Adapting Graph Structure for IoT Network: A Performance Analysis

Graph structures can effectively replicate intricate relationships among diverse entities, offering valuable insights into the interconnections among nodes. In this paper, we leverage graph neural network (GNN) algorithms for the classification of Internet of Things (IoT) nodes based on their connectivity and features. The node classification task involves transforming the IoT network into a graph structure, encompassing both fully connected and randomly connected graphs. In fully connected graph each node is connected to other nodes and randomly connected graphs have missing connections. Furthermore, two GNN methodologies, namely ARMAConv and Cluster-GCN, are employed for precise classification. The simulation results reveal that Cluster-GCN outperforms ARMAConv, achieving a higher success rate in node classification.