博士後期課程公聴会 （Faust Oliver）
日時
平成27年7月22日(水) 16:10-17:40
場所
自然科学総合研究1号館 3階 セミナー室
発表者
Faust Oliver
題目
Formal and model driven design for biomedical engineering
主査
羽石秀昭 教授
副査
伊藤智義 教授，中口俊哉 准教授，兪文偉 教授(指導教員)
要旨
Every biomedical health care system in existence is a physical realization of an idea or a strategy on how to solve a medical problem. These physical problem solutions must be safe, reliable and functional. The functionality is established through modeling. The reliability is tested by comparing the physical problem solution with the model. For most systems, the comparison takes the form of use and failure case testing. However, for complex physical systems, testing can only confirm the presence of a fault but never proof the absence of system faults. Therefore, it is impossible to establish safety through testing -- the safety critical fault might lurk in an untested system state. Safety is a design property, hence the way we design a system is very important.
With this thesis I propose a formal and model driven design methodology for biomedical systems. The idea is to extend the well-established systems engineering design methodology with formal and functional models. The systems engineering design methodology structures the design efforts. Within the structured design, models prove or at least estimate certain aspects of the system, such as safety, reliability and functionality. I demonstrate the efficacy of functional models for diseases, such as diabetes, sleep apnoea, and epilepsy. The functional models are tested and the test results are compared with published research work in the respective domain. The comparison shows that each of the proposed functional models constitutes an improvement on state of the art systems. To advance the safety of biomedical systems, I put forward formal models which proof that a specific system is deadlock and livelock free. Deadlock and livelock are systemic problems which cannot be addressed in the functional model. The proposed work benefits society, because (a) formal and model driven development leads to reliable systems which function according to specification, (b) the design methodology delivers physical problem solutions on time and within budget.
With this thesis I propose a formal and model driven design methodology for biomedical systems. The idea is to extend the well-established systems engineering design methodology with formal and functional models. The systems engineering design methodology structures the design efforts. Within the structured design, models prove or at least estimate certain aspects of the system, such as safety, reliability and functionality. I demonstrate the efficacy of functional models for diseases, such as diabetes, sleep apnoea, and epilepsy. The functional models are tested and the test results are compared with published research work in the respective domain. The comparison shows that each of the proposed functional models constitutes an improvement on state of the art systems. To advance the safety of biomedical systems, I put forward formal models which proof that a specific system is deadlock and livelock free. Deadlock and livelock are systemic problems which cannot be addressed in the functional model. The proposed work benefits society, because (a) formal and model driven development leads to reliable systems which function according to specification, (b) the design methodology delivers physical problem solutions on time and within budget.