ERCIM News 92

Maëlle Leroux | Download | HTML Embed
  • Jan 1, 1970
  • Views: 64
  • Page(s): 60
  • Size: 15.92 MB
  • Report

Share

Transcript

1 Number 92, January 2013 ERCIM www.ercim.eu NEWS Special theme: Smart Energy Systems Also in this issue: Keynote Smart Energy Systems A European Perspective by Ariane Sutor, Siemens AG Joint ERCIM Actions PaaSage An 8.4 Million Investment for Bridging Clouds Research and Innovation A Projector as Mobile Visualization Device on an Assistive Robot

2 Editorial Information ERCIM News is the magazine of ERCIM. Published quarterly, it reports on joint actions of the ERCIM partners, and aims to reflect the contribution made by ERCIM to the European Community in Information Technology and Applied Mathematics. Through short articles and news items, it provides a forum for the exchange of information between the institutes and also with the wider scien- tific community. This issue has a circulation of about 8,500 copies. The printed version of ERCIM News has a production cost of 8 per copy. Subscription is currently available free of charge. ERCIM News is published by ERCIMEEIG BP 93, F-06902 Sophia Antipolis Cedex, France Tel: +33 4 9238 5010, E-mail: [email protected] Director: Jrme Chailloux Smart Energy Systems ISSN 0926-4981 A European Perspective Editorial Board: Central editor: The European Union has adopted ambitious energy and cli- Peter Kunz, ERCIMoffice ([email protected]) mate change objectives to be achieved by the year 2020: Local Editors: greenhouse gas emissions should be reduced by 20% while Austria: Erwin Schoitsch, ([email protected]) the share of renewable energy is to be increased and energy Belgium:Benot Michel ([email protected]) efficiency to be improved by this amount. Furthermore, the Cyprus: George Papadopoulos ([email protected]) EU has made a long term commitment to cut emissions by at Czech Republic:Michal Haindl ([email protected]) least 80% by 2050. France: Thierry Priol ([email protected]) Germany: Michael Krapp ([email protected]) So far the current strategies are unlikely to achieve all 2020 Greece: Eleni Orphanoudakis ([email protected]), targets; whilst the goal to increase the share of renewables Artemios Voyiatzis ([email protected]) seems to be attainable, we are likely to fail on the improve- Hungary: Erzsbet Csuhaj-Varj ([email protected]) ment of efficiency. Also, the strategies in place seem inade- Italy: Carol Peters ([email protected]) quate to meet the longer term challenges. Use of renewable Luxembourg: Patrik Hitzelberger ([email protected]) energy is still limited because the development of efficient Norway:Truls Gjestland ([email protected]) energy transport is lagging. The internal energy market is still Poland: Hung Son Nguyen ([email protected]) fragmented and needs improvement regarding transparency, Portugal:Joaquim Jorge ([email protected]) accessibility and choice. Fragmentation of national regulation Spain: Silvia Abraho ([email protected]) limits companies to act on multi-national markets and sets Sweden: Kersti Hedman ([email protected]) barriers to fair competition. The EU must take urgent action Switzerland: Harry Rudin ([email protected]) to select the right tools to make the energy change happen and The Netherlands: Annette Kik ([email protected]) to move towards a sustainable energy future. United Kingdom: Martin Prime ([email protected]) W3C: Marie-Claire Forgue ([email protected]) The current situation exhibits a number of major challenges. Large-scale renewable energy generation is currently largely Contributions based on wind, and plants may be installed far from existing Contributions must be submitted to the local editor of your country power infrastructure, thus connection to the grid can present a challenge. Owing to the inherent variability and weather Copyright Notice dependence of energy production and the limited capacity All authors, as identified in each article, retain copyright of their work for energy storage, balancing supply and demand becomes also far more difficult. Advertising For current advertising rates and conditions, see The deployment of distributed energy resources can mitigate http://ercim-news.ercim.eu/ or contact [email protected] the uncertainties connected to large plants and at the same time exploit the potential of dispersed resources, although ERCIM News online edition the integration of distributed energy resources is extremely The online edition is published at http://ercim-news.ercim.eu/ challenging both from a market and a technical point of view. Decentralization supports scalability and robustness and Subscription facilitates access to the energy market for prosumers, thereby Subscribe to ERCIM News by sending email to [email protected] or by filling out the form at the ERCIM News website: http://ercim-news.ercim.eu/ Next issue April 2013, Special theme: Mobile Computing ERCIM NEWS 92 January 2013

3 Keynote Ariane Sutor Siemens AG and distribution system operators respectively. The European EIT ICT Labs Smart Energy Institute of Innovation and Technology (EIT) is an institution Systems Action Line Leader of the EU with the mission to increase European sustainable growth and competitiveness by reinforcing innovation capacity. Within EIT ICT Labs the action line Smart Energy Systems drives European innovation for smart energy sys- tems regarding user involvement, business models and ICT- supporting changes to consumer behaviour and social enabled infrastructures and mobilizes a strong network of acceptance. It is imperative that consumers understand and industrial partners, research institutes and technical universi- trust the process and receive clear benefits, eg energy sav- ties. ings, more transparent billing and a business case for electric vehicles, heat pumps and smart appliances. Demand A large number of European-level research, demonstration response is also a central theme where consumers load and deployment projects focusing on the development of reductions are aggregated to offer flexible services to other smart grids are currently underway. Most projects are sup- stakeholders in the electricity system. ported by FP6 and FP7 but many, including the Portuguese National Strategic Reference Framework (QREN), the One key to the integration of intermittent and dispersed Spanish Centre for Industrial Technology Development renewable energy and to increased energy efficiency is the (CDTI) and the German funding program E-Energy, also introduction of Smart Grids. In addition to building new lines benefit from substantial national co-funding. A major part of and substations, it is essential to make the electricity system future funding will be included in the upcoming Horizon smarter through the introduction of Information and 2020. The European Commission funds a whole series of dif- Communication Technologies (ICT). Smart Grids can be ferent issues concerning the implementation of smart grid described as advanced electricity networks enabling two- technologies. Large investments are foreseen to extend way exchange of power and information between suppliers existing grids to cope with the intermittent nature of renew- and consumers based on the introduction of intelligent com- able sources, eg new lines, additional capacities but also bal- munication, monitoring and management systems. An open ancing area extension, re-designed market mechanisms and and secure ICT infrastructure is at the core of the implemen- storage integration. Research projects are addressing the tation of the Smart Grid. Interoperability, data privacy and problem of distributed energy resources, with a focus on security thereby play a crucial role. Currently a convergence online coordination of distributed generators and storage towards proven communication standards and industry best devices such as electric vehicles to enhance grid stability and practices is observed, eg Internet Protocol communication. optimization of energy resources. For smart grids to exhibit their full potential the realization of Smart, sustainable and inclusive growth for Europe includes physical infrastructures is needed as well as new business a shift in our energy policy leading to increasing renewable models and regulations. There is a critical need to adopt a resources and improved energy efficiency. Smart Grids are European energy policy to overcome the fragmentation the key enabler to achieve the policy objectives and to main- related to national and regional policies. Management bodies tain a leading European technological and competitive posi- at EU level include the Directorate Generals for Energy and tion. In addition to new technologies, real pan-European reg- for Research (DG ENER, RTD) and the Joint Research ulations and markets are required. To this end, it will be para- Centre JRC.A recent update of the European Strategic mount that partners from industry, research institutes and Energy Technology Plan (SET Plan) describes the strategy to universities continue to join forces. accelerate the development and deployment of cost-effective low carbon technologies. Other related bodies include: the Ariane Sutor Smart Grids Task Force SGTF on European regulation and standardization, the Smart Grids European Technology Platform ETP providing the European strategic research agendas SRA 2020 and 2035 as well as industry driven con- sortia like ENTSO-E and EDSO of European transmission ERCIM NEWS 92 January 2013 3

4 Contents SPECIAL THEME This special theme section Smarty Energy Systems has been 2 Editorial Information coordinated by Carl Binding, IBM Research Lab, Switzerland and Han La Poutr, CWI and Utrecht University, the Netherlands KEYNOTE Introduction to the special theme 3 Smart Energy Systems A European 8 Smart Energy Systems Perspective by Carl Binding and Han La Poutr by Ariane Sutor Invited articles 10 Development of the European Virtual Smart Grid JOINT ERCIM ACTIONS Laboratory by Kai Strunz and and Christian Wiezorek 6 PaaSage An 8.4 Million Investment for Bridging Clouds 11 Forecasting the Conditional Dynamic Elasticity of Electricity by Pierre Guisset Consumers by Pierre Pinson and Henrik Madsen 6 Formal Methods for Intelligent Transporta- tion Systems - A track at ISOLA12 12 Putting Neurons in the Smart Grid by Alessandro Fantechi, Francesco Flammini by Bram Vonk, Robert de Groot and Han Slootweg and Stefania Gnesi 7 Pawe Parys Winner of the 2012 Cor Baayen IT for smart grids Award 14 Smart Energy Consumption Feedback Connecting Smartphones to Smart Meters by Markus Weiss, Friedemann Mattern and Christian Beckel 15 Meter Data Management for Smart Monitoring Power Networks by Agustn Yage, Juan Garbajosa and Mercedes Lopez-Perea 17 Designing and Simulating Smart Grids by Jennifer Prez, Jessica Daz and Eloy Gonzlez 18 Smart Management of Renewable Energy for Green Transport by Raffaele Bruno, Luca Valcarenghi, Molka Gharbaoui and Barbara Martini 20 Real-Time Visualization of MV/LV Energy Alarms on GIS Web Applications by Christophe Joubert, Vicente Monrabal, Miguel Montesinos and Carlos Snchez 21 Using Wireless Smart Meter Networks for Power Quality Monitoring Joel Hglund and Stamatis Karnouskos Supply/demand coordination 22 Smarter Energy: Opportunities and Challenges by Olle Sundstrm, Fabian Mller, Carl Binding, Bernhard Jansen and Dieter Gantenbein 24 Smart Demand-Side Response at Home by Armin Wolf, Thomas Luckenbach and Mario Schuster 25 Market Garden: A Scalable Research Environment for Heterogeneous Electricity Markets by Felix Claessen, Nicolas Hning, Bart Liefers, Han La Poutr and Peter Bosman 27 The Power Trading Agent Competition by Wolfgang Ketter and John Collins 4 ERCIM NEWS 92 January 2013

5 28 A Model-Free Flexibility Management System at KU Leuven and VITO by Stijn Vandael, Bert Claessens, Tom Holvoet and Geert Deconinck RESEARCH ANd INNOVATION 29 Demand Side Management for Multiple Devices by Albert Molderink, Vincent Bakker and Gerard J.M. Smit This section features news about research activities and innovative developments from 31 The Gamification of Agent-Based Smart Grids European research institutes by Radu-Casian Mihailescu, Matteo Vasirani and Sascha Ossowski 48 A Projector as Mobile Visualization Device 32 A Marketplace-Based Approach to Demand-Side Management on an Assistive Robot in the Smart Grid by Paul Panek, Christian Beck, Georg Edel- by Luigi Briguglio, Massimiliano Nigrelli, Frank Eichinger, Javier mayer, Peter Mayer and Wolfgang L. Zagler Lucio Ruiz-Andino and Valter Bella 49 ATLAAS-P2P: A Two-Layer Architecture for Approximated Search in Peer to Peer Smarter buildings by Ranieri Baraglia, Patrizio Dazzi, Matteo 34 Using an Intelligent Management System for Smart Mordacchini and Laura Ricci Residential Storage Systems by Vicente Botn, Mximo Prez, Adolfo Lozano-Tello and 50 Epeerdemics: A Peer-to-Peer Simulator Tar- Enrique Romero geting Epidemic-Based Protocols by Patrizio Dazzi and Emanuele Carlini 35 The Last One out Turns off the Light - Optimizing the Energy Efficiency of Buildings 51 gRecs: Exploiting the Power of Data Mining by Lutz Ehrig and Danilo Hollosi Techniques for Efficient Computation of Group Recommendations 36 Ambient Intelligence for Energy Efficiency in a Building by Kostas Stefanidis and Kjetil Nrvg Complex by Giuseppe Lo Re, Marco Ortolani and Giuseppe Anastasi 52 Utility-Theoretic Ranking for Semi-Auto- mated Text Classification Security by Giacomo Berardi, Andrea Esuli and Fabrizio 38 Secure Smart Grids or Say Goodnight Vienna! Sebastiani by Florian Skopik, Paul Smith and Thomas Bleier 53 A Radio Telescope of the Superlative 39 Preparing for the Smart Grids: Improving Information by Ton Engbersen Security Management in the Power Industry by Maria Bartnes Line 55 The Green-Wake Project Targets both Air Traffic Security and Airport Throughput 40 Cybersecurity in the Smart Grid by Sbastien Lugan and Benoit Michel by Magnus Almgren, Davide Balzarotti, Marina Papatriantafilou and Valentin Tudor EVENTS, IN BRIEf 41 CoppEnd A Security System for Power Equipment by Dimitrios Serpanos, Athanasios Safacas and Dimitrios 56 International Workshop on Information Stachoulis Technology for Energy Applications by Paulo Carreira and Vasco Amaral Green IT 43 PowerAPI: A Software Library to Monitor the Energy 57 ERCIM/EWICS/Embedded (Cyber-Physical) Consumed at the Process-Level Systems Workshop 2012 by Aurlien Bourdon, Adel Noureddine, Romain Rouvoy and by Erwin Schoitsch Lionel Seinturier 57 Announcements 44 Smart Energy Management for Greener Supercomputing by Mohammed el Mehdi Diouri, Olivier Glck and Laurent Lefvre 59 In Brief Novel energy forms 45 AI4B: Accountable IT Infrastructures for Optimizing Supply Chains in Bioenergy Symbiotic Networks by Theodore Dalamagas and Antonis Kokossis 47 Hydrodynamics-Biology Coupling for Algae Culture and Biofuel Production by Olivier Bernard, Jacques Sainte-Marie, Bruno Sialve and 5 ERCIM NEWS 92 January 2013 5

