Martin Breitbach, Christian Becker
Prof. Janick Edinger, Universität Hamburg
The demand for computing power in areas such as machine learning or computer vision has increased considerably. Computation offloading can meet this demand, even on resource-constrained devices such as smartphones. The devices offload workload to remote resource providers in the cloud or in the edge, which return the results via the network. In addition to the acceleration of computationally intensive tasks, computation offloading reduces the energy consumption of the offloading device.
The Tasklet project is a DFG-funded project that develops a computation offloading system for fast and energy-efficient edge computing. The Tasklet system allows heterogeneous devices to offload arbitrary tasks to a heterogeneous pool of resources. Thus, the approach is attractive for many Internet of Things applications.
As computing devices became more and more mobile, the research area of Cyber-Physical Systems (CPS) emerged out of the areas of Pervasive/
To simplify the design process as well as the operation and maintenance of a MC-CPN a comprehensive approach to handle the complexity of MC-CPNs autonomously at runtime is needed. We strive to develop learning self-adaptation mechanisms that enable and optimize the execution of tasks according to their Quality-of-Service requirements with respect to their criticality. Thereby, the whole management and adaptation capabilities are encapsulated into a middleware framework which provides transparency and realizes a robust and flexible MC-CPN.