Toward a Post-Quantum, Federated and QoE-Aware IoT Security Framework Using Hybrid Deep Learning

The blistering growth of the Internet of Things (IoT) has brought about serious concerns in terms of the security, privacy, scalability, and Quality of Experience (QoE). The current solutions, e.g., deep learning-based intrusion detection system (IDS), blockchain-based security, and federated learning (FL), are independent and lead to disjointed and inefficient solutions. In this paper, a hybrid model of convolutional neural networks (CNN), transformer-based attention, and federated learning are proposed to achieve multi-task IoT intelligence in order to detect intrusion and predict quality of experience simultaneously. Moreover, the framework includes post-quantum security-related considerations that can be used to combat future cryptographic threats. The suggested architecture provides the privacy in the distributed IoT setting (smart cities and 5G/6G networks), scalability, and adaptability. Evaluation of experimental results on datasets of CICIoT2023 and IoT-TON has shown higher detection precision, lower latency, and high QoE operation.