6 Joint ERCIM Actions PaaSage formal Methods An 8.4 Million Investment for Intelligent for Bridging Clouds Transportation Systems - by Pierre Guisset A track at ISOLA12 ERCIM, together with 13 European partners, has launched by Alessandro Fantechi, Francesco Flammini and PaaSage, a major research initiative with the goal of Stefania Gnesi developing an open and integrated platform to support model based lifecycle management of Cloud applications. The ERCIM Working Group on Formal Methods for Industrial Critical Systems (FMICS), organised a track at Cloud solutions are currently still insufficient and require a the 5th International Symposium on Leveraging high level of expertise on the part of the developer and the Applications of Formal Methods, Verification and provider to properly exploit the capabilities offered by Cloud Validation (ISOLA12), held in Crete on 15-18 October technologies. Cloud infrastructures are not standardized and 2012, to address the application of formal methods to porting an existing application to a Cloud platform is still a model and analyze complex systems in the context of very challenging task, leading to a strong interdependence Intelligent Transportation Systems. between the client application and the Cloud platform. Developing once and deploying on many Clouds is not a The term Intelligent Transportation Systems (ITS) refers viable proposition as things stand. This is the challenge that to Information and Communication Technology as applied to the PaaSage consortium will address. PaaSage will deliver transport infrastructure and vehicles with the aim of an open and integrated platform to support both design and improving transport outcomes such as transport safety, trans- deployment of Cloud applications, together with an accom- port productivity, travel reliability, informed travel choices, panying methodology that allows model-based development, social equity, environmental performance and network oper- configuration, optimisation, and deployment of existing and ation resilience. ITS is becoming increasingly important as new applications independently of the existing underlying novel driverless/pilotless applications are emerging. Cloud infrastructures. Based on discussions held by the ERCIM Working Group on PaaSage will provide the relevant means to significantly Formal Methods for Industrial Critical Systems (FMICS), a improve programmability, usability and performance of track was organized for the ISOLA12 Conference to address Clouds beyond current state of the art approaches, says the application of formal methods to model and analyse com- Keith G. Jeffery, scientific coordinator of the project. We plex systems in the context of ITS. In fact, modelling and have to admit that European industry is lagging behind in analysis activities are very important to optimize system life- business creation and development on the basis of Cloud cycle in the design, development, verification and opera- computing technologies, adds Pierre Guisset, project coor- tional stages, and they are essential whenever assessment dinator, Our objective with PaaSage is to develop the tools and certification is required by international standards. that will enable European small and large businesses to take a leading position in exploiting Clouds. Typically a business The contributions to the theme Formal Methods for will be developing its in-house server cluster to an in-house Intelligent Transportation Systems addressed three distinct Cloud to obtain benefits of elasticity and eco-friendliness. aspects. The first was a general perspective on the introduc- However, when elasticity needs to extend beyond the in- tion of formal methods in the development process of house environment to a public cloud there are interoper- safety-critical systems focusing on model-driven verifica- ability problems and provider proprietary solution con- tion techniques, both for functional and non-functional straints. These will be overcome by PaaSage. system properties. In particular, the expected impact of a novel software development guideline for safety-critical PaaSage is a collaborative research project co-funded under systems (namely, the avionic DO178-C standard) was eval- the ICT theme of the 7th framework programme of the uated. For the first time, that standard specifically includes European Union. In particular, PaaSage addresses the find- formal methods as one of the preferred development and ings highlighted by the Commissions Cloud Computing verification techniques. Expert Working Group. The total investment amounts to 8.4m, of which 6.3m is funded by the European Union. The other contributions related to two categories of systems PaaSage started on 1 October 12012 and will last 4 years. in the railway domain, which, owing to their complexity, pose several challenges to current software and system Links: development techniques. The first category addresses driver- http://www.paasage.eu/ less metros that integrate several subsystems within complex http://cordis.europa.eu/fp7/ ict/ssai/docs/future-cc-2may- architectures, which are geographically distributed, featuring finalreport-experts.pdf strict dependability requirements. The second category is represented by railway interlocking systems: here the com- Please contact: plexity lies in the geographical layout of the tracks, points Pierre Guisset, ERCIMOffice, and signals that can be found in stations or in railway yards. E-mail: [email protected] Within this context, a particularly challenging problem for 6 ERCIM NEWS 92 January 2013

7 current model-checking technology is to automatically We believe that, despite the limited space available in the verify that interlocking logics designed by railway engineers program of the track, the contributions succeeded in giving a actually satisfy safety properties (eg, two trains will never be good overview of the state-of-the-art and of the hard-to-solve assigned conflicting routes) for medium to large stations. open issues, as well as proposing significant directions for the future research in this field. The lively discussion that took place during the workshop on railway interlocking systems touched on topics such as: Link: http://www.cs.uni-potsdam.de/isola/isola2012/ possible ad hoc optimisations for symbolic model-check- ing, by means of specific re-orderings of BDD (Binary Please contact: Decision Diagrams) variables; Alessandro Fantechi, Universit di Firenze, Italy logical and physical distribution of the interlocking con- E-mail: [email protected] trol logics; the increasing complexity due to the integration and inter- Francesco Flammini, Ansaldo STS, Italy face of interlocking systems with other signalling systems; E-mail: [email protected] verification of legacy relay-based interlocking systems which are strictly related to the topology of the controlled Stefania Gnesi, ISTICNR, Italy layout. E-mail: [email protected] Pawe Parys Winner of the 2012 Cor Baayen Award ERCIM has selected Pawe Parys from the University of Warsaw as the winner of the 2012 Cor Baayen Award for a promising young researcher in computer science and applied mathematics. In a tight competition Pawe Parys was chosen from 21 excellent candidates from all over Europe as an exceptionally gifted young researcher. Pawes main fields of interest are algorithms and computa- tion theory. In his young career, Pawes research has already ERCIMpresident Keith Jeffery (left) presents the 2012 Cor Baayen led to two extraordinary results in very different fields such award to Pawe Parys. as linear time evaluation of XPath and higher-order push- down automata. correctly for all possible inputs? can be recast as a question about the language accepted by a deterministic higher-order The XPath (a language for addressing parts of an XML docu- pushdown automaton. One of the most important open prob- ment) evaluation problem is important in practice, in partic- lems in this area was about the equivalence of two models: ular for the performance of Web browsers which generally deterministic higher-order pushdown automata, and deter- include XPath evaluators. These evaluators are often quite ministic higher-order pushdown automata with the so-called inefficient, in many cases running in exponential time (in collapse operation. It has been conjectured that the collapse relation to the size of the query). In his groundbreaking work, operation gives more power to the automata. In a series of Pawe devised a system for evaluation of XPath in time papers, including two STACS papers and a paper accepted linear to the size of the document. Until recently, this was by LICS 2012, Pawe Parys has proved this conjecture. This widely believed to be impossible. The improvement is dra- settles an important, and technically difficult, open problem. matic for XML documents with billions of nodes, such as the DBLP database, for example. Although Pawe Parys just finished his PhD in January 2012, his work has already been recognized with best student paper In his research on higher-order pushdown automata, Pawe award at PODS 2009 (the top database theory conference), could prove that collapse increases the power of higher-order the Witold Lipski Prize for Young Researchers in Computer pushdown automata, which was a major open problem Science 2011 (an award for the best young Polish computer (pushdown automata are used for instance in theories about scientist). He was also the recipient of a "Start" stipend for what can be computed by machines). Higher-order push- young researchers awarded by the Foundation for Polish down automata are like pushdown automata, except that they Science in 2012 as well as several achievements in program- can use stacks of stacks, or stacks of stacks of stacks, and so ming competitions. on. They appear naturally when modelling the behaviour of recursive higher-order programs in functional languages. For more information about the ERCIM Cor Baayen The question does a recursive higher-order program behave Award, see http://www.ercim.eu/activity/cor-baayen-award ERCIM NEWS 92 January 2013 7

8 Special Theme: Smart Energy Systems Introduction to the Special Theme Smart Energy Systems by Carl Binding and Han La Poutr Modern, industrialized, society is heavily dependent on ubiquitous, cheap energy, which we expect to be readily available, not to be polluting, and to be convenient to use. Since the invention of the steam engine by James Watt, this paradigm has lead towards tremendous improvement of life quality in the developed world, and developing countries eagerly aspire to similar energy standards. However, the price of this hunger for energy is increasing. Fossil fuel resources such as oil or gas are becoming harder to explore, even leading to environmental disasters as with the Gulf of Mexico oil-platform, recently. Exploration of shale gas (fracking) causes negative environmental impact, beyond the well-known CO2 problematics. Besides the sheer availability of fossil energy, associated CO2 emis- sions have caused wide-spread concerns about impacts on climate and on human health (fine particle emissions). Another kind of fossil fuel energy source, namely uranium- based nuclear fission, has been explored for about 50 years. It is CO2 neutral, but has additional risks and thus costs asso- ciated with its operations. More importantly on a longer time scale, the nuclear waste disposal has not yet found a satisfac- tory answer. Furthermore, nuclear fission is a large consumer of water for reactor cooling, which is considered a negative environmental impact. It is against this background that concerned citizens, indus- tries, and their political representatives have shown increased interest in the use of renewable1, natural, energy resources. These are mainly represented by solar energy, be it in the form of photo-voltaic or simple heat, and by wind power. Other energy forms are based on bio-mass, wave energy, and of course the traditional hydro-power electrical generation plants. 1The term renewable being somewhat imprecise as it is the solar fusion process which has a long, but finite, life time, that keeps the wind blowing and the sun shining. 8 ERCIM NEWS 92 January 2013

9 An important challenge of wind and solar energy is their sto- Europe's North all the way to its far South and even chastic nature. Traditionally, power generation has been Northern Africa (spanning 24 countries and 400 million planned based on predictions of the aggregated energy con- users) cannot be rebuilt in a short time, given the tremen- sumption (load). Since wind and solar energy are hard to dous costs involved and given todays often vocal opposi- plan, their availability is uncertain. Hence, this fundamen- tion to power grid projects in many countries. Simulation, tally changes the traditional electrical power engineering addressed in some contributions, thus also plays a role in equation of generation follows load into the challenge of planning and designing the future power grid. load must follow generation. The ultimate power users, however, remain individuals. In addition, new devices are appearing more and more in our How can we motivate humans to use energy in a more effi- energy ecosystem, which will influence its characteristics. cient way? We have included some papers describing how to For example, electric vehicles and heat pumps are character- make the end-user more aware of his or her power usage and ized by heavy power consumption exceeding those of past the use of game theory to effectively tie the end-user into the domestic devices, but at the same time these new devices energy markets of the future. The challenge will be to make allow for the storage of energy. And micro-CHPs (combined the end-consumer price conscious for a commodity whose heat and power) allow for efficient generation of power in generation and transport only costs percents of a familys addition to heat. These developments give new control prob- budget. lems to be tackled, but also open up new ways of handling our power systems. Use of bio-mass is addressed in an interesting project which uses algae for energy production as well as the issues arising In this special theme of ERCIM News, we are happy to with wide-spread exploration of bio-mass. present a wide spectrum of ongoing research activities which attempt to address today's energy dilemma. The challenges ahead are huge. Fundamental science has found solutions to many of the smarter energy issues. It is Better building design and technology is seen as a path for- the large scale deployment in a cost-effective, safe, and ward to further decrease the volume of energy consumed environmentally sound manner which has to be addressed. always a good idea when a resource becomes scarce. In the Ultimately, the choice is ours: how much are we willing to same spirit, we see approaches to make better use of trans- reign in, respectively reconsider, our energy household of portation means by giving the end-user tools to select a con- the future and how much are we ready to spend on more venient yet ecological means of transportation. effective use of energy. The intelligent power grid is another large theme. By this, We hope that several of the projects presented here can make we understand a more optimal usage of existing power grid a contribution on the route to a sustainable energy system, resources to minimize waste as well as to align demand with which is, as the British physicist David JC MacKay terms it, renewable energy generation. This amounts not only to a without the hot air. large control and optimization problem, in the form of market-based approaches and automated demand side man- agement, but needs to be associated with a large communi- Please contact: cations and computer infrastructure. This, in turn, causes Carl Binding ample security concerns summarized under the notion of IBM Research Lab, Switzerland cyber security which is addressed by several of the E-mail: [email protected] ongoing projects. Han La Poutr A system as large as the electrical power grid an example CWI and Utrecht University, the Netherlands of which is the European ENTSO-E grid which ranges from E-mail: [email protected] ERCIM NEWS 92 January 2013 9

10 Special Theme: Smart Energy Systems development of the European Virtual Smart Grid Laboratory by Kai Strunz and and Christian Wiezorek The transition towards increasingly renewable energy systems calls for novel techniques of operation and control in response to the changing power transmission and distribution networks. A Smart Grid is expected to efficiently manage supply and demand of electricity. Electric networks will become more intelligent, bringing the worlds of IT, communications and energy systems closer together than ever before. The EIT ICT Labs, one of the Knowledge and Innovation Communities (KIC) at the European Institute of Innovation and Technology (EIT), are strongly supportive of work in this area. Founded in 2008, EIT aims to create synergies between education, research, and innovation. It promotes the systematic development of international networks and clusters of excellent institutions, universities and industrial research centres in Europe. Six nodes of the EIT ICT Labs across Europe coordinate different thematic topic areas. The action line Smart Energy Systems located at the Berlin Node addresses the above-mentioned synergies of ICT and energy systems. One of the activities coordinated within terms of information exchange and per- Research Institute for Mathematics and Smart Energy Systems is the sonnel involvement (see Figure 1). Computer Science (CWI), the German European Virtual Smart Grid Laboratory Involved are highly renowned universi- Research Center for Artificial (EVSGL). This activity develops a vir- ties such as KTH Stockholm, Imperial Intelligence (DFKI), and several insti- tual lab integrating the expertise and College London, TU Delft, TU Berlin tutes of the Fraunhofer Gesellschaft. capacities of each partner s lab and excellent research centres such as Industry partners, including Ericsson resources, thereby creating a platform the Technical Research Center of and Siemens, also participate. Each with a wide-ranging application field in Finland (VTT), the Dutch National partner contributes its expertise in its particular field to joint projects. As a Figure 1: the key project, the setup of the virtual lab- European Virtual oratory consists of three sequential Smart Grid stages: Laboratory Initialization stage: In the first stage, the fields of competences of each part- ner are discussed and put into context with those of the other partners to identify complementary expertise. Connection stage: During the second stage, the technical groundwork for the connection of the involved labs is laid. Demonstrators prove the functionality. At the end of this stage, first experi- ments at the labs can be observed by partners from remote sites. Implementation stage: Within this stage, the successfully implemented network will be used to run the Virtu- al Smart Grid Lab. The European Virtual Smart Grid Lab activity is led by the Chair of Sustainable Electric Networks and Sources of Energy (SENSE) at TU Berlin [1]. At SENSE, the Smart Grid Laboratory at TU Berlin has been estab- lished, integrated within EIT, and first presented at the 3rd IEEE PES Innovative Smart Technologies (ISGT) Europe 2012 conference from 14 to 17 October 2012 in Berlin [2]. The labora- tory comprises real power hardware with a low voltage distribution feeder integrating different load, storage, and generation components (see Figure 2). Figure 2: the Smart Grid Laboratory at TU Berlin Various smart control schemes are 10 ERCIM NEWS 92 January 2013

