Toward Trustworthy Digital Twinning: Taxonomy, Analysis, and Open Challenges

The proliferation of Digital Twins (DTs) across industries like manufacturing, healthcare, and logistics is leading to the formation of complex ecosystems where heterogeneous DTs must cooperate. In such environments, establishing trust becomes paramount. However, trust in DTs remains an under-investigated problem, with current research predominantly focused on security and privacy, which are prerequisites but not sole constituents of trust. This paper presents a comprehensive framework for analyzing and enhancing the trustworthiness of Digital Twins. First, we propose a novel five-layer symmetrical reference architecture (Asset, Synchronization, Data, Application, Integration) that models physical and digital twins as peers, improving reusability and maintainability. Using this architecture as a foundation, we then develop a multi-dimensional taxonomy to categorize DT trust issues from three critical perspectives: (1) an architectural perspective, which identifies and maps trust issues (e.g., model accuracy, data latency, application usability) to specific layers and behavioral attributes like conformance and dependability; (2) a massive twinning perspective, which explores emergent challenges in ecosystems of cooperating DTs, such as relationship complexity and data management; and (3) a stakeholder perspective, which addresses the need for both qualitative and quantitative trust assurances. Our analysis reveals that trust is a composite property requiring a holistic approach beyond conventional security. The paper concludes by synthesizing these perspectives into a unified view of DT trust and outlining critical open challenges and future research directions, providing a foundational roadmap for developing truly trustworthy Digital Twin systems.