Deep Learning-Based Reduction of Computational Overhead and Energy Consumption in IoT-Assisted WSNs

Localization of Internet of Things-assisted Wireless Sensor Networks has gained significant interest from the research community. The key issues of energy consumption and computational overhead in existing studies are addressed by proposing a deep learning-based method to reduce both in the Internet of Things-assisted Wireless Sensor Networks presented here. Network topology is built using the topographic network framework to optimize energy use and computational efficiency. Anchor node deployment is performed using the Upgraded Butterfly Backtracking Optimization (UBBO) algorithm. Pilot agents are responsible for clustering sensor nodes with the Enhanced Density Peak Clustering Algorithm (E-DenPeC), which solves device heterogeneity and orientation issues through bias regression. For clustered sensor nodes, localization is carried out using the Squeeze-Excitation Network embedded Skip Connection Gated Recurrent Unit and Accelerated Iterative Algorithm (SEN-GRU-AI), based on five different measurement techniques (FLMT). Finally, collaborative localization is achieved through deep learning technology by reducing energy consumption and computational overhead. The results show that the proposed method surpasses existing approaches.