Blockchain-based AI-enabled data protection for the Internet of Things

Industry 4.0 is all about doing things in a concurrent, secure, and fine-grained manner. IoT edge sensors and their associated data play a predominant role in today's Industry 4.0 ecosystem. Breaching data or forging source devices after injecting advanced persistent threats damages the industry capital and even loss of lives. The existing security challenges of Industry 4.0 include sophisticated cyberattack injection targeting vulnerable edge devices, insecure data transportation, trust inconsistencies among stakeholders, privacy-incompliant data storing mechanisms, etc. IoT edge servers often suffer because of their lightweight computation capacity to filter data anomalies properly, transport large-scale data securely and maintain data collaboratively with privacy, which altogether makes IoT systems exposed to attackers and hackers. Therefore, there has been an intense concern for security alternatives because of the recent rise of cyberattacks, mainly targeting enterprise, health and energy IoT infrastructures. Existing attack detection largely depends on machine learning techniques. However, the model trained on corrupted data should escalate suffering than defend the system. Moreover, conventional IoT security mechanisms employ trust from Public-key-Infrastructure third parties which can increase costs and mostly seems incompatible with the lightweight IoT ecosystems. While seeking alternatives, blockchain technology unveils promising security strengths to uphold a secure IoT setup. The loopholes of traditional attack detection, certificate-authority-dependent authentication, and privacy-incompetent communication have motivated the project to extract valid problems and solutions in the cybersecurity domain. After describing four research questions and methods, this dissertation focuses on how blockchain technology coupled with artificial intelligence foster certificate-less authentication and privacy-preserved cyberattack protection mechanisms. Firstly, the unique contributions include locally detecting malicious data to save the IoT system from failure. The blockchain-based solution demonstrates a novel false data detection and reputation preservation technique  based on the fuzzy rules and behaviors of the IoT-enabled cyber-physical system. It improves detection accuracy and eliminates single points of failure. Secondly, sensors and parties constitute a consortium blockchain network to authorise both data and sources that ensure trusted communication among participating devices. The demonstrated methods obviate the long-established certificate authority after enhancing the consortium Blockchain which reduces the data processing delay and increases cost-effective throughput. The distributed industry 4.0 security model entails cooperative trust using multi-signature than depending on a single party. Thirdly, it brings an efficient advanced persistent threat detection mechanism at the edge and transparent recording of the detection history in an immutable ledger where the edge-compliant storage technique facilitates efficient predictive maintenance. Finally, a blockchain-enabled, privacy-preserved collaborative machine learning model shields the IoT system internally at the edge and externally at the dedicated server. It allows other client devices to work flawlessly without being impacted by the delay of incurred by nearby clients and also empowers edge devices to contribute in training to downsize the server workload. Nonetheless, the outcomes of extensive experiments presented in this thesis explain the soundness and significance of the proposed research methods through respective qualitative and quantitative analysis. The evaluation results also confirm the applicability of the proposed methods for the real-time, large-scale and collaborative Industrial IoT system.