Toward A Real-Time Capable and Reconfigurable Middleware for Scalable and heterogeneous Distributed System

Abstract

The German Aerospace Center (DLR, Deutsches Zentrum für Luft- und Raumfahrt), Germany’s national research institution for aeronautics, Space, energy, transportation, and security, is mandated by the federal government to implement the national Space program and advance technological research and knowledge transfer. Within this framework, this thesis proposes a data-centric middleware architecture that can enable a Satellite-as-a-Service (SaaS) model, meeting stringent requirements for safety, real-time performance, predictability, reliability, scalability, and modularity. Current onboard computing platforms, such as Scalable Onboard Computing for Space Avionics (ScOSA), follow message-centric paradigms with centralized coordination. These architectures impose rigid reconfiguration procedures, suspend operations across healthy nodes during topology changes, and limit extensibility. In response, this work proposes a Data Distribution Service-based (DDS), data-centric middleware that ensures strict binary-level service isolation, peer-to-peer data exchange, and dynamic reconfiguration. The design enables unaffected nodes to continue running during topology changes, improving system availability and resilience. A prototype using OpenDDS demonstrates feasibility on flight-representative hardware, achieving low latency in varied operational scenarios and maintaining higher availability during reconfiguration events, with only a modest increase in reconfiguration time. The use of structured DDS data models and Quality of Service policies facilitates seamless integration of new services or subservices, accelerating feature deployment. Nonetheless, the study identifies limitations in OpenDDS for Space-grade applications, including reliance on dynamic memory allocation, non-deterministic behavior, and inefficiencies incompatible with safety standards such as MISRA C++ and JSF ++. Addressing these shortcomings will require a deterministic, static-memory DDS profile designed for safety-critical and resource-constrained missions. Overall, the results confirm that a safe, predictable, and reconfigurable data-centric middleware is technically viable and offers clear advantages over message-centric approaches. Realizing its operational potential will depend on optimizing data handling, refining communication strategies, strengthening cybersecurity, and developing a Space-qualified DDS implementation— paving the way for data-centric SaaS middleware as a baseline for future scalable, multi-tenant satellite systems. This thesis contributes an architectural prototype that demonstrates how data-centric middleware can sustain availability during reconfiguration, reduce integration effort for new services, and provide a pathway toward certifiable space-qualified implementations. These contributions can be applied directly to future satellite missions adopting multi-tenant payload models, where predictable service continuity and flexible adaptation are paramount.