11 applied. This includes operation in the Figure 3: the form of a microgrid. In addition, a real- physical structure time digital simulator serves for repro- of the lab ducing a possible connection to the transmission grid. A main scope of study of the TU Berlin Smart Grid Lab is the integration of renewable energy and electromobility into the power system. The physical structure of the lab is shown in Figure 3. The plug-and-supply interfaces to which resources can be connected are marked in green. Starting from the feed-in, the ring connects a solar photovoltaics (PV) simulator generator unit. The protective container analogies derived from computer shown in blue. It consists of a commer- allows for the integration of battery, engineering, IEEE Transactions on cially available inverter and a solar super capacitor, and cache control Power Systems, Vol. 24, No. 1, pp. 12- panel emulated by a controllable elec- testing [3], [4], as indicated in purple. 19, February 2009 tronic amplifier. Loads are shown in As a part of the Virtual European Smart [4] H. Louie, K. Strunz: yellow. A single family home and an Grid Laboratory, the installation will Superconducting magnetic energy apartment building consist of both off- support the development of Smart Grid storage (SMES) for energy cache the-shelf appliances such as dish- solutions in Europe and beyond [2]. control in modular distributed washers, refrigerators, and boilers, and hydrogen-electric energy systems, electronic loads. A small industrial unit References: IEEE Transactions on Applied is represented by different electronic [1] TU Berlin SENSE: Superconductivity, Vol. 17, No. 2, pp. loads capable of providing machinery http://www.sense.tu-berlin.de 2361-2364, June 2007 load profiles. Two electric vehicle-to- [2] IEEE PES ISGT Europe 2012: grid (V2G) charging stations are shown http://www.ieee-isgt-2012.eu Please contact: in cyan, one of which is for e-bikes. The [3] K. Strunz, H. Louie: Cache energy Kai Strunz simulator for a distributed wind energy control for storage: Power system Technische Universitt Berlin, Germany conversion system consists of a motor- integration and education based on E-mail: [email protected] forecasting the Conditional dynamic Elasticity of Electricity Consumers by Pierre Pinson and Henrik Madsen In the development of smarter energy systems, it is vital that we maximize flexibility of consumer demand. To this end, it will be of utmost importance to be able to predict the potential of electricity demand to respond dynamically to varied signals. Some countries aim to be deriving In contrast to large industrial con- Regardless of how demand flexibility is almost all of their power from renewable sumers, for which direct bilateral to be enhanced at the household level sources in the relatively near future agreements may be made and used on (electric heating, cooling, electric vehi- (Denmarks timeframe, for instance, is an ad-hoc basis, domestic consumption cles, etc.), identifying intelligent ways 2050). With greater integration of is a far greater challenge to manage to alter demand patterns is a stochastic renewable energy generation, demand owing to the large number of indi- optimization or control problem, com- flexibility will become ever more impor- vidual households, their distribution, prising a whole challenge in itself. This tant in supporting smart energy systems. the state of the art in ICT (Information question will depend, to some extent, This will translate to a paradigm shift, and Communication Technologies), upon the time scales considered (and from a system where demand drives the effectiveness of economic incen- corresponding mathematical formula- generation to a system where renewable tives, behavioural effects, etc. A tion), engineering considerations - for energy generation may influence number of research and demonstration instance related to ICT capabilities, but demand patterns. In practice this projects are investigating these factors, also on philosophical aspects of design. requires enhancing, and taking full including the iPower project in The two main approaches currently advantage of, the potential flexibility of Denmark, funded by the DSR-SPIR- under study are (i) direct distributed all electricity consumers, including program (project number: 10-095378, control, and (ii) the indirect control domestic households. see link below). approach based on price signals. The ERCIM NEWS 92 January 2013 11

12 Special Theme: Smart Energy Systems latter takes advantage of the elasticity of inefficient solution compared with consumers, ie the adaptation of con- alternatives, such as using storage or sumption in response to varying elec- expensive conventional generators, pos- tricity prices. Price signals are to be sent sibly even magnifying the fluctuations daily for optimal task assignment (bulk that we are aiming to dampen. Ideally, heating, washing machines, etc.), but these predictions should be of proba- also adapted in real time so as to take bilistic nature, in the form of scenarios, corrective action supporting the optimal so as to fully describe the range of matching of generation and consump- potential responses from the aggrega- tion. In this indirect control by price tion of household consumers to be setup, the stochastic optimization or influenced. control problem translates to issuing Figure 1: A conceptual view of the balance optimal price signals to be broadcast to between electricity production and Link: http://ipower-net.dk groups of consumers whose consump- consumption, as influenced by the external tion levels are to be influenced. environment and controlled through a price References: generator. [1] O. Corradi et al: Controlling elec- With this objective in mind, the core, tricity consumption by forecasting its and most crucial, aspect is to identify tric vehicles need to be charged at some response to varying prices, IEEE Trans. and be able to predict how small con- point before they are to be driven, while Power Syst., May 28, 2012, sumers respond to varying prices. We households need to be heated so as to http://dx.doi.org/ 10.1109/ refer to this as the conditional dynamic keep indoor temperature at an accept- TPWRS.2012.2197027 consumer elasticity. It is conditional able level. [2] D. Hammerstrom, et al: Pacific since the potential to affect the timing, NorthWest GridWise Testbed demon- and maybe even the magnitude, of the As a final point, this conditional stration projects Part I: Olympic flexible part of the load is clearly a dynamic elasticity of electricity con- Peninsula project, Tech. Rep., Pacific function of external conditions. If con- sumers may smoothly evolve with time, NorthWest National Laboratory, sidering space heating for instance, out- owing to changes in consumption pat- PNNL-17167, October, 2007 door temperature, as well as the settings terns, appliances and their functionali- [3] J. Torriti, M.G. Hassan, M. Leach: of the local heat controller, will directly ties, etc. As a consequence, one needs to Demand response experience in impact the potential demand response to employ a bottom-up approach and use Europe: Policies, programmes and prices. Similarly in the case of electric empirical data for the identification of implementation, Energy, 35(4), pp. vehicles, the demand response potential appropriate models, adaptive estimation 15751583, 2010, http://dx.doi.org/ will vary as a function of the time of the of their parameters, and continuous 10.1016/j.energy.2009.05.021 day when more or less electric vehicles monitoring of forecast quality. The may be plugged in and their batteries quality of such forecasts will be para- Please contact: made available for demand response. In mount since this data will directly Pierre Pinson, Henrik Madsen parallel this response is dynamic as impact the reliability of potential Technical University of Denmark most consumption patterns cannot be demand response. An unreliable E-mail: [email protected], deferred indefinitely: batteries of elec- demand response would make this an [email protected] Putting Neurons in the Smart Grid by Bram Vonk, Robert de Groot and Han Slootweg The collaboration between Eindhoven University of Technology and Dutch Distribution System Operator, Enexis, results in a direct implementation of scientific ideas and models in real world systems and faster feedback of data for analysis or validation. This is illustrated by the Smart Storage Unit: A grid connected battery in a residential area of 240 houses including photovoltaic generation and heat pumps, controlled by an artificial neural network based forecaster. The Smart Storage Unit (SSU) is the hours of the installed watt-peak PV In order to achieve these goals, the SSU result of a pilot project for development output in the neighborhood. The goals is equipped with an advanced control and deployment of a centralized storage set for the Smart Storage unit include: system capable of controlling battery unit in the low voltage (LV) distribution Increase of self-consumption (of the state-of-charge conform pre-specified network. The system is built around a photovoltaic (PV) generated energy) objectives and conditions (Figure 1). lithium-ion battery system consisting of Increase of reliability (in autonomous four separate strings, each with a operation acting as a back-up unit) The SSU is installed into an LV-grid in capacity of 58 kWh, having a total Maximization of utilization of local the Etten-Leur area, part of the Enexis capacity of 232 kWh and a nominal bat- infrastructure (common feed-in at distribution grid in the South of the tery voltage of 720 V. the total system is peak consumption of the households: Netherlands. The network connects 240 capable of storing approximately 1.2 peak shaving) households, of which 40 have locally 12 ERCIM NEWS 92 January 2013

13 information is the amount of reduction of uncertainty of one signal by another signal. In this case, the latter is a pos- sible input variable of the system and the former is the output of the fore- caster. An advantage of this approach is that it also recognizes non-linear rela- tions between signals. To avoid selec- tion of variables that contain the same information (which may be encoded differently) the partial mutual informa- tion is used. With this approach several weather forecast variables are selected as inputs as well as the realization of energy demand from the previous day. At the start of every forecast the system looks into a database containing data from the last two months, and selects the five sets of training data with the input that most resembles the current Figure 1: Schematic view of the grid with the SSU-1 input [2]. These sets are then used to train the ANN and specialize it to the current situation to make an accurate load forecast [3]. The developed forecaster will be imple- mented in the near future in the SSU. In the current version a simpler control algorithm is implemented, but the SSU is designed in such a way that new con- troller components can be installed, tested, evaluated and improved without much effort by using the SSUs internet Figure 2: the SSU in the residential area connection. References: installed PV systems with a total output caster with an artificial neural network [1] H.K. Alfares and M. Nazeeruddin: of 186 kWpeak. The LV-grid is con- (ANN). These mathematical networks Electric load forecasting: literature nected to the upstream medium voltage consist of nodes called neurons that col- survey and classification of methods, (MV) network through a 400 kVA trans- lect and sum incoming signals from International Journal of Systems former (0.4 / 10 kV), with an average other neurons and evaluate the result by Science, vol. 33 peak-load measured at 385 kW. a transfer function [1]. The advantage of [2] T.M. Peng et al: Advancement in using an ANN is that the forecaster can the application of neural networks for To ensure proper operation, the SSU be trained online by using previous short-term load forecasting, IEEE should be able to predict the future energy demands and matching input Trans. Power Syst., vol. 7 power demand of the residential area, data (eg wind, temperature, solar irradi- [3] B.M.J. Vonk et al: Improving since that determines when and with ation) from the residential area. In this short-term load forecasting for a local which rate the battery is charged and way, the forecaster adapts itself to a new energy storage system, in proc. UPEC discharged. This is not only necessary situation in less than two weeks. On a 2012 for peak shaving operations or for the neuron level this training means simply optimal use of locally generated energy, adjusting the connections between the Links: but also for autonomous operation when neurons. Since energy consumption on http://energyacademy.org/article/69/sm the SSU is disconnected from the main Saturday and Sunday is different than art-storage-to-counteract-intermittency- grid. A fully charged battery system is that of workdays, two different ANNs problems not optimal in this case, since the resi- are used for the weekend and workdays. http://www.emforce.nl/index.php?page dential area contains so many PV panels =smart-storage-eng that in summer there are times when An important issue regarding ANNs is energy is in surplus. With a full battery the selection of appropriate input vari- Please contact: this would cause the PV inverters to ables and training sets. The input vari- Bram Vonk shut down and the irradiated solar able selection for the forecaster is based ENEXIS, The Netherlands energy would be wasted. This short- on the mutual information theorem as Tel: +31 6543 00 521 term load forecasting is done by a fore- described by Shannon. The mutual E-mail: [email protected] ERCIM NEWS 92 January 2013 13

14 Special Theme: Smart Energy Systems Smart Energy Consumption feedback Connecting Smartphones to Smart Meters by Markus Weiss, Friedemann Mattern and Christian Beckel Smart information and communication technologies can contribute to a more thrifty use of energy. By connecting smartphones to digital electricity meters, we can process electricity consumption data and provide household-specific feedback. Bringing users into the loop then enables them to learn about their personal consumption-related behaviour and optimize it in order to conserve energy. Domestic electricity consumption is con- and handles incoming requests from the tinuously increasing and now accounts user interface. The carefully designed for about one third of the total electrical user interface, which forms the third energy produced in Europe. Many indi- component of our system, is realized as viduals would be interested in saving a mobile phone application on Android, energy (and thus also carbon dioxide as iOS, and Windows Phone platforms. well as money), but they lack informa- tion on their consumption. Indeed, feed- Occupants can easily familiarize them- back on household electricity usage is selves with their electricity consump- typically only provided by monthly (or tion and obtain feedback in real-time by even annual) utility bills and therefore using the following smartphone appli- remains opaque to most households. Few cation functions: people know how much electricity they Figure 1: Live visualization of current a live visualization of current total consume, and even fewer have any idea household electricity consumption electricity consumption (see Figure 1) how much electricity they use for a par- a historical view of electricity con- ticular purpose (eg lighting). And even sumption over time (also indicating those who do have a fair understanding the resulting costs) of their consumption patterns rarely a measurement function to interac- receive guidance about the changes that tively measure the consumption of would have the biggest impact on their any switchable electrical appliance in electricity bills. the house a household-specific recommenda- Fortunately, emerging smart informa- tion service on how to save energy. tion and communication technologies Most notably the latter two functions can help to make electricity consump- provide users with the simple feed- tion visible to individuals [1]. For back advice they require in order to instance, next-generation digital elec- ultimately conserve energy [3]. tricity meters (smart meters) enable Figure 2: Measuring the standby power of a detailed electricity consumption infor- specific household appliance The measurement function (see Figure 2) mation to be captured, processed, and allows users to break down their con- communicated at frequent intervals (eg mation in real-time, enabling occupants sumption and learn how much a specific once per second). As smart meters are to better understand their electricity household appliance or device con- replacing traditional electricity meters in consumption. sumes. To perform a measurement, the large parts of Europe, there is now a user activates the process by pressing unique opportunity to realize compre- The sensing and feedback system we the start button on the smartphone hensive consumer feedback systems that developed consists of three loosely application and then turns a device on or consist of much more than mere remote coupled components that are responsible off. Within seconds, the measurement metering applications. for data acquisition, data handling, and algorithm then determines and displays data visualization. The first component the electricity consumption of the At ETH Zurich we seized the opportu- consists of a state-of-the-art smart elec- device together with information on its nity offered by this development. Within tricity meter that monitors the total energy efficiency and possible energy the eMeter project we connected smart household load. We extended the saving tips. meters to smartphones [2]. As they are meters functionality by incorporating a almost always connected to the Internet gateway module implemented on an The recommendation service analyzes and within reach of the user, using embedded computing device equipped the captured electricity consumption smartphones to visualize electricity con- with flash storage and a WiFi communi- data. It provides feedback that is indi- sumption eliminates the need for costly cation module. This matchbox-sized vidually tailored to the household and in-house displays. The resulting system gateway module also holds the second explains how its consumption relates to not only fits unobtrusively into the home component of our system, a web server that of comparable households. It does environment, but also provides fine- with a database. It manages the recorded so by computing the households grained electricity consumption infor- metering data, performs data analysis, standby power as well as the consump- 14 ERCIM NEWS 92 January 2013

