Research
For many application domains, such as the Internet of Things (IoT) and Cyber-Physical Systems, intelligent computer systems are required to interact with and respond to physical processes in a timely and reliable manner. Supporting such requirements is enabled by distributed systems that need to orchestrate massive processing and communication resources to analyze location-dependent data from interacting devices, e.g., sensors, actuators, users, or other computer systems. Especially for low-latency and energy-efficient responses, a good choice of resource location and type is the key to fulfilling complex application requirements.
Our research contributes to new mechanisms and methods for self-adaptive distributed and network-centric intelligent computing systems. This comprises methods and mechanisms for
- Coping with the tremendous volume of data streams, e.g., in the form of highly flexible data processing systems supporting functions like inference serving, event- and stream processing, or asynchronous communication of data (publish/subscribe),
- Efficiently utilizing resources like the network infrastructure, processing resources, and storage by building on novel programming models, e.g., by building on abstractions of software-defined networks and network function virtualization,
- Ensuring information consistency and reliability by analyzing the characteristics and forms of specific intelligent computing functions, and
- Adapting intelligent computer systems to account for several constraints, e.g., accounting for privacy and security policies for accessing information, quality of data, quality of service (QoS), and energy-efficiency.
Consequently, our research addresses problems in the deeply intertwined research fields of distributed data analytics and management, parallel and distributed computing, and networked systems. Our core research areas can be summarized by
- Distributed Computing
- Cloud and Edge Computing
- Computer Networks
- Middleware