A widely known and promoted short-term goal EU has set is about reducing its emissions to 20% below 1990 levels by 2020. This goal is also reflected to the Challenge 6: ICT for low carbon economy and Objective ICT-2013.6.2: Data centres in energy efficient and environmentally friendly Internet of the FP7 ICT 2013 Work Programme. The GEYSER project, addressing the objectives set by the aforementioned call, aims at realizing energy-efficient data centres, which promote green energy usage both locally and at the smart grid or smart city level. To this end, the GEYSER data centres realize demand side management through flexible dynamic energy consumption, storage or disposal of energy, meant as electricity, heat, cold, and so on. GEYSER envisions data centres as conscious energy prosumers in the smart grid and smart city paradigm, providing or acquiring energy through local energy marketplaces, also developed within the project. In parallel, data centres may participate in demand response programs, offering ancillary services to the local DSO to secure the electricity grid.
The GEYSER technical solution is based on the so called MAPE approach: Monitor – Optimize – Plan – Execute. On the monitor side, the project innovates by introducing a layer of software intelligence on top of physical measurements, trading off extra cost for incremental hardware with the level of in-formation provided. Non-intrusive appliance load monitoring algorithms (NIALM) are used within GEYSER as a solution for disaggregating loads and for tracking information not possible to be provided by physical meters. The multi-criteria optimization problem requires specific search strategies capable of identifying the optimal or near optimal solutions. In consequence the project scientific team employs evolutionary techniques that combine the strength elements of different bio-inspired meta-heuristics. Such as combining population-based algorithms with trajectory-based algorithms so as to find the optimal balance between intensification and diversification aspects of the optimization problem.