A system is said to be adaptive if it responds to environmental changes such as hardware/software defects, resource changes, and non-continual feature usage, in ways that extend the area of operation and improve the efficiency in the use of system resources. Adaptivity often incurs overhead in terms of system complexity and resource requirements, but unfortunately the resulting tradeoffs are usually ignored until a very late stage of the system development process. Retrofitting existing prototypes, architectures, middleware, operating systems, and protocols with concepts and means for flexibility such as run-time system reconfiguration or reflexive diagnostics and steering methods, typically leads to disproportionate overhead, unusual tradeoffs, and less satisfactory results. There is a strong need for adaptivity-centered research. To exploit adaptivity, new specification methods are needed, to define acceptable adaptation ranges which will be explored by the system at run-time to improve a given performance metric. Current operating systems and network protocols are generally not designed to support such flexible requirements nor complementary reflexive mechanisms to help applications adjusting their operation to the current conditions. The same can be said about current fault tolerance mechanisms, which are usually not designed such that they can adapt to different dependability goals during system operation. Programming such systems also needs adequate middleware layers that provide adequate interfaces for the development of adaptive applications. Building such middleware so that it preserves adaptive properties while providing performance guarantees together with satisfying other usual goals, such as modularity, reusability and scalability, is a challenge still to be conquered. In general, flexibility and complexity are counterpoised to dependability, but these notions must be reconciled in order to design dependable adaptive systems. This challenge encompasses aspects such as investigation on how adaptivity can be used as a means to achieve improved performance and efficiency without sacrificing dependability (for instance through reconfiguration upon failures) and investigation on how to guarantee that the adaptive mechanisms themselves are dependable, e.g. reliable, available, safe, etc. This workshop brings together experts in the development and use of adaptive and reconfigurable embedded systems and researchers from the embedded systems community at large. Of particular interest are new concepts and ideas for modeling and analyzing tradeoffs of embedded and real-time systems, novel algorithms and mechanisms to realize adaptation and reconfigurability, and experience reports with practical case studies.
04月13日
2015
会议日期
注册截止日期
2014年04月14日 德国
第六届自适应和可重构嵌入式系统研讨会
留言