A Novel Security Framework for IoT Networks using Blockchain Technology and Adaptive Residual LSTM with Attention Mechanism

Background: Real-world objects are now connected to the world's largest network because of the Internet of Things' (IoT) quick development, which has improved digital communication while posing serious security risks. In order to detect threats and defend IoT devices from cyberattacks, Intrusion Detection Systems (IDS) are crucial. The researchers throughout the world work together effectively on addressing the difficulties of IoT security. Here, a secure, decentralized platform made possible by blockchain technology, which is considered as one of the best options. Aim: This research developed a blockchain framework for strengthening the IoT network security. The process of performing secure communication in an IoT network by integrating blockchain and deep learning models while ensuring data protection is the main objective of the research work. Most importantly, the trustworthy data management in the IoT environment is ensured by the proposed framework. Methodology: The attack data related to IoT is garnered from public data sources. Next, the deep features are selected with the help of a conditional autoencoder. Subsequently, the selected features are subjected to an Adaptive Residual Long Short-Term Memory with Attention Mechanism (ARes-LSTM-AM) for effective intrusion detection. Moreover, the parameters of ARes-LSTM-AM are tuned using the Enhanced Osprey Optimization Algorithm (EOOA). Intrusion detection is performed to detect trusted users in Blockchain transactions and storage. For distributed intrusion detection engines, the improved privacy is provided by the Ethereum-based smart contract. Results: The proposed model attained promising results in accuracy of 93.6%, recall of 93.65%, and precision of 93.6% in the task of intrusion detection. Conclusion: Thus, the results showed that the recommended framework can ensure safe decentralized information sharing within the IoT network.