Throughout the world society is demanding changes to energy systems with the overriding concerns of sustainability, and environmental and economic accountability. The electricity distribution systems that serve consumers have evolved over more than a hundred years. The supply paradigm is infrastructure-driven with a model of over-provisioning to guarantee availability at all time with very rare failure. Energy users are not “in the loop”, and often not aware that a large proportion of their electricity costs are driven not by their average usage, but by the peak usage of the grid. The capacity in the supply chain for demand response is very slow, and energy providers generate significantly more energy than is actually used by the connected consumers. This supply model is based on conventional generation in large power plants that provide energy at controlled rates; it presents a major barrier for larger-scale integration of renewable energy sources that are characterized by smaller outputs, distributed generation and high intermittency dependent on localized generation conditions.

Advances in ICT open up new avenues for innovation of how energy is supplied and consumed. ICT components and sensors have become miniaturized and give rise to pervasive embedding of sensing, computing and digital networking in the world; to the “Internet of Things” and real-time economies; and to “ambient intelligence” and digital services that blend with the everyday. Energy research programs reflect the vital importance of advances in ICT toward “smart grids”, and smart meter initiatives are an initial step toward interactivity and two-way communication between supplier and consumer. The potential of ICT for addressing the energy problem however is much larger, and as yet underexplored and underdeveloped.

The objective of this project is to develop methodologies and information and communication technologies towards future generation, user/consumption centric (“user in the loop”), implicit interaction-based energy management systems, PowerIT.

PowerIT, in proliferating ICT for activity-based implicit energy management, will assess and improve on context and activity recognition as the key methodological underpinning of implicit interaction. It will (i) design and develop a sensor ensemble management system supporting opportunistic (i.e., involving dedicated but also spontaneously available sensors), (ii) define abstract feature sets for context and activity recognition problems and show how such features can be computed from different sensor configuration, and (iii) use machine learning techniques to develop improved classification algorithms for context and activity recognition.