Producing a system allowing the protection of an information system (IS) interacting with connected objects and their management infrastructures was the objective of the thesis presented here. This system reacts in real time to the apperance of a connected object within the perimeter of the IS. To do so, the system uses a set of probes that has been spread out the IS to detect new connected objects. Once an object has been detected, a knowledge base and a linear programming model are used to determine, if any, which counter-measures must be inforced in order to protect the IS in terms of cost and performance. The knowledge base used in order to propose efficient counter-measures results in the compilation of the results of risk analysis conducted for several objects. These analyses follow the method developped during this thesis, which is based on the well known EBIOS method. Using our method, an analyst can take into account IoT specific needs as well as the interactions that exist between the IoT and an IS.