15 tion of cooling appliances (ie fridges Using methods developed in machine Link: and freezers) based on their character- learning and data mining, we are further www.vs.inf.ethz.ch/res/show.html?wha istic quasi-cyclic load curves. investigating ways of generating a pie t=energy chart that shows the contribution of Our eMeter prototype was used in sev- each relevant appliance to the overall References: eral households in Zurich for more than consumption bill. To validate our algo- [1] F. Mattern, T. Staake, M. Weiss: a year. Study participants reported that rithms relative to ground truth, we ICT for Green How Computers Can their energy-saving knowledge substan- equipped six of our test households with Help Us to Conserve Energy, in proc. tially increased thanks to the system. additional sensors that measure elec- e-Energy 2010, ACM, pp. 1-10. However, the applications usage rate tricity consumption at individual power [2] M. Weiss, T. Graml, T. Staake, F. dropped over the course of the experi- outlets and reliably determine when Mattern, E. Fleisch: Handy feedback: ment once the participants initial occupants are at home. Connecting smart meters with mobile curiosity had been satisfied. Owing to phones, in proc. MUM 2009, ACM. this phenomenon, which is well known The eMeter system demonstrates that [3] M. Weiss, C. Loock, T. Staake, F. in the scientific community, concepts smart ICT can collect, analyze, and Mattern, E. Fleisch: Evaluating are required that motivate users to be present valuable real-time data on elec- Mobile Phones as Energy involved with the system over an tricity consumption without costly addi- Consumption Feedback Devices, in: extended period of time. tional hardware. It provides guidance P. Snac, M. Ott, A. Seneviratne (for example in the form of household- (Eds.): MobiQuitous 2010, LNICST In a follow-up project, we are currently specific recommendations) that is crucial 73, pp. 63-77, 2012. investigating algorithms to automati- for occupants who want to conserve cally generate enhanced household-spe- energy. Transforming our prototype, or Please contact: cific saving recommendations. Inferring at least parts of it, into a real product is Friedemann Mattern occupancy from smart meter data, for the next logical step. We are cooperating ETH Zurich, Switzerland instance, enables us to provide feedback with smart meter manufacturers and E-mail: [email protected] on how much electricity a household electric utility companies to work consumes when no one is at home. towards this goal. Meter data Management for Smart Monitoring Power Networks by Agustn Yage, Juan Garbajosa and Mercedes Lopez-Perea The electric power distribution and commercialization scenario is evolving worldwide, and electricity companies, faced with the challenge of new information requirements, are demanding IT solutions to deal with the smart monitoring of power networks. Two main challenges arise from data management and smart monitoring of power networks: real-time data acquisition and big data processing over short time periods. We present a solution in the form of a system architecture that conveys real time issues and has the capacity for big data management. Electric power distribution markets meters throughout the day, and analysed and Software Technology (SYST) of worldwide are evolving, and power off-line in time-constrained periods the Technical University of Madrid companies are demanding solutions that (quarterly, hourly, daily and monthly). (UPM) are working in close collabora- help them manage real time problems Consequently, electricity companies are tion on the ITEA2 project IMPONET to more efficiently and safely, whilst facili- demanding IT solutions to deal with the deal with these challenges. IMPONET tating the integration of real time infor- smart monitoring of power networks has developed a flexible platform that mation into all corporate applications. and, in turn, to be able to respond to the allows the continuous monitoring of the These requirements present new data current market energy needs in real network with real time or quasi real management challenges, which must be time. However, there are two main chal- time data processing and the configura- addressed by smart grids [1] with their lenges behind the smart monitoring of tion of the customer consumption pro- capacity for smart metering and real power networks: real-time data acquisi- file to monitor its evolution and to make time monitoring [2]. Smart grid manage- tion and big data processing in a short decisions based on its information. ment requires dealing with huge period time. We present a solution: a amounts of data including data collected system architecture that conveys real The IMPONET platform includes com- from smart meters and other devices time issues and has the capacity for big ponents/modules for the remote control connected to the power network. The data management. and smart metering of distributed enormous volume of data scales up the energy sources and mobile energy con- data management to a Big Data issue The company Indra Software Labs sumption (eg electric vehicles) by using [3]. Currently, data are acquired from (ISL) and the research group SYstems smart devices and following the ERCIM NEWS 92 January 2013 15

16 Special Theme: Smart Energy Systems The information processing involves handling raw data to obtain optimal data by performing two operations: (i) vali- dation of meter data according to an established validation formula, (ii) cal- culation of the best available energy data from different sources for a meas- uring point, magnitude and period of energy data. This information is then provided to the customer, the retailer and the distributor from different per- spectives, enabling different stake- holders to analyse the results from their own particular point of view (see Figure 2). Link: IMPONET: Intelligent Monitoring of POwer NETworks: http://www.innovationenergy.org/imponet/ Figure 1: IMPONET Data Management Platform References: [1] S. Massoud and B.F. Wollenberg: Service-Oriented Architecture (SOA) nents/modules of the platform to guar- Toward a Smart Grid, IEEE P&E approach. To provide this functionality, antee the real-time capability. The big Magazine 3(5), 2005 these components/modules have the data processing has to be performed in [2] J. Byun et al: A smart energy capacity to respond in real time to the such a way that the massive amounts of distribution and management system massive amounts of information that information extracted from smart for renewable energy distribution and will be received from the network and meters can be processed on time (see context-aware services based on user to store them, as well as the signals that Figure 1). Big data processing has been patterns and load forecasting, IEEE will have to be transmitted to the addressed by using Hadoop and Oracle Transactions on Consumer Electronics, devices in the field. Berkeley DB Non-SQL database. vol. 57 , n.2, 2011 Hadoop Map-Reduce provides parallel [3] S. Madden: From Databases to The data metering is achieved with a and distributed data processing with Big Data, IEEE Internet Computing, powerful communication bus based on Non-SQL approach. This implementa- vol.16 n.3, 2012. the standard data distribution service tion has been validated through dif- (DDS), a real-time publish-subscribe ferent smart grid scenarios obtaining Please contact: paradigm, and the storage of informa- the promising results of storing and Catherine Murphy Oconnor tion into a real-time database (see processing around 150 M measures in Indra Software Labs, Spain Figure 1). The DDS bus serves as a less than three hours. Tel: +34 914806774 coordinator for the rest of the compo- E-mail: [email protected] Figure 2: IMPONET Web Portal: distributor, retailer and customer views 16 ERCIM NEWS 92 January 2013

17 designing and Simulating Smart Grids by Jennifer Prez, Jessica Daz and Eloy Gonzlez Growing energy demands and the increased use of renewable energies have changed the landscape of power networks leading to new challenges. Smart Grids have emerged to cope with these challenges by facilitating the integration of traditional and renewable energy resources in distributed, open, and self-managed ways. Innovative models are needed to design energy infrastructures that can enable self-management of the power grid. Software architectures smoothly integrate the software that provides self-management to Smart Grids and their hardware infrastructures. We present a framework to design the software architectures of autonomous Smart Grids in an intuitive domain-oriented way and to simulate their execution by automatically generating the code from the designed autonomous smart grid architectures. Power networks are undergoing contin- Smart Grids [1] promote the integration the scalability of smart grids and enable uous change due largely to increased of traditional and renewable energy the management of autonomous opera- power consumption and the availability resources, with new elements in a dis- tions throughout the power network. As of renewable energy. These changes tributed, open, and self-managed way. a result, it is vital that we address both require the evolution of power networks This requires innovative models for the definition of self-management poli- towards new software architectures of energy infrastructure and the self-man- cies and the design of the software to be control and operation. These software agement of the power grid [2], whose deployed throughout the different architectures will soon enable great openness and distributed nature devices of the Smart Grid. The design improvements to the automation and requires a shift from the current central- of this software layer involves con- distributed communications of power ized infrastructures towards more dis- structing the Smart Grid software archi- networks. tributed ones. Such a shift would ensure tecture and establishing the autonomous behaviour of the grid. This autonomous behaviour has to define the properties that are going to be monitored and their acceptable values, as well as the poli- cies that have to be followed and auto- matically executed at run-time. The company Indra Software Labs (ISL) and the research group SYstems and Software Technology (SYST) of the Technical University of Madrid (UPM) are working in close collabora- tion on the CENIT Spanish project ENERGOS and the ITEA project NEMO & CODED to address these Smart Grid challenges. From the results of ENERGOS and NEMO & CODED, ISL and UPM are constructing a frame- work to support the design of Smart Grids by providing mechanisms for: (i) designing distributed software architec- Figure 1: Autonomous Smart Grid Architectural Model tures in an intuitive way by using con- cepts of the power network domain [3], (ii) defining the autonomy capabilities of software components, (iii) generating code from these designs, and (iv) simu- lating the behaviour of the Smart Grid software architecture and its autonomous capabilities by executing the generated code. This framework is based on an innova- tive approach which integrates a model Figure 2: Process for automatically for specifying autonomous behaviour, simulating Autonomous Smart Grid and an architectural model designed for Architectural Models specifying Smart Grids. The autonomous model establishes a loop ERCIM NEWS 92 January 2013 17

18 Special Theme: Smart Energy Systems for monitoring those properties that properties, acceptable values and poli- Links: have to be controlled, analysing their cies are to be considered for its http://kwz.me/4L values, processing the collected values autonomous behaviour. This model is http://innovationenergy.org/energos/ through the use of a set of established provided by the framework that has policies, and executing the needed com- been constructed, which offers an intu- References: mands to update the properties (see M2, itive and graphical power network lan- [1] S. Massoud and B.F. Wollenberg: Figure 1). On the other hand, the smart guage (see M3, Figure 1). Toward a Smart Grid, IEEE P&E grid architectural model extends soft- Magazine 3(5), 2005 ware architectural models with the con- This framework follows a well defined [2] J.O. Kephart and D.M. Chess: The cepts of the power network domain to process that consists of the configura- Vision of Autonomic Computing, facilitate the design and comprehension tion of autonomous smart grid architec- Computer, 36(1), IEEE, 2003 of the architectural design (see M1, tures, the execution of model-to-code [3] S. Kelly and J.-P.Tolvanen: Figure 1). The resulting model of this generation patterns and the execution of Domain-Specific Modeling: Enabling integration allows autonomous smart the generated code [3]. Specifically, the Full Code Generation, John Wiley & grid architectures to be defined. As a generated code is Jade, and we are now Sons, 2008. result, any specialist of the energy area working on refining this generation and would be able to define the configura- the platform to systematize the simula- Please contact: tion of power network devices by inher- tion by executing the generated code Jennifer Prez ently defining the software architecture (see Figure 2). Technical University of Madrid and specifying the network devices in (UPM), Spain which the software components are to Tel: +34 913367882 be deployed, and by defining which E-mail: [email protected] Smart Management of Renewable Energy for Green Transport by Raffaele Bruno, Luca Valcarenghi, Molka Gharbaoui and Barbara Martini Concerns about the impact of climate change mandate the drastic reduction of green house gas (GHG) emissions through the increased utilization of renewable energy sources (eg wind or solar energy), as well as hybrid and electric vehicles (EVs). To achieve this goal we outline a smart management system for community-wide public charging infrastructures, which can foster the market penetration of EVs by effectively complementing home-based charging while coping with the power fluctuations of renewable energy sources. One of the main targets for smart cities ventional household electrical outlets community/city-wide level. is a new environmentally conscious (albeit with significant charge delays), Specifically, we envision urban areas in form of urban living, which entails a we are not yet close to developing a which public charging infrastructures more intelligent use of locally generated large-scale public charging infrastruc- are built upon an interconnected mesh renewable energy and the integration of ture, and such an infrastructure will be of residential/commercial charging sustainable and greener sources into the costly to build. Moreover, while the EV docks, or charging stations (an illustra- large-scale car fuelling system [1]. technology enables the shift from fossil tive example is provided in Figure 1). Liberalization and deregulation of elec- fuel dependent combustion engines to The SMS-EV will need to: (i) ensure a tricity markets mean that citizens are no electrical engines, much of the elec- balanced usage of the electrical infra- longer passive consumers of electricity tricity used to recharge the batteries will structure minimizing the impact of EVs provided by large-scale distribution sys- still be produced via unsustainable on grid reliability and peak demands; tems, but active market players and methods. Finally, with increased use of (ii) improve the sustainability of the EV micro-producers at individual and com- EVs, the demand for electricity will charging infrastructure by optimizing munity levels. Moreover, electric cars soar, and, if charging is unregulated, the utilization of locally generated have gained popularity over the last two will be accompanied by the risk of grid renewable energy; and (iii) maximize years thanks to increasing concern about overload. user satisfaction with the EV charging the environment, public health, the service (eg, minimizing EV charging unsustainable nature of fossil fuel use, We argue that to address the above tech- times). This vision entails the use of and the release of a handful of affordable nical challenges there is a need for emerging wireless networking para- electric vehicles (EVs). smart management systems for electric digms (eg, VANETs) for information vehicle recharge (SMS-EV), which dissemination to achieve: good cov- Several infrastructure problems stand in could jointly and autonomously control erage without dense deployments, scal- the way of widespread use of EVs. EV charging and distributed energy ability, communication reliability, and While EVs could be charged using con- resources (ie, renewable sources) at a fast and incremental deployability with 18 ERCIM NEWS 92 January 2013

19 pheric conditions even during high insolation daytime hours (in Figure 2 variability is assumed to be 50% of clear sky nominal power). Figure 2 reports the observed charging delays versus the arrival rate of charging requests for the case of a charging infra- structure deployed in 5x5 grid road net- work, in which each block size is 1km x 1km. The results show that the MWT policy is benecial because it reduces the average waiting time with respect to the CS policy for both the types of power sources. It must also be noticed that for high request arrival rates the MWT policy combined with the utilization of solar power outper- forms the CS policy even if utilizes tra- ditional power sources. Furthermore, these results clearly indicate that there Figure 1: Example of a SME-EV architecture is a need for more sophisticated charging strategies that take into account vehicular mobility as well as a low initial investment. Most impor- EV to the closest charging station; and the temporal variability of PV output tantly, the SMS-EV must feature an (ii) minimum waiting time policy power. intelligent service layer capable of pro- (MWT), which assigns an EV to the sta- viding customized information and tion that minimizes the charging delay Link: services to drivers while ensuring an taking into account the time to complete http://cnd.iit.cnr.it/index.html efficient use of energy resources at indi- the charging of other EVs currently vidual and community levels. present at each charging station. References: Intuitively, another essential factor that [1] C. Rissen et al: Electric Vehicle We are currently exploring the potential may affect the system performance is Charging Solution for Smart Cities, benefits offered by the SMS-EV [2], the type of power resources deployed in Hitachi Review, vol. 61, no. 1, Feb. and preliminary results are reported in SMS-EV. Specifically, we consider two 2012 Figure 2. The considered scenario fea- types of power sources: (i) standard [2] M. Gharbaoui et al: Effective tures an SMS-EV that applies two dif- grid-based power sources, which can be Management of a Public Charging ferent policies for assigning an EV to assumed to have a constant power pro- Infrastructure through a Smart charging stations when a new charging duction profile during the day; and (ii) Management System for Electric service request is used by an EV: (i) solar panels, which are known to suffer Vehicles, in proc. IEEE closest station policy (CS), which mini- from short-term variability of their ENERGYCON-STS12, Florence, mizes travelling times by assigning an output power due to changing atmos- Italy, 2012. Please contact: Barbara Martini CNIT, Pisa, Italy E-mail: [email protected] Raffaele Bruno, Marco Conti IIT-CNR, Italy E-mail: [email protected], [email protected] Luca Valcarenghi, Molka Gharbaoui, Piero Castoldi Scuola Superiore SantAnna, Italy E-mail:[email protected], [email protected], [email protected] Figure 2: Comparison of charging times with different policies for charging station selection and different mixes of power sources ERCIM NEWS 92 January 2013 19

