Adaptation of Open Component-Based Systems

Abstract

Abstract. Software adaptation aims at generating software pieces called adaptors to compensate interface and behavioural mismatch between components or services. This is crucial to foster reuse. So far, adaptation techniques have proceeded by computing global adaptors for closed systems made up of a fixed set of components. This is not satisfactory when the systems may evolve, with components entering or leaving it at any time, e.g., for pervasive computing. To enable adaptation on such systems, we propose tool-equipped adaptation techniques for the computation of open systems adaptors. Our proposal also support incremental adaptation to avoid the computation of global adaptors.

Software Components Behavioral Adaptation Based on LTS

“…The signature and protocol layers are mainly concerned when adaptation. Based on abstract model, they can be classified as method based on finite state machine [9]- [11], method based on process algebra [12] [13], method based on Petri net [14]- [16] and method based on LTS [6] [17].…”

Robust reconfiguration of cloud applications

“…Dynamic reconfiguration has been extensively studied in the last 20 years in the context of, e.g., software architectures [16,19,15,4,17], graph transformation [3,24], software adaptation [22,21,8], metamodelling [14,18], or reconfiguration patterns [7]. In software architectures, for example, the authors proposed various formal models, such as Darwin [16] or Wright [4], in order to specify dynamic reconfiguration of component-based systems whose architectures can evolve (adding or removing components and connections) at run-time.…”

Behavioral adaptation of component compositions based on process algebra encodings

“…Restrictive adaptation may restrict interactions between components in order to make the system work. This makes the assessment (evaluation) of adaptors and adapted systems a mandatory step of an adaptation approach in order to be sure they fulfil one's needs [28,31]. An additional benefit of our Lotos encoding is to support both the generation of adaptors and the assessment of adapted systems in a common framework.…”

“…Some work have recently tried to address this issue either at the computation time -by supporting the incremental computation of adaptors [31] -or at deployment timeby distributing the adaptors over the components [3]. Yet, they build on algorithms for the computation of global adaptors previously defined by their authors.…”

“…Compared to the existing adaptation approaches, ours also benefits from being based on a formal toolset which has been intensively validated on large case studies over the last years, namely Cadp. In particular, compared to our own previous approaches [14,31], the one at hand deals with the same mismatch but follows a simplified process and has a better complexity. Using Compositor to get a Lotos specification, one can directly rely on Cadp and our external module for it, Scrutator to get adaptors.…”