An application-driven architecture for residential energy management with wireless sensor networks

This paper presents an application-driven multi-agent architecture to address the evolving use of wireless sensor networks as an enabling technology for autonomous residential energy management. Paradigm shifts in the electrical utility industry are changing the requirements for isolated residential devices. Dynamic electricity pricing rather than static flat rates requires increased automation and awareness to manage occupant comfort and price preferences. Automated load reduction programs, called Demand Response, require asynchronous control of electrical devices in response to signals from a utility provider. The architecture addresses key issues for a scalable, extensible system. A tiered design provides multiple levels of processing and communication performance and allows measurement, data synthesis, and control functionality to exist at various logical and physical locations. A Java environment embodies this architecture with multi-agent concepts to connect wireless sensor nodes, other environments, and the Internet. An implementation and deployment results show feasibility.