FOR 1085: OC-TRUST - Trustworthiness of Organic Computing Systems

点击翻译按钮获取中文摘要

Organic Computing systems differ from classical software systems as the topology of the system and the participating components of the system are not predefined and are subject to unforeseeable change during the system's runtime. Thus, completely new challenges to the verification and validation of such systems as well as for interactions between system components and, of course, the user interface arise. These challenges can be subsumed by the term trustworthiness or trust. Trust encompasses different aspects, as, for example, safety, reliability, credibility and usability, the investigation of which requires appropriate methods for each case. The aim of the Research Unit is to develop techniques that allow the consideration of trust during analysis, design and implementation of Organic Computing systems. Apart from the conscious generation of trustworthiness, evaluation and comparison of different facettes of trust or metrics thereof are parts of this endeavour. Organic Computing systems are highly dynamic, composed of a vast number of changeable components and are located in an ever changing environment. These facts yield a number of highly desirable properties, e.g. the ability to self-heal, to self-adapt or to self-configure. However, classical techniques for analysis and design of software systems are no longer suitable for systems of such complex structure. Novel aspects that could not be observed in other systems, such as emergent properties, and the extreme dynamics of OC-systems require a new way to think about such systems as well as the development of new mechanisms. These mechanisms include formal methods for functional correctness, safety and security as well as methods to describe and measure the trustworthy interaction of parts of the system, the observation of predefined policies at runtime, and the development of algorithms that take into account aspects of trust in self-organising systems. In particular, the user interface can no longer be implemented in a conventional fashion. Questions dealing with the graphical representation of self-organising system structures, the adaptive representation of information on different kinds of displays, and finally the privacy of data have to be investigated.