Security and Privacy of Digital Twins for Advanced Manufacturing: A Survey

TL;DR

The paper surveys security and privacy challenges of digital twins in advanced manufacturing, organizing threats into data collection data sharing ML DL and system level categories. It surveys concrete attacks and defenses across hardware, software, data provenance, DP HE SMPC TEEs adversarial and poisoning phenotypes, and anomaly detection frameworks, as well as model updating and uncertainty quantification. It also discusses data governance and architectural choices including blockchain based provenance and decentralized data sharing, as well as the role of uncertainty quantification in robust decision making. The findings highlight that realizing trustworthy digital twins requires defense in depth, privacy preserving ML techniques, secure model updates, and governance frameworks to balance innovation with resilience against cyber threats.

Abstract

In Industry 4.0, the digital twin is one of the emerging technologies, offering simulation abilities to predict, refine, and interpret conditions and operations, where it is crucial to emphasize a heightened concentration on the associated security and privacy risks. To be more specific, the adoption of digital twins in the manufacturing industry relies on integrating technologies like cyber-physical systems, the Industrial Internet of Things, virtualization, and advanced manufacturing. The interactions of these technologies give rise to numerous security and privacy vulnerabilities that remain inadequately explored. Towards that end, this paper analyzes the cybersecurity threats of digital twins for advanced manufacturing in the context of data collection, data sharing, machine learning and deep learning, and system-level security and privacy. We also provide several solutions to the threats in those four categories that can help establish more trust in digital twins.

Figures (6)

• Figure 1: Key Components of Advanced Manufacturing
• Figure 2: Digital Twin for Advanced Manufacturing
• Figure 3: Overview of Entities in Data Collection
• Figure 4: Data Sharing Challenges Areas: 1) Data Storage, 2) Data Access, 3) Data Provenance
• Figure 5: The overview of attacks and defenses in ML or DL inspired from the paper 9294026. We comprehensively review the existing privacy and security issues based on the DL life cycle. In addition, we also analyzed the defense methods. The arrows have the following meanings: Data Poisoning and Backdoor Attacks target at the training phase of the lifecycle, where we could apply different defence techniques (Poisoning Defense, Improving Models' Robustness, and Differential Privacy) to the Preprocessing part or the Training part, respectively. Additionally, Model Extraction, Model Inversion, Membership Inference and Adversarial Attacks target at the testing phase of the lifecycle, where we could apply different defence techniques (Homomorphic Encryption, Secure Multi-party Computation, Trusted Execution Environment and Malware Detection) to the Applying part.