DS4DT: An IDS-Compliant Data Space Reference Architecture for Sovereignty-Preserving Industrial Digital Twin Federations

n multi-stakeholder industrial environments, the strategic value of Digital Twins (DTs) hinges on secure data exchange and coordinated intelligence across organizational boundaries while safeguarding data sovereignty, trade secrets, and regulatory compliance. This paper introduces DS4DT, a reference architecture that operationalizes an International Data Space (IDS)-compliant Data Space (DS) layer for industrial DT federations by consolidating established standards and mature open-source components into a reproducible, edge-deployable blueprint. The architecture treats model updates and coordination signals as sovereign data flows subject to the same trust, identity, and policy mechanisms as raw data, thereby enabling decentralized learning under policy-driven constraints through a dedicated Federated Intelligence Layer (FIL). An Industrial Energy Community (IEC) case study, instantiated via a Smart PhotoVoltaic Station (SPVS) as a controllable proxy for industrial flexible assets, demonstrates how the pattern scales from individual energy nodes to multi-tenant federations. Empirical characterization on an edge cluster quantifies both the aggregated end-to-end overhead of the DS layer relative to baseline connector-less DT deployment and the impact of policy complexity. Results reveal predictable, single-digit-millisecond steady-state overhead that amortizes through caching mechanisms, with policy-evaluation cost scaling gracefully within acceptable bounds. Analytical compatibility assessment translates these measurements into capacity bounds for Federated Learning (FL) rounds on the same hardware. A concrete prototype and end-to-end FL validation are identified as primary future directions.