Towards Trusted and Intelligent Cyber-Physical Systems: A Security-by-Design Approach

TL;DR

This paper proposes a trusted, intelligent Digital Twin framework for Cyber-Physical Systems that integrates blockchain and AI to enhance security, data integrity, and real-time threat detection throughout the product lifecycle.

Contribution

It introduces a novel hybrid framework combining blockchain and AI for trustworthy, security-aware Digital Twins in CPSs, addressing data trustworthiness and real-time responsiveness.

Findings

Framework effectively integrates blockchain and AI for CPS security.

Application demonstrated in automotive industry use case.

Identifies open challenges for implementing intelligence-driven CPS architectures.

Abstract

The complexity of cyberattacks in Cyber-Physical Systems (CPSs) calls for a mechanism that can evaluate the operational behaviour and security without negatively affecting the operation of live systems. In this regard, Digital Twins (DTs) are revolutionizing the CPSs. DTs strengthen the security of CPSs throughout the product lifecycle, while assuming that the DT data is trusted, providing agility to predict and respond to real-time changes. However, existing DTs solutions in CPS are constrained with untrustworthy data dissemination among multiple stakeholders and timely course correction. Such limitations reinforce the significance of designing trustworthy distributed solutions with the ability to create actionable insights in real-time. To do so, we propose a framework that focuses on trusted and intelligent DT by integrating blockchain and Artificial Intelligence (AI). Following a hybrid approach, the proposed framework not only acquires process knowledge from the specifications of the CPS, but also relies on AI to learn security threats based on sensor data. Furthermore, we integrate blockchain to safeguard product lifecycle data. We discuss the applicability of the proposed framework for the automotive industry as a CPS use case. Finally, we identify the open challenges that impede the implementation of intelligence-driven architectures in CPSs.