Assuring Intelligent Ambient Assisted Living Solutions by Statistical Model Checking.

A modern way of enhancing elderly people’s quality of life is by employing various Ambient Assisted Living solutions that facilitate an independent and safe living for their users. This is achieved by integrating computerized functions such as health and home monitoring, fall detection, reminders, etc. Such systems are safety critical, therefore ensuring at design time that they operate correctly, but also in a timely and robust manner is important. Most of the solutions are not analyzed formally at design time, especially if such Ambient Assisted Living functions are integrated within the same design. To address this concern, we propose a framework that relies on an abstract component-based description of the system’s architecture in the Architecture Analysis and Design Language. To ensure scalability of analysis, we transform the AADL models into a network of stochastic timed automata amenable to statistical analysis of various quality-of-service attributes. The architecture that we analyze is developed as part of the project CAMI, co-financed by the European Commission, and consists of a variety of health and home sensors, a data collector, local and cloud processing, as well as an artificial-intelligence-based decision support system. Our contribution paves the way towards achieving design-time assured integrated Ambient Assisted Living solutions, which in turn could reduce verification effort at later stages.