20 Special Theme: Smart Energy Systems Real-Time Visualization of MV/LV Energy Alarms on GIS Web Applications by Christophe Joubert, Vicente Monrabal, Miguel Montesinos and Carlos Snchez In April 2006, in its Directive 2006/32/EC on energy end-use efficiency and energy services, the European Parliament and Council encouraged the installation of smart meters at every end-user in Europe with a complete coverage planned for 2018. This power network evolution implies new needs such as real-time management of energy flows to increase connectivity, automation and coordination of producers, providers and consumers in the distribution network. Our work contributes to the attainment of this goal by providing real-time visualization of energy alarms detected at the end-user end, ie medium voltage (MV) / low voltage (LV) energy alarms. In the ITEA2 international project requirements in energy management, in to visualize energy alarms, faults, and IMPONET (AVANZA TSI-020400- particular power quality monitoring and other energy information on a map [1]. 2010-103, ITEA2 N 09030), a compre- the remote control and smart metering This project offers novel power quality hensive, flexible and configurable infor- platform. To date, power companies monitoring solutions for power compa- mation system has been developed to have been using geographical informa- nies to visualize real-time information support the most complex and advanced tion systems (GIS) with historical data coming from the medium voltage (MV) / low voltage (LV) distribution network. In order to operate the network more efficiently, it is necessary to have a sys- tematic and standardized method of exploiting data from a huge number of electronic devices involved in control and protection of power systems. One solution investigated in IMPONET is the use of the object management group (OMG) data distribution service (DDS) open standard for messaging that sup- ports real-time systems [2]. Also, to achieve the ambitious goals that are envisioned for Smart Grids of the future, a new conception of network monitoring systems is required. Within the power quality monitoring force task, we developed a geoportal that allows the continuous monitoring of the net- work with real-time data processing capabilities. It includes features and Figure 1: Geoportal for real-time and B.I. visualization of energy alarms services that enable providers to detect issues on the power network, to monitor their evolution and to make decisions Figure 2: Real-time connection based on this information. architecture between DDS and geoportal The power quality monitoring geoportal is divided in four parts as described in Figure 1: layer tree, substation selector, map viewer, and business intelligence (B.I.) viewer. The layer tree contains a set of common and open base layers as well as several specific overlays: real- time faults, power network compo- nents, and B.I. layer for statistics. The map viewer allows a user to interact with any element of selected layers, and to show historical data or real-time events directly from the DDS. Finally, the map viewer interacts with the B.I. viewer, which allows a user to request 20 ERCIM NEWS 92 January 2013

21 any kind of statistics over the power Slovenia, Spain and Turkey) and fifteen quality OLAP cubes preprocessed from project partners: Answare, Deusto monitoring campaigns. University, Tecnalia, INDRA, Innova, Kapion, Kema, LNL, UPM, Union The novelty of our approach resides in Fenosa, Girona University, Ljubljana the processing of real-time power fault University, LDOS, Wooam. Our work is events from DDS to GIS web clients in partially supported by the Spanish MEC three stages, as illustrated in Figure 2: INNCORPORA-PTQ 2011 program. 1) When the fault overlay is selected, a DDS subscriber is internally Link: launched and connected to the GIS http://www.innovationenergy.org/imponet client in order to capture fault events, ie voltage sags, swells and other References: power disturbances detected at the Figure 3: Real-time layer with geolocated [1] G. Aydin: Service oriented smart meters and substations. fault events coming from DDS architecture for geographic information 2) Through a permanent link between systems supporting real time data the GIS application and the DDS sub- grids, Ph.D. Thesis: Indiana scriber, based on an atmosphere serv- scale in order to show the alarm University, 2007 er for asynchronous web applications severity based on duration and mag- [2] J. A. Dianes et al: Using standards and Socket.IO for communication nitude of the fault. to integrate soft real-time components channels, fault events are transmitted into dynamic distributed architec- in real-time to the GIS client. This The results are very encouraging: power tures, Comp. Stand. & Interfaces, vol. technology allows any web client to companies have already shown an 34, no. 2, pp. 238-262, 2012 interact with the service without interest in such architecture, which pro- [3] E. Bozdag et al: Performance needing to refresh the web client [3]. vides features not yet covered by cur- testing of data delivery techniques for 3) Fault events being geolocated, they rent power company solutions, in par- AJAX applications, J. of Web Eng., are represented on the power network ticular real-time geolocated information vol. 8, no. 4, pp. 287-315, 2009. where faults occur (as indicated on for the lower part of the power network, Figure 3) as soon as they become directly from the outputs of end-user Please contact: available on the DDS. In the case of smart meters. Christophe Joubert MV / LV faults, they appear close to Prodevelop, Spain the substations. Faults are depicted in Collaboration for this project spanned Tel: +34 963 510 612 different colours on a green to red four countries (Korea, Republic of E-mail: [email protected] Using Wireless Smart Meter Networks for Power Quality Monitoring Joel Hglund and Stamatis Karnouskos The future of Smart Grids lies in standardization and open protocols. By using the capabilities of modern smart meters, which communicate over wireless radio, grid power quality monitoring can be improved while keeping costs low. In NOBEL we have been developing mid- include: SICS from Sweden, University enable electricity providers to better opti- dleware and services for the future smart of Duisburg-Essen and SAP from mize the voltage delivered, which may electricity grids. One of the key issues, in Germany, ETRA and Alginet from Spain lead to a reduction of the energy con- the context of lower level networking, is and CERTH from Greece. sumed of up to 4% [1] (for Sweden that how best to make use of the more fine would correspond to 5.6 TWh in 2011). grained smart meter measurement data. Part of the work has been to explore and We have studied how smart meter data can outline the business advantages that can The next generation WSN based tests be used to complement and partially be gained by improved power quality have been performed at a distributed replace more expensive grid monitoring monitoring [1]. Improved monitoring testbed at SICS, Stockholm, Sweden, equipment without comprising on power gives electricity providers proof of with the receiving enterprise services quality monitoring [1]. This is achieved power quality that can be delivered to running at UDE in Duisburg. Larger using smart meters that communicate over residential users and be used to attract scale user trials using several energy low power radio, with the smart meters new customers. It would allow residen- services have been performed while forming a wireless sensor network (WSN). tial customers to monitor their own tapping into the existing smart meter power quality. It offers greater potential infrastructure in Alginet, a small town The European Commission FP7 NOBEL for early problem identification and pre- outside of Valencia, Spain. In the lab (www.ict-nobel.eu) projects partners ventive maintenance. Finally, it could tests, the smart meters are modelled ERCIM NEWS 92 January 2013 21

22 Special Theme: Smart Energy Systems with relatively cheap, low power, hard- dled by a 6LoWPAN layer which access to or tampering of energy data ware. As an important partial goal is to ensures the maximum packet propor- may raise security and privacy con- reduce the overall consumption, new tions can be used for actual program cerns. Since the design is standard infrastructure should not offset the sav- data. Still, it is worth spending some based, it is well prepared to be extended ings by its own energy consumption. design effort to keep messages meter with IPSec [3], something which we readings short so as to limit the delays currently investigating. By being based on the Contiki operating of fragmentation which would increase system [2], the smart meter middleware network delays. References: is using implementations of the wireless [1] J. Hglund, D. Ilic, S. Karnouskos, standards which have gone through In our evaluations we have shown that R. Sauter and P. Goncalves Da Silva: interoperability testing across different the testbed could handle meters Using a 6LoWPAN Smart Meter platforms. Contiki is being supported reporting data every 60 seconds. On top Mesh Network for Event-Driven Mon- by a growing number of hardware plat- of that we made extra subscriptions itoring of Power Quality, in 3rd IEEE forms, and has an active developer com- when the system detected a voltage International Conference on Smart munity. Using the latest routing stan- deviation, where a subset of the meters Grid Communications (SmartGrid- dard for wireless networks, RPL, the started reporting every 10 seconds to Comm), Tainan City, Taiwan, 2012 smart meter network becomes self-con- aid the tracing of quality problem ori- [2] A. Dunkels, B. Grnvall, and T. figuring. A newly installed device will gins. These results were communicated Voigt: Contiki - a lightweight and automatically connect with the rest of to an enterprise system via the NOBEL flexible operating system for tiny net- the network, and acquire the needed energy services and show that the sug- worked sensors, 1st IEEE Workshop network configuration data. This lowers gested architecture is feasible with on Networked Embedded Sensors installation costs, and makes it easy to acceptable overhead [1]. The power net- (EmNetS-I), Tampa, Florida, USA, add new meters. work will continue to need high cost November 2004 network analysers to get instantaneous [3] S. Raza, S. Duquennoy, T. Chung, Based on local voltage deviation detec- state information at critical points in the D. Yazar, T. Voigt, and U. Roedig: tions, smart meters can trigger events to grid. Nevertheless, there are clear Securing Communication in 6LoW- alert the local grid operator. The events advantages to complementing the archi- PAN with Compressed IPsec, IEEE could either trigger a request to auto- tecture with modern smart meters International Conference on Distrib- matically increase the sampling rate in capable of delivering fine grained moni- uted Computing in Sensor Systems the affected area, or be forwarded to a toring data. (DCOSS '11), 27-29 June 2011, grid operator for manual decision Barcelona, Spain. making. The NOBEL project started in spring 2010, and finished in December 2012. As the radio packet sizes in WSNs are Security in the wirless networks has not Please contact: smaller than on the wired internet, frag- been addressed within the project, but Joel Hglund, SICS mentation and address compression for real life deployments the customer Tel: +46 70 775 16 09 become important issues. This is han- data needs to be further protected, as E-mail: [email protected] Smarter Energy: Opportunities and Challenges by Olle Sundstrm, Fabian Mller, Carl Binding, Bernhard Jansen and Dieter Gantenbein To increase the share of renewable energy in the electric power system, smarter solutions are needed in many areas. The way we consume electricity needs to change so that consumption is to a greater degree influenced by the availability of energy. In a system with high penetration of fluctuating production, such as wind and solar power, all kinds of flexible loads need to be identified and utilized to balance the supply and demand. At IBM Research Zurich, we are A first area of investigation has been Our general approach is structured as involved in various research projects the use of a large fleet of electrical follows: that focus on enabling an increased vehicles (EVs) in the context of the Identify a mathematical model penetration of renewable energy in the EDISON project, where energy describing the underlying physical power system. Energy can be buffered buffering was to be provided by the EV model. For that purpose, we use sys- directly by, for example, the thermal batteries [1]. Currently, we focus on tem identification techniques based mass of a house or the energy storage thermal storage, be it for industrial on observed historical data. in a battery. Energy can also be indi- cooling and refrigeration purposes or Using observed data, we derive fore- rectly buffered by time-shifting the for domestic heating applications in the casts of future use of these systems. consumption, for example, by starting context of the EcoGrid EU project [3]. For example, we formulate trip pre- an appliance at a different point in time dictions for EVs or predictions on than intended. hot-water usage in the case of boilers. 22 ERCIM NEWS 92 January 2013

23 the main energy buffering is represented by home-heating systems driven by heat pumps. The system's flexibility is bound by the comfort zone set by the user, who may tolerate a temperature band around a set point. We have achieved initial results in system identification based on the data collected in the pilot setup. Figure 3 illustrates the quality of the system modeling using the indoor and the out- door temperature as well as the influ- ence of solar radiance. References: Figure 1: System architecture [1] O. Sundstrm and C. Binding: Flexible charging optimization for electric vehicles considering Based on the predictions and the sys- imum and maximum usage boundaries distribution grid constraints, IEEE tem model, we compute the flexibili- and, based on these, derive a minimum Transactions on Smart Grid, Volume 3, ty in terms of the electricity needed and a maximum inflow of energy Issue 1, pp. 26-37, 2012 from the power system. needed to satisfy demand. Figure 2 [2] O. Sundstrm, C. Binding, D. This flexibility allows us to allocate illustrates this flexibility. The energy Gantenbein, D. Berner, and W. power over time in such a way that and power taken from the power system Rumsch: Aggregating the flexibility the demand requirements are fulfilled over time are constrained to remain of domestic hot-water boilers to offer while, for example, following sto- within the energy and power flexibili- tertiary regulation power in chastic renewable generation. ties. All trajectories within the flexi- Switzerland, in proc. IEEE Intelligent The optimal power schedule is com- bility bounds would satisfy user Smart Grid Technology Conference, puted and communicated to the flexi- demand, but differ in the timing of Berlin, 2012 ble energy resource as a control signal. replenishment. [3] D. Gantenbein, C. Binding, B. Jansen, A. Mishra, and O. Sundstrm: A schematic architecture of our A fourth layer aggregates the flexibility, EcoGrid EU: An efficient ICT approach is shown in Figure 1. We dis- and the fifth optimizes the use of the approach for a sustainable power tinguish several layers. First, there is the flexibility. Here, optimization tech- system, in proc. Second IFIP data-collection layer, in which we use a niques are used to achieve the best Conference on Sustainable Internet and large-scale communication and sensing usage of energy generation versus ICT for Sustainability, Pisa, Italy, 2012. infrastructure to harvest usage and gen- demand in general. eration information. A second functional Please contact: layer performs the system identification. Our current focus is on applying the Olle Sundstrm above energy-buffering paradigm and IBM Research, Switzerland A third layer computes the flexibility of the system architecture in the frame- Tel: +41 44 724 8871 the energy systems. We indicate min- work of the EU project EcoGrid, where E-mail: [email protected] Figure 2: Schematic flexibility, with minimum and maximum energy needs and minimum and maximum power flexibility, together with a realization of the flexibility (blue curves) Figure 3: Modelling a home-heating system ERCIM NEWS 92 January 2013 23

24 Special Theme: Smart Energy Systems Smart demand-Side Response at Home by Armin Wolf, Thomas Luckenbach and Mario Schuster The threat of global climate change and the challenge of reducing carbon emissions necessitate new ways to collaboratively manage and optimize energy production and load. Home and living areas can be part of such an approach thanks to secured and standardized smart metering environments. The availability of renewable energy profiles of the considered electricity tariff information. For instance, a typ- fluctuates heavily. Because of the sto- consumers on a common cost aware ical dish washer has a low power chastic nature of energy production, cumulative resource [2], respecting demand while pre-/post-rinsing (eg 200 energy consumption should follow its temporal constraints (eg washing W) but has a high power demand production. The introduction of time- before laundry drying) and capacity during the washing/drying phase (eg 1 variable energy tariffs is an approach to restrictions (eg determined by the main kW) such that a naive management attract consumers to shift their energy fuse). Therefore, the system receives approach, which starts dish-washing loads into low demand periods. We have the energy tariff for the considered when the energy price is low, may not developed a smart energy management scheduling horizon typically 24 hours be the most economical option since system that reacts to flexible electricity with fixed prices for a time resolution the energy intensive part of the cycle prices. The system optimally schedules of 15 minutes as well as the load pro- might occur within the high-price electricity consumers, minimizing the files to be scheduled, the overall period. The temporal constraints allow overall energy cost while respecting capacity level which must not be the formulation of: device-specific load profiles and indi- exceeded, and the temporal constraints time slots, ie earliest start and latest vidual constraints defined by the user. on the load profile. The load profiles completion times of the energy loads, According to a recent study on demand are represented by discrete pricewise mutual exclusions of several loads, side integration, there is the potential to constant functions showing the average eg caused by the same device or load shift more than 10 GW in private power demand of an electricity con- requiring an additional but exclusive- households in Germany [1]. sumer per time unit during its usage ly available resource, eg an operator, which has to be scheduled. The chosen temporal relationships between the The core of the realized energy opti- time resolution depends on the applica- start and end times of different loads, mization system comprises a constraint- tion domain typical resolutions are eg before, while, after, earli- based scheduler that schedules the load five or 15 minutes according to the est, or latest conditions. Figure 1: Example of a dynamic load management based on individual preferences and load profiles of typical devices 24 ERCIM NEWS 92 January 2013

