Enforcing safety in pervasive computing environments

Devices controlled by embedded software applications are finding increasing use in a variety of environments such as hospitals, health-care units, aircrafts, automobiles, and homes. Such devices may allow remote access to their operations via an Intranet or an Internet. Current efforts in the area of pervasive computing focus on the integration of such “smart” devices into these environments with little or no human assistance. Such integration creates potential for unsafe situations due to the interactions of two or more devices in the environment. The statement of this thesis is that, under certain conditions, it is feasible to enforce safety in pervasive computing environments through the (a) automatic and dynamic synthesis of safety controllers and (b) dynamic enforcement of control actions that prevent the environment from moving to an unsafe state due to the interactions among devices. The novel notions of Connected Spaces and Digital Device Manuals for modeling pervasive computing environments and devices, respectively, are introduced. The safety requirements for the environment are specified as a set of safety policies. Procedures based on control-theoretic and algorithmic techniques are designed for the automatic and dynamic synthesis of centralized and decentralized safety controllers. A safety control protocol used by the controllers for the safety enforcement is introduced. An infrastructure that provides generic capabilities for the online monitoring and control of “smart” devices is designed and built. A safety enforcement mechanism consisting of the safety controllers is implemented using this infrastructure.