Energy-Efficient IoT Network Routing Model Based on Multi-Layer Clustering and Deep Learning

The increasing numbers of sensors or actuators in the Internet of Things (IoT) networks make it essential to develop smart routing solutions whose goals are to extend the network's lifetime and keep reliable communication.In this paper, we present a new hybrid routing model based on multi-layer hierarchical clustering with the hybrid Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU) model to find the most reasonable routing paths in a dynamic IoT environment.The CNN-GRU model utilizes a sequential sliding window of historical node attributes residual energy, hop count, and data rate for predicting the best energy-efficient next-hop node as closely as possible in pipeline.This predictability allows the protocol to rely less on static heuristics, and to adapt its behavior to varying network conditions.Observations from experiments also illustrate that our model achieves superior performance over other baseline protocols in terms of network lifetime of 134 rounds the first node death (FND) following 202 rounds the half-node death (HND), energy consumption of 0.041 J/round, packet delivery ratio of 96.8%, end-to-end delay of 244 ms, and the routing overhead of 28cp/round.These findings demonstrate that the proposed Multi-Layer Clustering with CNN-GRU model can enhance the energy efficiency, reliability, and scalability of IoT networks.