Diagnosis of Ambient Systems Based on the Modeling of Effects

International audienceComplex ambient systems are composed of many heterogeneous systems; each system interacts with its surroundings using collected environmental data from sensors' readings. Diagnosis for the overall system as well as the sub-systems composing it must address many challenges caused by the dynamic nature of these systems and the impossibility to pre-define control loops between sensors and actuators at design time as they are discovered dynamically at run-time. This paper presents a design technique that is based on defining for each component the expected effect — the “physical phenomenon” — it is supposed to produce or receive. The proposed approach facilitates the comparison of the produced effects in the environment and the actual readings of the sensors, and thus simplifies the diagnosis task. To do so, we propose a precise definition of the concept of effect. The approach is validated by implementing a simple yet complete example taking place in an ambient environment