EHO-CNN-TBR: A Trust-Based Secure Routing Model for IoT Using Elephant Herding Optimization and Convolutional Neural Networks

The fast expansion of the Internet of Things (IoT) has enabled autonomous communication across massive, heterogeneous device networks.However, the dynamic topology and resource limitations of IoT environments expose routing processes to severe security threats, particularly malicious attacks that disrupt reliability and network stability.This study introduces EHO-CNN-TBR, a hybrid secure routing framework that integrates Elephant Herding Optimization (EHO) for adaptive trust evaluation with a Convolutional Neural Network (CNN) for real-time anomaly detection.EHO optimizes trust based on energy level, packet forwarding behavior, and historical performance, while CNN identifies malicious or abnormal traffic patterns.Using OMNeT++ with the INET Framework, the proposed model is evaluated against ACO-TBR [11] and FL-TBR [18].Comprehensive results across PDR, delay, throughput, energy, routing overhead, and detection metrics demonstrate that EHO-CNN-TBR substantially outperforms existing approaches, offering a more reliable and attack-resilient routing solution for secure IoT communication.