization and Enhanced Security in Wireless Sensor Networks

Abstract Many wireless sensors are placed in an ad hoc way to create a wireless sensor network (WSN), which is used to monitor system, physical, and environmental conditions and transmit the collected data to a centralized point. Base stations and several nodes (wireless sensors) make up the system. The base station of a WSN System is connected to the Internet to share data, and these networks are used to cooperatively transfer data via the network to the base station while monitoring physical or environmental factors like sound, pressure, and temperature. These data can be processed, analyzed, stored, and mined by WSN. In this study, additional optimization and a deep learning approach were used to separate a rogue node from the network's busiest node based on a variety of criteria. A deep learning model for identifying the malicious node has been offered as a solution to these challenges. This model works by computing the probability of request forwarding, reply forwarding, and data dropping in a sum-rule weighted method. It has been determined that the planned task would have both a high throughput and a decreased necessary amount of time. There has been a decrease in the overall rate of packet loss. There has been a drop from 70ms to 42ms in the delay-related hyper metrics. There has been a near threefold reduction in the percentage of missing packages, from 23–8%. The adoption of deep learning has removed hostile node behaviour that might bring down a network as a potential failure mode.