25 The system transforms this kind of input This gateway is a programmable com- [2] H. Simonis and T. Hadzic: A data and constraint definitions (defined munication device that can take over the resource cost aware cumulative, in in XML) into a mathematical, constraint role of a gateway for Controllable Local proc. ModRef 2010 workshop at CP optimization problem using a Constraint Systems (CLS) to allow electrical 2010 Programming (CP) approach. Our con- devices to be switched based on the cur- [3] A. Wolf: firstCS New Aspects straint-solving library firstCS [3] rent load management. The CLS on Combining Constraint offers the necessary primitives and approach is also specified in the Programming with Object-Orientation Operations Research (OR) functionality German guidelines for smart metering in Java. KI Knstliche Intelligenz, to define and solve the necessary equali- systems and gateways and the respec- vol. 26, no. 1, 2012 ties and inequalities resulting from vari- tive protection profile [4]. Additional [4] Federal Office for Information able tariffs as well as the temporal and software modules of our gateway Security, Protection Profile for the capacitive restrictions. The achieved enable demand side integration (eg our Gateway of a Smart Metering System, solution consists of two parts: a cost- smart energy management) and load Gateway PP, v01.01.01(final draft), minimal schedule of the loads and the control. For comfortable use, the 25.08.2011. according load distribution. Again, the gateway offers a unified web-based output is in XML format. In particular, user-interface on PC, appliances, and Please contact: for the load distribution the interface smartphones (using HTML5). The solu- Armin Wolf standard for meter reading and control, tion is an open interoperable gateway Fraunhofer FOKUS, Germany IEC 61968-9, is used. We have chosen having a standardized interface to Tel: +49 30 6392 1864 this format for easy comparison of the access certified smart meter gateways E-mail: [email protected] scheduled load with the real energy con- (SMGW), smart meters, and other sumption measured by smart meters. machine-to-machine (M2M) devices. It Thomas Luckenbach These data in the same format show the also supports state-of-the-art home Fraunhofer FOKUS, Germany difference between target and reality as automation and facility control tech- Tel: +49 30 3463 7245 well as the financial loss compared to nologies such as digitalSTROM, KNX, E-mail: the cost-optimal schedule. In order to ZigBee and 868MHz RF. [email protected] perform the necessary analysis and per- form a comfortable automatic control of References: Mario Schuster the involved devices (causing the loads), [1] Demand Side Integration Fraunhofer FOKUS, Germany the energy management system can run Lastverschiebungspotenziale in Tel: +49 30 3463 7302 on our own energy management and Deutschland, VDE-ETG-Studie, June E-mail: control gateway. 2012 [email protected] Market Garden: A Scalable Research Environment for Heterogeneous Electricity Markets by Felix Claessen, Nicolas Hning, Bart Liefers, Han La Poutr and Peter Bosman How will we trade energy in the future? We can expect vast technological changes in our energy systems, leading to high heterogeneity in both supply and demand. Knowing the true value of energy at a given time and location will be crucial. Market mechanisms are methods of determining prices in complex, multi-actor settings. In the Intelligent Systems group at CWI, we have developed a research environment in which different market mechanisms for electricity can be studied and evaluated - in interaction, remotely and in a scalable manner. In order to allocate energy efficiently in It is important to design and study a (see [2], for example). However, a dif- future energy systems, it will be vital to range of possible market mechanisms ferent mechanism is needed if it is cru- take an economic perspective. This is and their properties, since it is unlikely cial for a substantial number of devices due to three major trends: the rise of that a single market mechanism, for or participants to plan ahead, eg because intermittent renewable generation, new example one spot market [1], can be electric vehicles in a residential area steerable demand appliances (eg electric used in all local settings. Consider flat- need to charge their batteries in time vehicles and heat pumps) and market tening the loads of the heating devices before their owners go to work [3]. A deregulation. The Intelligent Systems in a large office building: This can, in market mechanism for this situation group at CWI has developed a software principle, be accomplished by continu- should allow bidders to announce pref- research environment to study market ously collecting price bids from the erences for future time slots (eg when mechanisms for various economic devices for the next 10 minute interval they need to leave) and incorporate that future scenarios. and updating allocations accordingly into the computation of an efficient out- ERCIM NEWS 92 January 2013 25

26 Special Theme: Smart Energy Systems !

27 Figure 1: Simulation of a future scenario in Market Garden. A scenario consists of a network topology, a market hierarchy and a market timeline. come. Other goals of market mecha- remotely accessed on powerful servers. strategies, novel market mechanisms nisms can be to reward flexibility and to In 2011, a collaboration funded by the and their interaction, thereby combining provide a minimum level of fairness. European Institute of Innovation & various research areas in computer sci- Not all such goals are compatible with Technology (EIT), with participants ence, including auction design theory, each other and even if they are, from science and business, such as optimization and agent technology. The including them can make a market Siemens, Imperial College London, group also studies the use of sensor net- mechanism too complex to be usable. Ericsson and also our group at CWI, has works in smart buildings and electricity developed four Future Scenarios for network planning. This has two major consequences for energy systems, forming a basis for fur- research. First, it becomes important to ther research into valuable business It is planned that Market Garden will be study which market mechanism works models. Partly funded by the EIT, the available in 2013. If you are interested best in a given setting. Second, market development and use of Market Garden in writing and testing a scenario or mechanisms will differ in important aims to act as a catalyst for innovation joining a simulation, please contact us. aspects such as timing and expressive in this scenario research by benefiting power of bids, so how can we expect the partners. Market Garden is therefore Link: results of various local markets (on the planned to be made available as an http://virtualsmartgrid.project.cwi.nl/wi low voltage level) to be aggregated and online simulation tool as part of the ki/Market_Garden be used in national markets (on the European Virtual Lab. This initiative middle and high voltage levels)? enables distributed co-simulation by References: connecting existing hardware labs and [1] Schweppe, Fred C., et al: Spot To help answer that question, we have simulators which model wind farms, pricing of electricity. (1988) developed Market Garden, a scalable coal plants, smart offices or houses, etc. [2] J. K. Kok, C. J. Warmer, and I. G. research environment for market mecha- Kamphuis: PowerMatcher: multiagent nisms, written in Java. In Market The second advantage is accelerated control in the electricity Garden, users can model a physical grid development through cooperation. infrastructure, in proc. 4th and then associate the agents which are Market mechanisms are set up in a mod- international joint conference on controlling devices on the grid in mar- ular fashion, allowing users to create autonomous agents and multiagent kets. Market Garden can host different different scenarios by coupling and systems, ACM, 2005 market mechanisms, which define the interchanging mechanisms. [3] V. Robu et al: An online formats of bids, the protocol of bidding Furthermore, scenarios, mechanisms mechanism for multi-speed electric and how an outcome is computed. For and strategies can easily be reused. vehicle charging, in proc. Second instance, one general market mechanism International Conference on Auctions, that is currently included in Market In addition to accelerating research, Market Mechanisms and their Garden is a clearing market with various Market Garden will be useful to end Applications (AMMA'11). possible bid formats (linear, piecewise users, who can gain experience with linear, quadratic). Given a grid model, their possible future roles and policy Please contact: one or more market mechanisms and makers, who can be assisted in Felix Claessen, Nicolas Hning, Bart bidding strategies for agents, Market designing laws and regulation that Liefers, Han La Poutr, Peter Bosman Garden runs agent-based simulations to hinder strategic market exploits and CWI, The Netherlands compute the results. ensure societal benefits. Tel: +31 20 592 4330 E-Mail: {felix.claessen, The first advantage of Market Garden is The Intelligent Systems group at CWI nicolas.honing, bart.liefers, that simulations can be hosted and uses Market Garden to research trade han.la.poutre, peter.bosman}@cwi.nl 26 ERCIM NEWS 92 January 2013

28 The Power Trading Agent Competition by Wolfgang Ketter and John Collins Energy is at the foundation of modern society, and almost everything about the way we produce, use, and manage energy needs to change over the next few decades. Our work is focused on understanding this transition, particularly in the electric power sector. We have developed Power TAC, a simu- regional distribution network, and is ronment, and so accurate predictions lation environment that models a liber- responsible for maintenance of its infra- may need to account for the effects of a alized retail electrical energy market structure and for real-time balancing of brokers own actions in order to model that includes a significant fraction of supply and demand. The balancing the effects of other brokers actions. retail-level production (eg solar, wind) process is a market-based mechanism Successful broker agent designs will and storage (eg electric vehicles) capaci- that uses economic incentives to typically integrate a variety of tech- ties [1], and allows us to experiment encourage brokers to achieve balance niques from artificial intelligence, with market structures in a variety of within their portfolios of tariff sub- machine learning, and game theory [2]. potential scenarios for the future energy scribers and wholesale market posi- landscape. Figure 1 shows a schematic tions, in the face of stochastic customer There are many important open ques- view of the simulation environment, behaviours and weather-dependent tions and research challenges posed by which models a wholesale market, a reg- renewable energy sources. The broker a power grid with large numbers of ulated distribution utility, and a popula- with the highest profit at the end of the active participants. Examples include tion of retail energy customers and pro- simulation wins. the role of retail brokers and their cus- ducers, situated in a real location on tomers in grid balancing; demand-side Earth during a specific period for the The Power TAC wholesale market is a management using dynamic pricing; weather data that is available. relatively simple call market, similar to forecasting supply, demand, and prices many existing wholesale electric power in the future grid; tariff terms and cus- In this market, competing business enti- markets, such as Nord Pool in tomer preferences; and impact of elec- ties or brokers offer energy services to Scandinavia or FERC markets in North tric vehicle penetration on the grid. A customers through tariff contracts, and America, but unlike the FERC markets number of these questions concern the must then serve those customers by we model a single region, and therefore structure of markets and the behaviours trading in the wholesale market. Brokers we do not model locational-marginal of market participants. Some of these are challenged to maximize their profits pricing. can be addressed by game-theoretic by buying and selling energy in the analysis, but many are sufficiently com- wholesale and retail markets, subject to Broker developers face a number of plex that they cannot be effectively fixed costs and constraints. Costs include interesting challenges. Broker agents addressed by formal methods. To fees for publication and withdrawal of operate in a fast-paced information-rich address these more complex issues, tariffs, and distribution fees for trans- environment. Customer behaviour is simulation-based techniques are porting energy to their contracted cus- stochastic, and depends partially on required. Unlike more traditional simu- tomers. Costs are also incurred whenever weather and the actions of brokers. lation methods, Power TAC is a com- there is an imbalance between a brokers Brokers are challenged to predict cus- petitive simulation [2] that invites total contracted energy supply and tomer power usage and wholesale research teams to write their own demand within a given time slot. The market prices up to 24 hours in advance, autonomous software agents that com- competition revolves around brokers who and have multiple available actions to pete with each other in rich economic play against each other to maximize their interact with the markets and influence simulation environments through profits. customer behaviour. The competitive annual competitions. Competitions pro- environment is oligopolistic, which duce large quantities of data, as well as The distribution utility models the regu- means that broker actions have observ- the competing agent software, which lated natural monopoly that owns the able impacts on the competitive envi- will be publically available for analysis Power TAC Figure 1: Power TAC scenario pants overview Partici Brokers build portfolios, P buy & sell power Tar ff market Tariff Large Energy Retail Customers Suppliers producers, consumers Environment Simulation Distribution Utility owns/operates local grid (regulated monopoly) ERCIM NEWS 92 January 2013 27

29 Special Theme: Smart Energy Systems by any interested parties. The ongoing Link: http://www.powertac.org Competition; AI Magazine 33 (1), 2012 discussion of Smart Markets [3] recom- [3] M. Bichler, A. Gupta, A., W. mends rich market simulations such as References: Ketter: Designing smart markets; Power TAC to validate market struc- [1] W. Ketter, et al: The 2012 Power Information Systems Research 21 (4), tures and to minimize real-world risks. Trading Agent Competition, Tech. 688699, 2010. Rep. ERS-2012-010-LIS, RSM The 2013 competition is scheduled in Erasmus University, Rotterdam, The Please contact: conjunction with Association for the Netherlands (2012) Wolfgang Ketter Advanced of Artificial Intelligence http://ssrn.com/paper=2144644 Erasmus University, The Netherlands (AAAI) conference for July 2013 in [2] W. Ketter, A. Symeonidis: Tel: +31104082624 Bellevue, WA, USA. Interested individ- Competitive benchmarking: Lessons E-mail: [email protected] uals and groups are invited to participate. learned from the Trading Agent http://www.rsm.nl/energy A Model-free flexibility Management System at KU Leuven and VITO by Stijn Vandael, Bert Claessens, Tom Holvoet and Geert Deconinck In the future smart energy system, millions of domestic appliances will be managed in order to support the electricity grid. Is it realistic to build exact mathematical models and controllers for each individual appliance? In Belgium, at KU Leuven, the University of Leuven, and VITO, the Flemish Institute for Technological Research, we take a different approach. Recent years have seen an explosive number of devices quickly become large number of household devices, growth in renewable energy. The computationally intractable, while con- without the need for complex mathe- European target is to achieve 60% elec- sumers require a fast response. For matical models at each individual tricity generation from renewable example, is it realistic to keep an EV household [1]. Essentially, this manage- sources by 2050. Owing to the low con- owner waiting for 15 minutes before ment system consists of two layers: an trollability of renewable electricity gen- charging, because calculations have to energy controller and a power controller eration, engineers are looking for flexi- be done? (Figure 1). The energy controller trades bility in household power consumption In a joint effort between the Belgian on electricity markets, and steers the patterns. For instance, heat pumps can university KU Leuven and the research collective flexibility of a cluster of store warm water in a buffer at flexible institute VITO, we have developed a household devices accordingly. Based times. Similarly, electric vehicles (EV) flexibility management system for a on this steering behaviour, the power can be charged at flexible controller decides on the times during the night. Figure 1: A model-free power consumption of the flexibility management system clusters individual house- In smart energy systems, hold devices, and strictly millions of flexible house- respects each devices com- hold devices, which can fort settings. intelligently manage their flexibility, will become The energy controller trades available. However, building on electricity markets to an exact mathematical acquire the best deal for the model and controller for flexibility of its devices. In each individual device electricity markets, elec- would be impractical and tricity is traded in energy complex, because each has amounts per timeslot. For different characteristics, and example, at Belpex, the its usage is difficult to pre- Belgian day-ahead market, dict. For example, different electricity is traded in MWh types of EV batteries will (megawatt hour) on an have different charging hourly basis. Depending on characteristics, and EV the composition of the owners may depart at dif- cluster, different short- and ferent times than expected. long-term contracts can be Furthermore, algorithms for negotiated. At KU Leuven optimally controlling a large and VITO, a team of econo- 28 ERCIM NEWS 92 January 2013

