Run-time environment for resource-scarce networked systems

点击翻译按钮获取中文摘要

The dynamic resource management of processor, memory, and input/output of the different system classes (mobile nodes, ground nodes) as well as the resulting, potentially dynamic adaptation of the system software (adaptive run-time environment, ARTE) are in the focus of the research work. Besides application-oriented reasons for dynamic adaptations, it is especially important to consider the system's energy state of (mobile) nodes as a trigger for such adaptations. Energy models are being developed not only for single, isolated nodes but also for the distributed overall system that consists of mobile and ground nodes including (energy) costs resulting from their interactions. Based on these energy models, the (worst-case) energy consumption of (partial) program code is predicted. In particular, the system needs to react to environmental changes and varying requirements on behalf of the biologic applications by means of optimized and dynamic adaptation capabilities at any time. A distributed middleware, which stretches across all system classes, is being developed in order to manage and distribute required reconfiguration and reprogramming tasks. To maximize the energy awareness of the overall system, low-level system components are being developed in an energy-aware manner by using previously developed (energy) analysis tools. With the aid of the modules static analysis, energy analysis, and dynamic adaptation it is possible to exploit energy-saving and energy-harvesting mechanisms of the target platforms at run-time and, in case of energy shortage, to facilitate exception-handling decisions. At the end of the development, a dynamically adaptable, energy-aware system is available which provides a high degree of flexibility.