Research Topics and Projects
Energy-efficient Parallel Computing
We investigate static scheduling of tasks with real-time properties on multicore processors with frequency scaling, to meet deadlines and minimize energy consumption. Both independent tasks, task graphs and streaming applications are considered for homogeneous and heterogeneous platforms. Newer approaches target energy cost, which can vary over time and place. Beyond theoretical results, heuristics are tested on real systems. Additionally, the energy-efficient and fault-tolerant execution of task graphs on parallel platforms is investigated. Finally, numerical and combinatorial algorithms for multicores and GPUs are parallelized.
Safety & Security
Since some years, we focus on hidden channels in networks and systems, investigating both new channels and countermeasures as hidden channels are a dual-use good. Furthermore, the structure of state spaces for cryptographic pseudo random number generators, stream ciphers and cryptographic hash functions is investigated, to find hints for weaknesses that might enable attacks, to find energy-efficient solutions, and to find possibilities to "repair" short cycles without re-design. Moreover, tools for forensic investigations in clouds are developed. In the future, the relatioin between security level, energy consumption and performance of cryptographic systems will be investigated. The research on safety/fault-tolerance is described under parallel computing.
Given that FernUni is a distance-teaching institution, and given the focus on computer engineering, the group develops virtual labs in different parts of computer science and engineering.