30 mists is studying these complex market values for individual devices. In con- ment approach. Examples of devices, interactions. trast to the once-per-timeslot control of which are currently being managed, are the energy controller, the power con- electric scooters and washing machines. Based upon the agreed contract, the troller continuously reacts to events (for Both simulations and real-world experi- energy controller steers the flexible example, a heat pump which is switched ments show that our approach is able to energy demand of its cluster. Before the on) to control devices in its cluster, and find close to optimal solutions, without beginning of each market timeslot, the respects comfort settings of consumers. the need for exact mathematical models. energy controller sends a request to the The power control values are deter- power controller for the consumption of mined by a priority scheme, which References: an amount of energy by the whole assigns higher power values to devices [1] S. Vandael et al: A scalable three- cluster. At the end of each market with a higher priority. A devices pri- step approach for demand side timeslot, the energy controller receives ority is based on the imminence of its management of plug-in hybrid feedback on the actual amount of energy consumption. For example, an vehicles, IEEE Transactions on Smart energy consumed by the cluster. This EV with an SOC (state of charge) of Grid simple feedback mechanism hides any 20% will have a higher priority than an [2] B. Claessens et al: Self-learning information about individual devices, EV with an SOC of 80%. In general, the Demand Side Management for a and allows us to learn the appropriate priority can easily be determined by a Heterogeneous Cluster of Devices with energy requests, which are evaluated by few key parameters of the device. Binary Control Actions, ISGT-2012. a reward function (Figure 1). This reward function depends on the negoti- In a simulation framework developed at Links: ated contracts. The key advantage of KU Leuven and VITO, the two-layer http://distrinet.cs.kuleuven.be/ our approach is its independence of any flexibility management concept has http://esat.kuleuven.be/electa/ underlying model, which means the been extensively tested with heteroge- http://vito.be/ energy controller can learn the behav- neous clusters of devices (eg electric iour of the cluster simply by observing vehicles, heat pumps and boilers). These Please contact: the reaction to its energy requests [2]. simulations show that it is possible to Stijn Vandael The ability to learn these reactions continuously improve control actions KU Leuven, Belgium arises from the repetitive collective over a two weeks period, until an appro- Tel: +32 16 327826 behaviour of consumers. priate set of control actions is found. E-mail: [email protected] Furthermore, at VITO, the first real- The power controller translates energy world experiments are being set up amounts per timeslot to power control using the two-layer flexibility manage- demand Side Management for Multiple devices by Albert Molderink, Vincent Bakker and Gerard J.M. Smit Steering a heterogeneous set of devices in a Smart Grid using cost functions: Demand Side Management (DSM) is an important element in smart grids. DSM is already in operation for large consumers, but thorough research is required into DSM on a building level within the distribution grid. Decreasing flexibility on the production with this approach is that it introduces (see Figure 1). The outcome of such an side of the supply chain, caused by a device-level constraints, resulting in a optimization problem is an energy price growing share of (uncontrollable) very complex optimization problem (Aimp), which in combination with the renewable generation, and an increasing that is subject to many constraints. The cost function of the devices specifies load on the grid, due to consumer side complexity can be overcome by using the resulting dispatch for each device. developments as distributed generation generic functions expressing the device and electrical cars, require flexibility on constraints and ranking, in order of In all control methodologies, a decision the consumer side of the supply chain to preference, different options relating to for every device is taken, often in a hier- maintain a stable and affordable energy the device. archical way to maintain scalability. supply: Demand Side Management Therefore, the cost functions of mul- (DSM). A range of DSM methodologies Cost functions, which reduce the com- tiple devices need to be aggregated. have been proposed [1], based around plexity of the optimization problem to a Adding the cost functions of two strategies to optimize consumers con- cost optimization problem with a lim- devices means adding two partial linear sumption patterns and to exploit the ited number of constraints [2], express non-continuous functions. The sum of potential of buffers and distributed gen- the relationship between energy price the two functions consists of all pos- eration. Most proposed methodologies and level of consumption (Xd). We use sible combinations of the individual optimize at a device level. The problem partial linear non-continuous functions parts of the two individual functions ERCIM NEWS 92 January 2013 29

31 Special Theme: Smart Energy Systems (see Figure 2), where some combina- cannot be switched on while the freezer) and sources (import, generator tions may not be applicable. freezer is switched off. The cost func- and battery) are combined, the amount tion of the freezer depends on the tem- of consumption depends on the price, We use a freezer and a fan to demon- perature of the freezer. Figure 3(b) but the price (per unit) depends on the strate how the cost functions of two shows multiple values of the energy amount of consumption. For example, consuming devices are combined. The price and multiple temperatures for the matching the above mentioned costs energy price determines which option freezer when both the fan and the for demand and supply, both the fan (eg switching on/off) will be chosen freezer are switched on, when only the and the battery would be supplied since for each device; costs can be calculated freezer is switched on, when only the the battery can deliver enough cheap based on the energy price, where the fan is switched on and when both electricity for both. However, when the cheapest option will be chosen. Figure devices are switched off. capacity of the buffer would have been 3(a) shows the total costs for the pos- limited on 100W, only the freezer sible scenarios as a function of energy As a second example, Figure 4(a) would have been supplied. price: when the energy price is lower shows the cost functions of three elec- than 1000, both devices will be tricity supply sources (production): In conclusion, the cost functions for switched on (the line with the lowest electricity import, a generator and a devices in combination with the options cost will be chosen); when the energy battery. The battery is the cheapest per device are a very flexible way to price lies between 1000 and 1100 only source (when it is full), but has a lim- express the status of the device and the freezer will be switched on; and ited capacity. The combined function desirability of different options. The when the price is higher than 1100 both (Figure 4(b)), shows which (mix of) control methodologies act on a homoge- devices will be switched off. So, in the source(s) is used for a given level of neous set of cost functions that keeps combined cost function with one electricity consumption. When the the algorithms much easier and less steering signal for both devices, the fan above mentioned consumers (fan and computationally intensive. Cost func- tions of multiple devices can be com- bined into one cost function to aggre- gate them for scalability, even when Figure 1: Example consumption and production are com- cost function bined. A single energy price for a group of devices (combined cost functions) is a trade-off between complexity and flexibility: it disables certain combina- tion of options and may consequently limit the application of a single price for steering. For more information, an example of steering a group of 400 houses with multiple devices and dif- ferent control methodologies can be found in [3]. Figure 2: Combining two non-continuous, partial linear functions References: [1] A. Molderink, et al: Comparing Demand Side Management approaches, international conference on Innovative Smart Grid Technologies Europe, October 2012 [2] A. Molderink, et al: Management and control of domestic smart grid technology, IEEE transactions on Smart Grid, vol. 1, no. 2, pp. 109119, September 2010 [3] F. Claessen: Smart grid control, Figure 3: Combined behaviour of a freezer and a fan master thesis, TU Delft, the Nether- lands, 2012. Please contact: Gerard J.M. Smit University of Twente, The Netherlands E-mail: [email protected] Figure 4: Combination of multiple electricity supply sources 30 ERCIM NEWS 92 January 2013

32 The Gamification of Agent-Based Smart Grids by Radu-Casian Mihailescu, Matteo Vasirani and Sascha Ossowski Envisioning a smart grid scenario pervaded with controllable loads, we are working on the use of game mechanics to drive consumer behaviour towards efficient grid-wise use of energy. In order to cope with the challenges faced by current electricity networks, we aim to build a game layer on top of the electricity grid infrastructure and to use gamification as a catalyst for change, encouraging participation of customers in the energy field towards lower carbon generation and increased grid resilience. Whilst the worldwide energy demand is erators (photovoltaics, wind turbines) is enabling the network to work more effi- expected to be 30 percent higher in 2040 expected to significantly increase the ciently is developing solutions that compared to 2010, more of the same will volatility in generation. enable consumers to benefit from oper- not be sufficient to meet the challenges ating under a set of given constraints, in this sector. In our work we advocate a With these considerations in mind, we provided by the network operator for consumer-centric approach, by encour- have developed a mechanism for coor- flattening demand. To this end, a non- aging a collaborative participation of dination of the various actors in the cooperative game is proposed in which customers for a more efficient use of grid, which operates on two levels agents representing consumers deter- energy. Presently, in the traditional (Figure 1). The distinction is correlated mine optimal strategies that can be model of electricity networks, users are with the general organization of elec- translated into adaptations of their not active participants, but merely repre- tricity markets. On one hand, the base- demand in order to maximize user satis- sent sinks for electricity. Much like the load energy requirements are traded in faction, represented as a utility function internet, the electricity grid will be inter- the day-ahead market, which settles the that aggregates energy and comfort active, in the sense that power flow will amount and price of electricity for each costs. We then analytically derive the become bidirectional and energy man- time-slot for the following day. On the Nash equilibrium of the game - a strategy profile at which no user may gain by unilateral deviations - proving the existence and uniqueness of such an operating point for our given setting. Under this framework we introduced decentralized algorithms for non-coop- erative environments where agents self- ishly aim to maximize their utility, which we prove to converge to the Nash solution. Secondly, the intraday market poses additional challenges, given that the system needs to react in real-time to sudden changes of the aggregated gen- eration profile, in order to balance supply from intermittent renewable resources. Of course, agents as con- Figure 1: Consumer-centric smart grid configuration sumers in the network need to cooperate in order to attain the critical mass for significantly influencing the total agement will become distributed in the other hand, the intraday market is demand curve and thus be enabled to grid due to the many actors involved in responsible for mitigating the trading of participate in this market. This can be the operation of the system (eg micro- short-term energy delivery. The differ- made possible assuming that consumers generation, plug-in vehicles, control- ence in terms of time-horizon, means can engage in an online, self-interested lable loads) [1]. that different approaches are required to negotiation for shifting loads and thus determine a viable scheduling solution adapting their demand. However, con- Instead of varying supply in order to for these two markets. sidering the real-time constraints we match demand, in a consumer-centric reorient towards protocols that mini- model users can adapt their energy con- Firstly, to address the day-ahead market mize computational and communica- sumption to balance excesses or short- requirements we introduced a game- tion costs, unlike the classical stability ages in the grid. This is particularly theoretic framework for analysing our concepts in coalitional game theory (eg important during peakload intervals, setting given that consumers can be the Core), where the computational when expensive, carbon-intensive gen- modelled as individual and independent complexity of the problem makes it erators need to be activated to cope with decision makers whose behaviour intractable for all but the simplest of the high demand. Moreover, the impacts all other consumers and the problems at hand. Here, we formulate increased penetration of renewable gen- grid as a whole. The key issue for the problem in terms of a cooperative ERCIM NEWS 92 January 2013 31

33 Special Theme: Smart Energy Systems game where agents learn prediction increasing participation of customers in uncertainty and minimal information in models regarding potential coalition the energy field. energy domains, in proc.PRIMA'11, partners and thus, can respond in an D. Kinny, J. Yung-jen Hsu, G. agile manner to situations that are Acknowledgements: This work is sup- Governatori, and A. K. Ghose (Eds.). occurring in the grid, by means of nego- ported by the projects AT (CON- Springer-Verlag, Berlin, Heidelberg, tiating and formulating a priori solu- SOLIDER CSD2007-0022, INGENIO 472-483. tions, with respect to the estimated 2010) and OVAMAH (TIN2009- behaviour of the system. 13839-C03-02). Please contact: To sum up, we are providing a novel Link: Radu-Casian Mihailescu, Matteo framework for managing the grid effi- http://www.agreement-technologies.org Vasirani, Sascha Ossowski ciently by envisioning a game layer on Rey Juan Carlos University, Madrid, top of the electricity grid infrastructure References: Spain that allows us to implement a consumer- [1] R-C. Mihailescu, M. Vasirani, and Centre for Intelligent Information centric approach as one key driving S. Ossowski: An investigation of Technologies factor for a new vision of the grid via emergent collaboration under E-mail: [email protected] A Marketplace-Based Approach to demand- Side Management in the Smart Grid by Luigi Briguglio, Massimiliano Nigrelli, Frank Eichinger, Javier Lucio Ruiz-Andino and Valter Bella Market mechanisms facilitated by Future Internet technologies will help manage energy demand in the smart grid. New scenarios, stakeholders and services will need to be considered in the development of strategies to manage energy generation, distribution and consumption. Power grids are facing new challenges as detailed information on energy con- To this end, our team is defining and a consequence of an increase in the use of sumption and the introduction of energy describing the architecture for a market- renewable energy sources (RES). The contracts with (dis)incentives will raise place-based approach to demand-side fluctuation in power supply from these awareness of how the behaviour of management. In particular, the architec- energy sources results in insecure and groups of consumers could impact on ture is based on the idea of having two unstable operating points. Furthermore, energy production/distribution and con- marketplaces (see Figure 2): the integration of distributed energy sequently encourage more efficient 1. A B2C marketplace where consumers resources (DER), as well as distributed energy use. can subscribe to different demand- information systems, poses serious sta- side-management programmes; bility and security challenges in power In this context, the role of marketplaces 2. A B2B marketplace where demand- network operation and management, is fundamental for the management of side managers can trade their bids necessitating reinforcement of the distri- energy prices, contracts and incentives for shifting energy demand. bution grid. In the scientific literature, a based on the energy available/used in wide range of applications of control the grid (as shown in the Figure 1). The The marketplaces are enabled by Future theory have been suggested for balancing following scenario demonstrates how Internet technologies (identified by the renewable energy generation and con- the eco-system could work. yellow boxes in Figure 2) coming from sumption. Above all, to overcome these the FI-WARE Generic Enablers (GE) problems we need to start considering When energy retailers experience infrastructure [2]. new scenarios, stakeholders and services. energy shortages, they buy energy at intra-day exchanges. When renewable The role of the demand-side manager In this article, we present a proposal for production is low, this can be very might be taken on by the local distribu- the adoption of market mechanisms in expensive and a better option might be tion system operator, by energy retailers order to allow grid users and grid opera- to ask their customers to consume less or by dedicated demand-side-manager tors to implement, for instance, contrac- energy within a certain time frame. companies. tual peak shaving policies, thus avoiding Similarly, grid operators monitor the issues associated with the stability of electricity grid and may want to ask con- The architecture assumes an infrastruc- electricity distribution networks. sumers to temporarily reduce their con- ture where consumers have installed at sumption in order to achieve grid sta- their premises an energy-efficiency The proposed solution is based on the bility. In a nutshell, such actions con- control system (EECS) that monitors engagement of all stakeholders (such as tribute to peak shaving in order to and controls energy consumption by energy retailers, grid operators, cus- improve grid performance and reduce changing their programming parame- tomers) in the controlled eco-system the need for auxiliary non-environmen- ters, and allows DSM signals to be [1]. We make the assumption that more tally friendly peak production. received from the demand-side man- 32 ERCIM NEWS 92 January 2013

