BlockWave: A Blockchain-Enabled Security Architecture for Trustworthy Data Exchange in 6G-Powered IoT Networks

Abstract

The evolution of 6G networks brings unprecedented connectivity and processing capabilities to Internet of Things environments, yet it also raises more challenges in data security, intrusion detection, and computational efficiency. In this paper, a complete end-to-end framework with the inclusion of Artificial Intelligence, blockchain technology, and novel encryption techniques is proposed to address these challenges in 6G-IoT networks. Then, the AI-Powered Cybersecurity Events Dataset is utilized first, wherein network traffic is normalized through Min-Max normalization to normalize heterogeneous features to a uniform scale. A Gated Recurrent Unit-based neural network is then trained on this normalized data to detect real-time intrusions by learning complex temporal dependencies. Upon detecting anomalies, a blockchain layer is called to execute smart contracts that apply automatic security measures, e.g., quarantining affected nodes. For efficient processing, the architecture accommodates GRU-assisted task offloading optimization on a multi-factor delay, energy, and network load optimization model. Blockchain smart contracts manage load balancing and delegation verification autonomously without any central authority. Homomorphic encryption and proxy re-encryption also ensure data confidentiality in edge computing and secure multi-party computation. All encryption and task offloading operations are permanently stored onto the blockchain, ensuring system-wide transparency and auditability. Federated learning support is built-in to enable privacy-preserving decentralized AI training from distributed edge nodes. The entire framework is implemented in Python using TensorFlow/Keras for the GRU model and web3.py for blockchain-related interactions. Experimental measurements demonstrated the efficiency and stability of the system by exhibiting a percentage of accuracy of 98.9% in intrusion detection, justifying the applicability for future 6G-IoT secure and scalable operations. The experiments utilize the AI-Powered Cybersecurity Events Dataset containing approximately 1.5 million network flow records with class distribution of 52% benign traffic, 21% DoS, 15% probing, and 12% privilege escalation events.