CNN–Transformer-based Hybrid Deep Learning Framework with PSO Optimization for IoT Intrusion Detection

The Growth of IoT (Internet of Things) Networks is exacerbating Security Vulnerabilities That Can Be Exploited by Increasingly Advanced Cyber Attacks Targeting All Types of Connected Devices. This Research Presents a Hybrid Deep Learning Framework for IoT Intrusion Detection Based On CNN-Transformer Models Combined with Particle Swarm Optimization (PSO). The Hybrid Deep Learning Framework Combines a Combination of Feature Extraction by Convolutional Neural Networks (CNN), And Long-Range Dependency Learning Through Transformer Models, To Help Accurately Identify Malicious Network Activities. To Improve Classification Performance, PSO is Used for Optimal Hyper Parameter Tuning and Feature Selection, Resulting in Higher Detection Rates with Lower Computational Costs. The Framework Will Be Tested Using Existing Bench Marked Datasets of IoT Intrusion Detection and Compared to Previous Machine Learning and Deep Learning Methods for Their Performance Measurement of Accuracy, Precision, Recall, F1-Score, And False Alarms. The Hybrid Architecture Will Be Able to Represent Both Local and Global Traffic Patterns, Allowing for High Levels of Robustness in Detecting Many Types of Attacks. The Proposed Hybrid Deep Learning Framework Can Provide a Comprehensive, Intelligent, Scalable Cybersecurity Solution for Protecting Innovative, Cutting-Edge IoT Environments from Emerging Threats.