Resilient Semantic Sensor Middleware

Resilience is the capability of a system to absorb and mitigate unexpected faults and risks. This paper describes the definition of a resilient middleware for sensor network management in dynamic environments for supporting Situation Awareness processes in security scenarios. The proposed approach is based on Quality of Service of sensors for identifying faults, disturbances, incompleteness and uncertainty in raw data. Moreover, Dempster-Shafer Theory is employed to aggregate sensor data and abstract them to obtain coherent observations about the monitored environment. Lastly, machine learning techniques are used to discover association rules in order to handle the absence of relevant observations.