Over the last two decades, embedded systems have been introduced in several application areas (transportation, industry, housing, medical...). These systems have achieved more important tasks for delivering new services to users with ever shorter time-to-market deadlines at lower cost. The rapid integration of these systems within manufactured products is a competitive edge for companies. However, breakdowns related to these systems, along with the increasing level of equipment complexity, have made maintenance interventions much more difficult. Identifying root causes of some breakdowns currently represent a real challenge in the maintenance activities. They lead to an excessive downtime of equipment. This thesis proposes a versatile "active" monitoring architecture for the maintenance assistance of mobile systems. This architecture relies on "smart" monitoring entities that can assess the health state of monitored equipment. Our contribution gathers different Research communities and relies particularly on concepts developed by the PHM (Pronostics and Health Management) community. The proposed architecture is implemented and applied for monitoring a real railway transportation system within the SURFER project (SURveillance active FERroviaire) led by Bombardier-Transport.
