Signal processing and AI-based data evaluation
In order to realize an efficient data evaluation in terms of Green ICT and under real-time conditions we investigate signal processing and AI methods which can be used for resource-limited systems. Especially for AI methods the emphasis is on the realization of hardware accelerators, e.g. for the detection of attacks on the basis of processor data or for the detection of freezing of gait and the triggering of counter measures within less than 10 milliseconds for Parkinson patients. These approaches are also used to evaluate the condition of the systems in order to control their resilience by initiating appropriate measures such as relocation of activities to neighboring systems, restarts for rather transient faults, etc.