Our research is focused on Interconnected Resource-aware Intelligent Systems. In this context, we design, analyze, develop, and evaluate concepts, models, algorithms, protocols, and tools that optimize (distributed embedded) systems performance in terms of timing behavior, dependability, programmability, reliability, robustness, scalability, accuracy, energy and data computation efficiency, and trustworthiness.
The following three pillars, and their interaction, form the basis of IRIS research:
- Pervasive networking: Edge networking, (IoT) communication and interoperability, (IoT) Network resource management
- Pervasive computing: Edge computing and federated learning, Edge-driven data analytics and machine learning, Embedded AI, Explainable AI, Data flow computing
- Predictable performance: Time-sensitive networks, Realtime distributed systems, Resource management and scheduling, Formal modeling, analysis & verification, Model-based engineering of CPS
The interaction between these research pillars is important as, optimization of resource usage against the required quality of services often requires a cross-layer and system-wide approach. Such interaction is not only ensured through “optimization by design”, i.e., careful analysis of design implications on resource usage quality of service indictors/guarantees, but also through implementation in real-world applications (as much as possible) and the over-arching research theme of IRIS, namely, System Architecture.
Research at IRIS is of both practical and theoretical nature. We consider experiments analyzing contexts of use in multiple industrial application areas and consumer market such as automotive, predictive maintenance, smart industry, logistics, health-care and intelligent lighting, essential for applicability and validation of theories, models, algorithms, protocols, and tools that we develop.