Autoregressive Diffusion Networks for Trustworthy Node Identification and Optimized Routing in Wireless Sensor Networks

Finding and eliminating unwanted nodes is a foremost challenge in wireless sensor networks (WSNs). This paper proposes an Autoregressive Diffusion Network (ADN) to find and categorize the nodes in the network as affected or trustworthy. Here, the proposed ADN provides a strong foundation for finding uncommon or insecure actions by effectively recording the characteristics of the sensor over a period of time. Thus, using the capability to capture successive data. In addition, an improved routing mechanism is developed using the Low-Energy Adaptive Clustering Hierarchy (LEACH) protocols. Then, the Pelican Optimisation Algorithm (POA) is used to optimize the cluster heads optimally. To further improve the routing mechanism using fitness proportional selection technique to transfer the data safely and effectively. Simulation results show that the proposed technique enhances the performance by reducing energy consumption and increasing the detection accuracy of affected nodes. The proposed LEACH + POA + ADN model shows the improved detection accuracy of 77%, false positive rate of 7.50%, average energy consumption of 8.02 J, network lifetime of 92.94 rounds, and packet delivery ratio of 92.97%, which is high compared to other benchmark models.