CAEHV: um método para verificação de modelos do tratamento de exceção sensível ao contexto em sistemas ubíquos

Abstract

The context-aware adaptive software, also known as self-adaptive software, represents a specic class of complex software systems, which has as primary requirement the ability to monitor its execution environment (physical and logical) and reacts upon change, either adapting its internal structure and behavior or performing automatic tasks. In ubiquitous environments, context-aware adaptive software plays a fundamental role, as it must be designed to interact with the physical environment, aiming to transparently supporting users in their daily activities, improving or hiding the interaction between human and computer. In that direction, the application domain in which context-aware adaptive software may be applied is broad (e.g., smart home, mobile visit guide, gaming, health or rescue missions), requiring a high-degree of reliability, not only capable of performing its functional requirements, but also dealing with abnormal conditions, avoiding occasional failures. Although context-aware exception handling has been used to improve robustness and reliability levels of context-aware adaptive software, it requires from the software designers a good understanding of the application domain, turning out to be a complex and error-prone activity. In order to achieve the expected results of using context-aware exception handling, it is mandatory to follow a rigorous design approach, trying to minimize the number of design faults. In this scenario, this doctoral thesis proposes a method for model checking context-aware exception handling, named CAEH!(Context-Aware Exception Handling Verication). It provides a set of abstractions that permits designers to model the context-aware exceptional behavior and translates it to a Kripke structure. Additionally, a set of behavior properties is established, attempting to aid designers to identify specic types of design faults. To evaluate the feasibility of the proposed method, a tool, called JCAEH!, for modeling and verication of context-aware exceptional behavior is developed, and fault injection scenarios are modeled to analyze its eectiveness in identifying design faults.