Predictive IoT Network Routing Optimization Using Hybrid Augmented Gradient Boosting Classifier Algorithm

Abstract The exponential development of Internet of Things (IoT) systems lead to network traffic challenges that require advanced classification methods for optimizing resource distribution, congestion control and QoS enhancement. Both rule-based and deep packet inspection classification techniques show limited capacity to respond to IoT traffic changes because of its dynamic complexity together with encryption barriers. The proposed research work developed a machine learning-based technique for IoT traffic classification and routing by using HAGBC alongside LDA and best fit feature techniques which enabled traffic clustering using Tuna Swarm-inspired Jaccard and Interpolation Fuzzy C-Means Clustering. The model achieves optimal weight adjustment through Self-Updated Dung Beetle Optimization which results in superior classification accuracy together with reduced computational complexity. The experimental results have demonstrated increased scalability, decreased energy usage, and enhanced accuracy with the proposed algorithm. The proposed model maintains adaptive functionality toward network changes to execute real-time traffic management efficiently for developing automatic network infrastructure systems across extensive IoT settings.