In contemporary urban environments, the federation of IoT-empowered data spaces is gaining ground as a concept, however a single unifying approach to federation is still elusive. As a result, the exchange of data across heterogeneous federated data spaces often encounters challenges, such as different data models and data exchange protocols, or stringent policies prohibiting data sharing across federations. This paper introduces CCDUIT, a software overlay architecture designed to address these issues, facilitating seamless crossfederation collaboration. As a comprehensive solution, CCDUIT offers modularity, scalability, and interoperability, enabling efficient and sovereignty-preserving data exchange across diverse federations. CCDUIT leverages rich property graph models for context modeling of federations, which are exchanged across federations via publish/subscribe topic schemes, with data sharing, and access control governed by policy mechanisms matching the topics. Our experimental results demonstrate that CCDUIT significantly reduces the complexity and effort involved in data management and sharing across federations, with a quantifiable decrease in operational overhead by approximately 40% to 60%. This streamlines collaboration while ensuring compliance with data sovereignty and sharing policies, providing a solution to a longstanding challenge in federation-based data ecosystems.