Decision theoretic resource management for intelligent environments

Abstract

Intelligent environments are becoming more complex, involving greater numbers of more sophisticated devices and agents. As complexity increases, it becomes difficult to build agents that effectively use the resources in their space. Agents may not know which resources are present in a given space, and they are almost certainly do not know which resources are inoperative or being used by other agents. To address this problem, this paper introduces PrefMan, a new resource management system for intelligent environments. To use this system, agents simply request a particular service type and give preferences over resource attributes. T he system keeps track of resources, their types, and their availability, and uses the preferences to intelligently allocate resources to the requesting agent. This paper describes PrefMan in detail and illustrates why it is an effective solution to the resource management problem.