Abstract: Many software systems operate in environments where change and uncertainty are the rule, rather than exceptions. Techniques for self-adaptation allow these systems to automatically respond to environmental changes, yet they do not handle changes to the adaptive system itself, such as the addition or removal of adaptation tactics. Instead, changes in a self-adaptive system often require a human planner to redo an expensive planning process to allow the system to continue satisfying its quality requirements under different conditions; automated techniques typically must replan from scratch. We propose to address this problem by reusing prior planning knowledge to adapt in the face of unexpected situations. We present a planner based on genetic programming that reuses existing plans. While reuse of material in genetic algorithms has recently applied successfully in the area of automated program repair, we find that naïvely reusing existing plans for self-* planning actually results in a loss of utility. Furthermore, we propose a series of techniques to lower the costs of reuse, allowing genetic techniques to leverage existing information to improve planning utility when replanning for unexpected changes.

Abstract: An important issue for the specification and design of software architectures is how to combine the analysis capabilities of multiple architectural definition languages (ADLs) and their supporting toolsets. In this paper, we describe our experience of integrating three ADLs: Wright, Rapide, and Aesop. We discovered that it is possible achieve interoperability in ADL tools for a non-trivial subset of the systems describable by these languages, even though the languages have different views about architectural structure and semantics. To carry out the integration we used the Acme architectural interchange language and its supporting tools.

Abstract: We are moving towards a world of pervasive computing in which users can access and manipulate information from anywhere at anytime. Computing devices and networks are becoming ubiquitous. In this new world, computing will no longer be tethered to desktops: users will become increasingly mobile. As users move across environments, they will have access to a dynamic range of computing devices and software services. They will want to use the resources to perform computing tasks. Today's computing infrastructure does not support this model of computing very well because computers interact with users in terms of low level abstractions: applications and individual devices. Today, if a mobile user wants to use the computing resources of a new environment, he has to manually figure out how to perform a computing task using local resources and/or to migrate his computing context from another environment. Such manual operation is unacceptable in a pervasive computing world because it does not scale with the increasing amount of software services, user mobility and resource dynamism. It also demands that users spend too much time on non-task related configuration activities. This technical report describes an approach called task-driven computing that can be used to solve this problem. The approach is based on the insight that with appropriate system support it is possible to let users interact with their computing environments in terms of high level tasks and free them from low level configuration activities.