Hybrid Optimized Deep Neural Network-Based Intrusion Node Detection and Modified Energy Efficient Centralized Clustering Routing Protocol for Wireless Sensor Network

Abstract

As Internet of Things (IoT) technologies advance, applications such as smart cities, healthcare, and smart grids will become increasingly commonplace. A wireless sensor network (WSN) is one of the futuristic technologies used in IoT-enabled applications for sensing and data transmission. An IoT-enabled WSN (IWSN) is characterized by several sensors dispersed randomly in open and harsh environments. Given the resource constraints of sensor nodes (SNs) and the hostile deployment environments, designing routing protocols for WSNs necessitates a focus on energy efficiency and security. An optimized hybrid model, Hybrid Optimized Deep Neural Network (HODNN), is designed using Deep Neural Networks (DNNs) to maximize its detection accuracy. The source node determines the shortest path to the destination after detecting malicious nodes and performs secure routing without malicious nodes. A modified energy-efficient centralized clustering routing protocol determines the optimum path for routing data in the proposed model (MEECRP). The paper presents HMRP-IWSN, HODNN-based intrusion detection and MEECR protocol for securing IWSN data. Through comprehensive evaluation using various performance metrics, HMRP-IWSN demonstrates superior outcomes compared to existing methods, including a higher packet-delivery ratio (PDR), detection rate, lower delay and energy usage, and an extended network lifespan.