Using Role-Based Composition to Support Unanticipated, Dynamic Adaptation - Smart Application Grids

Due to the wide acceptance and distribution of mobile devices, it has become increasingly important that an application is able to adapt to a changing environment. This implies the necessity to integrate varying functionality at runtime being activated depending on the current context. A common approach is to foresee and model all possible influencing factors and to integrate the required software building blocks in advance. But, due to the constant change of the environment, as described by Lehman’s laws, it is impossible to anticipate all future situations. Hence, modeling the entire adaptation process at design time prohibits the adaptation to unanticipated scenarios and, thus, is likely to lead to the malfunctioning of the adaptive application in the future. In this paper we focus on unanticipated, dynamic selfvariation of applications (i.e., without a central coordinator) and propose a role-based composition system that enables the adjustment of the structure and functionality of softwareobjects in a fine-grained manner. Systems following our proposed approach form a Smart Application Grid (SMAG). The SMAGs-Approach is putting emphasis on dynamic collaborations between components within an application and between several different software systems. Therefore, role-modeling is used to model and perform dynamic variation of applications at runtime, whereby roles are stored in central repositories. This allows the integration of previously unknown softwarebuilding-blocks and the dynamic adaptation to situations that were not foreseen. Keywords-Dynamic Variation; Unanticipated Adaptation; Role-Modeling; Composition; Repository.