Enhanced Node Localization in IoT Sensor Networks via Deep Learning and Virtual Node Modeling

With the rapid increase in the deployment of Internet of Things (IoT) devices, numerous innovative and practical applications are being developed. IoT enables these applications through the use of resource-constrained fixed and mobile nodes. Using the whale optimization algorithm (WOA), we develop a deep learning technique that overcomes traditional DV-Hop localization challenges. In our approach, deep neural networks (DNNs) are used to estimate distances between nodes that have not yet been determined (non-coordinated) and anchor nodes that have been determined (coordinated) at low costs for IoT infrastructure. In contrast to DL methods that typically require substantial training data to achieve high accuracy, this problem can be solved by implementing a data augmentation strategy (DAS). DNNs are improved by creating virtual anchors around actual anchors to increase the dataset size and enhance their performance. Simulation results indicate that our deep learning-based DV-Hop localization method significantly surpasses other localization techniques, especially in terms of positional accuracy.