I finished my Ph.D in the System Software Research Group (SRG)
at University of
Illinois at Urbana-Champaign. I worked on a pervasive computing project called Active Spaces. The vision of the Active Spaces project was to develop a distributed computing system from everyday computing and consumer appliances such as computers, televisions, cameras, sensors and actuators. The active space system is managed by a middleware system, called Gaia, that provides numerous distributed services and orchestrates the functioning of various components.
Commercial deployment of an active space in offices, classrooms and other organizations mandates the system to follow the guidelines set by the organization. The organization may require the active space to be functional for specific hours of the day, provide different qualities of service to different users, enforce different access rights to users and so on. Therefore, an administrative interface is necessary to encode and enforce these guidelines.
Policy-based management is an administrative approach where organization guidelines are specified using Event-Condition-Action (ECA) rules. This approach has been successfully used in managing network switches, content distributed networks and distributed systems. In my thesis, I developed a policy-based management framework for active spaces that enables an active space administrator to specify corrective actions when certain event-conditions are observed in the system. I discovered that the ECA framework provides very poor guarantees with regard to conflict detection, cycles, semantics of enforcement, exceptions and so on and I proposed an extension to the ECA model, called Event-Condition-Precondition-Action-Postcondition, for designing policy rules for an active space. ECPAP model combines axiomatic and behavioral specifications of actions with ECA rules and thus enables complex reasoning and guarantees to be provided. This framework has been used to manage active spaces and a mobile version of active spaces, called Mobile Gaia.
My Ph.D Thesis: Policy-based Pervasive Systems Management using Specification-enhanced Rules