A Adaptive Dilated Cascaded Recurrent Neural Network Framework for Energy-Efficient Routing in Wireless Sensor Networks

Abstract

Energy-Efficient Routing in Wireless Sensor Networks is crucial for extending network lifetime and guaranteeing dependable communication, especially in large-scale battlefield environments. Wireless Sensor Networks (WSNs) are made up of resource-constrained sensor nodes that communicate thru multi-hop paths and have limited battery capacity. The limited energy of sensor nodes is frequently overlooked by traditional routing protocols. Energy histograms are used to create a workable energy-aware routing method in order to overcome this difficulty. Outlier detection and classification are preceded by data preprocessing methods such as data cleaning, shuffling, and normalisation. To get the best chi-squared value, a ADRNN is combined with a chi-squared filter-based feature-selection technique. The best routing paths for effective data transfer are then found using ADRNN. The conveyed information is further analysed thru an Adaptive Dilated Cascaded Recurrent Neural Network (ADRNN) for the purpose of energy-efficient routing, whereas ADRNN fine-tunes the parameters of the ADRNN model. The suggested ADRNN framework achieves a high classification accuracy of 94.08% by successfully integrating intelligent energy-efficient routing, energy-aware routing, and feature selection. The findings show that in energy-constrained wireless sensor network situations, the suggested ADRNN framework can enhance communication effectiveness and offer precise illness diagnosis.