A framework for threat recognition in physical security information management

In modern society, the capability to ensure an adequate level of security to persons and infrastructures is essential for the development of a territory. Malicious acts as well as adverse natural events can pose a threat to the physical security. Whatever the application domain, the protection of complex, extended and critical environments requires the development of innovative approaches to the security. They must aim at recognizing threats scenarios as early as possible, providing superior situation awareness and decision support, in order to activate a quick and focused response. The research presented in this thesis addresses that issue, on different levels. At a methodological level, by defining a general paradigm of ࢠaugmented surveillanceࢠ, thanks to information fusion strategies. At the application level, by developing a framework aimed at the automatic and early detection of threat scenarios, thanks to a model-based logical, spatial and temporal correlation of events. In order to improve the detection effectiveness and efficiency, the work introduces a heuristic situation recognition, based on ad-hoc distance metrics; and a real-time trustworthiness evaluation of the detected threat scenarios, based on uncertainty parameters characterizing sensors and models. Finally the thesis includes the application of those techniques to railway and mass-transit domain and the overall integration of the framework with an existing PSIM system.