An energy efficient data fault prediction based clustering and routing protocol using hybrid ASSO with MERNN in wireless sensor network

Abstract

Wireless sensor networks (WSNs) and Internet of Things (IoT) are essential for numerous applications. WSN nodes often operate on limited battery capacity, so energy efficiency is a significant problem for clustering and routing. In addition to these limitations, one of the primary issues of WSNs is achieving reliability and security of transmitted data in vulnerable environments to prevent malicious node attacks. This work aims to develop a secure and energy-efficient routing protocol for fault data prediction to enhance WSNs network lifespan and data reliability. The proposed technique has three major phases: cluster construction, optimal route selection, and intrusion detection. The adaptive shark smell optimization (ASSO) technique was initially used with three input parameters for CH selection. These parameters are the residual energy, the distance to the BS, and the node density. After clustering, salp swarm optimization (SSO) is used to select the optimum path for data transmission between clusters, resulting in an energy-efficient WSN. Finally, to ensure the security of cluster-based WSNs, an effective intrusion detection system based on a modified Elman recurrent neural network (MERNN) is implemented to detect the presence of intrusions in the network. The experimental results show that it outperforms the competing methods in various performance metrics. The performance results of quality of service (QoS) parameters are expressed as dispersion value (0.8072), packet delivery rate (98%), average delay (160 ms), network lifetime (3200 rounds), and the accuracy of this method is 99.2%. Compared to the SVM, ELM, HMM, and MK-ELM protocols, the proposed protocol increases network lifetime by 77%, 60%, 45.4%, and 14.2%, respectively.