Assessing Dependability for Mobile and Ubiquitous Systems: Is there a Role for Software Architectures?

A traditional research direction in SA and dependability is to deduce system dependability properties from the Knowledge of the system Software Architecture. This will reflect the fact that traditional systems are built by using the closed world assumption. In mobile and ubiquitous systems this line of reasoning becomes too restrictive to apply due to the inherent dynamicity and heterogeneity of the systems under consideration. Indeed these systems need to relax the closed world assumption and to consider an open world where the system/component/user context is not fixed. In other words the assumption that the system SA is known and fixed at an early stage of the system development does not apply anymore. On the contrary the ubiquitous scenario promotes the view that systems can be dynamically composed out of available components whose dependability can at most be assessed in terms of components assumptions on the system context. Moreover dependability cannot be anymore designed as an absolute context free property of the system rather it may change as long as it allows the satisfaction of the user's requirements and needs. In this setting SA can only be dynamically induced by taking into consideration the respective assumptions of the system components and the current user needs. The talk will illustrate this challenge and will discuss a set of possible future research directions.