Abstract

Software complexity is the main obstacle to further progress in IT industry, as the difficulty of managing complex and massive computing systems goes well beyond IT administrators’ capabilities. One of the remaining options is autonomic computing, which helps to address complexity by using technology to manage technology in terms of hiding and removing low level complexities from end users.
Real-time reactive systems are some of the most complex systems that have become increasingly heterogeneous and intelligent. Thus, we want to add autonomic features to real-time reactive systems by building a formal framework, Reactive Autonomic Systems Framework (RASF), which can leverage specification, modeling and development of Reactive Autonomic Systems (RAS). With autonomic behavior, the real-time reactive systems are more self-managed to themselves and more adaptive to their environment.
Formal methods are proven approaches to ensure the correct operation of complex interacting systems. However, many current formal approaches do not have appropriate mechanisms to specify RAS and have not addressed well on verifying self-management behavior, which is one of the most important features of the RAS. The management of evolving specifications and analysis of changes require a specification structure, which can isolate those changes in a small number of components and analyze the impacts of a change on interconnected components. Category theory has been proposed as a framework to offer that structure; it has a rich body of theory to reason about objects and their relations. Furthermore, category theory adopts a correct by construction approach by which components can be specified, proved and composed in the way of preserving their properties.
In the multi-agent community, agent-based approach is considered as a natural way to model and implement autonomic systems, as the ability of an autonomous agent can be easily mapped to the self-management behaviors in autonomic systems. Thus, many ideas from the Multi-Agent Systems (MAS) community can be adapted to implement the autonomic systems, such as the self-management behavior, automatic group formation, interfacing and evolution.
Therefore, in terms of achieving our research goal, we need to i) build an architecture and corresponding communication mechanism for modeling both reactive and autonomic behavior of the RAS, ii) formally specify the architecture, communication and behavior above using category theory, iii) design and implement the architecture, communication as well as behavior of the RAS model by the MAS approach with its implementation and iv) illustrate our RASF methodology and approach with case studies.