34 Market Mechanisms ager based on the subscribed conditions for Solving Market Market and user preferences. This is only pos- Grid Issues Electricity Price Definition sible with an entity (normally software) DSM Contract that can ensure security and effective Management programming of the appliances to avoid Customer Energy Information Service Enterprise Enterprise any inconvenience. Shape Demand Monitor In the architecture, customers have Electricity Home Energy usage Operation Operation access to the B2C marketplace where Network optimization Energy Monitoring they can see different offers for demand-side management programs Generation Generation from demand-side managers. These Transmission Transmission DER DER Customer Customer offers could be based on real-time tariff Distribution Distribution schemes and users can then choose to enter into a contract with one of these Figure 1: Marketplaces for DSM: key functions services with the aim of reducing the monthly bills. The offers might be cou- pled with energy contracts and the based on surveys (among different [2] FI-WARE Security Generic Enablers: incentives may vary in terms of how stakeholders) and a selection process https://forge.fi-ware.eu/plugins/ much money is paid (or deducted from which allowed identification of relevant mediawiki/wiki/fiware/index.php the bill) when a customer actually reacts use-case scenarios both from a business [3] FINSENY - Future Internet for to load-shifting requests. By using such and a technical perspective. The specifi- Smart Energy: a service, customers allow the grid cation of the architecture and its compo- http://www.fi-ppp-finseny.eu/ operator to send demand-response sig- nents will be completed by February nals to their energy-efficiency control 2013. system. The signals are used to initiate Please contact: actions in the appliances connected to Acknowledgments: Luigi Briguglio, Massimiliano Nigrelli the energy-efficiency control system, in This work was partially supported by the R&D Lab - Engineering Ingegneria order to schedule operations (e.g. elec- EU project FINSENY - Future INternet Informatica S.p.A, Italy tric-vehicle charging, starting a for Smart ENergY (FP7/2007-2013, E-mail: {luigi.briguglio, machine or appliance, reducing or Grant Agreement n. 285135). The massimiliano.nigrelli}@eng.it increasing the temperature of building authors gratefully acknowledge by a given amount). It is important that Pasquale Andriani, Giuseppa Caruso Frank Eichinger, contracts allow the customer to override and Pierre Plaza Tron. Their activities in SAP Research Karlsruhe, Germany DSM signals at any time in order to the first year of the FINSENY project E-mail: [email protected] make manual decisions about how and allowed to provide the useful inputs for when to use the contracted energy. achieving the current results. Javier Lucio Ruiz-Andino, Telefonica Investigacin y Desarrollo, The results, partially presented in this References: Spain article, will be further refined in the [1] EU Commission Task Force for E-mail: [email protected] European project FINSENY (Future Smart Grids Expert Group 3: Roles INternet for Smart ENergY) [3] and in and Responsibilities of Actors involved Valter Bella particular in the work package 6 in the Smart Grids Deployment, Telecom Italia, Italy Electronic Marketplace for Energy. http://ec.europa.eu/energy/gas_electrici E-mail: Modelling and specification activity is ty/smartgrids/doc/expert_group3.pdf [email protected] # ( # (+ ! ! ! # ( " # (+ ! # ' & # ( ! # ( * ! $ % ! + * " # ( + ! # (' #.+ & # (# , ! * & & * " ) # (+ ' & & " &* , ! - , " # ) # ( " Figure 2: Electronic Marketplace for # ( * ! Energy: High-Level Architecture ERCIM NEWS 92 January 2013 33

35 Special Theme: Smart Energy Systems Using an Intelligent Management System for Smart Residential Storage Systems by Vicente Botn, Mximo Prez, Adolfo Lozano-Tello and Enrique Romero In order to transition the present day electrical grids into the smart grids of the future, it is essential that we develop effective energy storage systems. Intelligent software systems can be employed to manage stored energy and smart domestic devices in order to optimize local energy consumption whilst taking into account user behaviour and environmental conditions. An effective energy storage system will provide economic benefits and enable us to optimize use of the distribution grid. Figure 1: Schematic diagram of the interconnection of home appliances, and the local energy management unit (with energy storage system). Energy storage systems (ESS) [1,2] are energy and common behaviour of the residential buildings with the aim of necessary to facilitate integration of both electric distribution grids in an unobtru- controlling the load curve, and a central distributed generation systems, based on sive and transparent way. system that receives data from a group renewable energy sources, and smart of LEMUs via a conventional connec- devices, located in houses, residential and An appropriate way to control smart tion, eg ADSL, operated by telephone commercial buildings, into the power houses is to use ontologies [3] to clas- companies, that is typically found in grid. Such systems can thereby help to sify the types of appliances and sensors houses and residential buildings. The manage the load profile, optimizing the and their functionality to gain an under- data conveyed by the LEMUs are performance of the distribution grid. This standing of the load profile of a house or analysed by the central system using kind of system is typically associated residential building. Ontologies and behavioural algorithms in order to with distributed generation plants - SWRL (Semantic Web Rule Language) decide on an action and this information mainly with photovoltaic installations - rules provide a precise definition of is also communicated to the distribution but an ESS could also be used in the smart grid taxonomy and are reusable, grid operator to enable it to optimize the absence of an energy generation system, so other systems can use them to clas- load curves. In this way, the system with the aim of smoothing the load curve sify their own components and to build becomes a Central Energy Management or even controlling energy consumption rules that will allow new information to and Intelligent System (CEMIS). depending on its price-by-hour. be inferred. One proposed system to integrate smart home devices and Using this system, each event captured These environments require both the energy storage systems is IntelliDomo by sensors and actuators is recorded. ability to adapt themselves to the resi- [3]. This system (see Figure 1) com- With these data, there is a learning dents consumption habits and the versa- prises Local Energy Management units module which incorporates algorithms tility to make decisions in a variety of (LEMU), which are located in houses or in order to acquire users habits and situations. In this sense, identifying con- sumption patterns within a sequence of events in order to predict future peaks on the load curve can help us achieve energy optimization. This information could enable the system to recognize power consumption patterns and deter- mine the best way to smooth future peaks. Identification of these patterns obviously requires a previous task of learning. In a smart grid environment, learning means that the system has to Figure 2: Example of the power consumed over time in the IntelliDomo system used to analyse acquire knowledge about the use of the behaviour of the installation. 34 ERCIM NEWS 92 January 2013

36 automatically generate production rules will send the results of the data mining [2] M. Guerrero, et al: Supercapacitors: for behaviour patterns to anticipate the process to the IntelliDomoRules tool, Alternative Energy Storage Systems, users periodic actions. As a result, the which can build the rules and save them PRZEGLD system can adapt to changes in the iden- into the ontology. Thus, IntelliDomo ELEKTROTECHNICZNY, 2009 tified patterns based on the user implicit can control several LEMUS to optimize [3] A. Lozano-Tello, V. Boton: Analy- and explicit feedback. The algorithms energy consumption and manage smart sis of Sequential Events for the Recog- will analyse this dataset and look for home devices remotely. nition of Human Behaviour Patterns in frequent peaks in the load curves (see Home Automation Systems in proc. Figure 2). Once they recognize the pat- Link: DCAI 2012, AISC 151, Springer-Ver- terns, they will design a strategy to http://www.intellidomo.es lag. pp 511-518, 2012. coordinate the behaviour of the LEMU and optimize the use of the resources. References: Please contact: This strategy is carried out by gener- [1] C. Pieper, H. Rubel: Revisiting Adolfo Lozano-Tello ating SWRL rules that automate the cor- Energy Storage: There Is a Business Telefonica Chair of Extremadura responding actions. Therefore, the Case, the Boston Consulting Group, University, Spain Energy Optimization Module (EOM) (2011) E-mail: [email protected] The Last One out Turns off the Light - Optimizing the Energy Efficiency of Buildings by Lutz Ehrig and Danilo Hollosi The project S4EeB (Sounds for Energy Efficient Buildings), which has been running since October 2011, aims to optimize the performance of existing building management systems by taking into account a buildings occupancy rate using audio sensor networks as a new source of information. The overall goal of this demand driven approach is to reduce unnecessary consumption of energy for heating, ventilation, air conditioning, and lighting. In the course of the project the Fraunhofer Institute for Digital Media Technology IDMT, located in Ilmenau and Oldenburg in Germany, has been developing procedures and methods for analysing audio data in order to gain information about the occupancy rate of buildings, on the basis of which the energy consumption of a building can be optimized automatically. A major proportion (35%) of electricity are unobtrusive and have a higher user tions, this may not always be possible. consumed in commercial and public acceptance compared to video surveil- Hence, microphone arrays will be used, office buildings is attributed to heating, lance systems, for example. Unlike allowing analysis of the spatial ventilation, and air conditioning cameras, acoustic monitoring does not acoustics of the respective areas in (HVAC). Together with street and com- require a free line of sight. Also, sound order to localize sound sources. mercial lighting this accounts for more based systems require less computing than 60% of the electricity consumed in power, which is appropriate given our All acoustic sensors are connected to an office buildings across the European goal of increased energy efficiency in acoustic processing unit (APU). The Union [1]. With 50 million public build- buildings. APU combines the microphone signals ings existing across Europe, this sector of one area or a particular part of it. has a huge potential for improving The S4EeB system consists of three Based on suitable signal parameters energy efficiency. main components that have already extracted from the audio signals in the been, or will be, developed, as depicted APU, sound source localization and While modern buildings and public in Figure 1: the audio system for sound sound source separation are performed spaces use sensing technologies on a and noise recording, the acoustic pro- and the occupancy rate of a building is broad scale, for instance motion detec- cessing system for detecting acoustic determined. Initial experiments have tion, video surveillance, temperature events using a machine learning shown that the occupancy rate can be measurements and gas detection, the approach, and the management system estimated with high accuracy by using potential of sound and noise has not yet for monitoring occupancy rates and con- approaches for machine learning based been utilized for the purpose of building trolling the building automation system. acoustic event detection. Furthermore, automation. Given the importance that the modular layout of the APU allows humans attribute to sound and noise in The audio system consists of micro- users to easily modify the system so that their indoor and outdoor environments, phones to be installed inside the respec- it is capable of detecting security related sound based sensing seems an obvious tive building. Ideally, the acoustic sen- acoustic events, such as cries for help or choice to provide valuable information, sors should be distributed evenly across breaking glass. Hence, using the S4EeB such as estimates of the occupancy rate the space. However, due to architectural system will not be limited to the pro- in buildings. Furthermore, audio sensors or infrastructure conditions and limita- jects objectives. ERCIM NEWS 92 January 2013 35

37 Special Theme: Smart Energy Systems 1 8 1 8 1 8 1 8 1 8 1 8 ... ... ... ... ... ... Figure 1 system Audio System Satellite Satellite Satellite Satellite Satellite Satellite components for the Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 S4EeB project 8 Ch. ADAT Audio Signal ... 8 Ch. ADAT Audio Signal Acoustic Acoustic Acoustic Processing Processing Processing System Unit A Unit X Room 1 Room X TCP/IP Occupancy Data and Training Audio Data Building Sensors Building Management Building Management Management System Optimizer and Automation System HVACL All the data and semantic information audio data captured is analysed by algo- Milano-Linate airport and two shopping from the APU are collected and analyzed rithms developed by Fraunhofer IDMT, malls in Spain, Principe Pio in Madrid by the building management system allowing the buildings occupancy rate to and Maremagnum in Barcelona. optimizer, which is the interface to the be determined for the purpose of inte- classic building management system. grating this data into the building energy Based on the buildings occupancy rate, management system. Links: its thermal characteristics, outside http://www.s4eeb.org weather conditions, and other parame- This three-year project is funded by the http://www.idmt.fraunhofer.de ters, the optimal settings with respect to European Union, and its consortium energy efficiency and user comfort will comprises research institutes and Reference: be determined. Thermal modelling of the industry partners from four European [1] P. Bertoldi and B. Atanasiu: building is done beforehand, providing countries, who have long-standing Electricity Consumption and the basis for the best strategy considering experience in building control strate- Efficiency Trends in the Enlarged the buildings energy consumption rate gies, audiovisual applications, micro- European Union, European and the interaction of the building man- electronics and mechanical components Commission, Institute for Environment agement and automation system with the as well as in consulting and dissemina- and Sustainability, 2007. HVAC system. tion of results. The project recently fin- ished its first year of collaborative Please contact: The main contributions of Fraunhofer development of a prototype system and Lutz Ehrig, Danilo Hollosi, Fraunhofer IDMT to the project are: sound the corresponding components. In the IDMT, Germany recording, audio signal processing, and first quarter of 2013, field tests will start E-mail: [email protected], acoustic event detection. In particular, at the S4EeB demo sites, namely [email protected] Ambient Intelligence for Energy Efficiency in a Building Complex by Giuseppe Lo Re, Marco Ortolani and Giuseppe Anastasi The quest for energy efficiency currently represents one of the most stimulating challenges both for academic and industrial organizations. We address the issue of ensuring timely and ubiquitous monitoring of a potentially large building complex in order to optimize their energy consumption. Over 50% of energy produced worldwide respectively by the year 2030. Similarly, in the long term this is not an effective is consumed by the industrial sector, the goal is to reduce energy consumption approach [3]. Using an automated whilst residential and commercial build- by artificial lighting by 3% and 14% for Building Management System (BMS), ings account for about 20%, mainly due residential and commercial buildings in addition to user cooperation, is a to inappropriate use of appliances, such respectively [2]. Consequently, recent more viable solution, especially in the as heating, ventilation and air condi- years have seen a growth in research on context of ambient intelligence (AmI). tioning (HVAC) systems and artificial energy efficiency in residential/commer- AmI is a new paradigm in Artificial lighting [1]. Hence, the International cial buildings. Intelligence that relies on the assump- Energy Agencys (IEA) roadmap has set tion that the environment is permeated the goal of reducing energy consumption Studies show that providing appropriate by a set of sensors and actuators, by HVAC systems by 30% and 17% in feedback to building occupants can help remotely controllable according to residential and commercial buildings reduce overall energy consumption, but some policy, in order to bring the envi- 36 ERCIM NEWS 92 January 2013

38 ronmental conditions closer to the users ideal conditions while taking into account particular global constraints. Within the SmartBuildings project, we are currently designing and proto- typing an AmI-based BMS targeted to a building complex (eg, a campus or a res- idential complex), rather than just a single building. In our approach, the remote sensor infrastructure acts as the termination of a centralized reasoner, where sensed data are processed to extract higher-level information and perceive high-level features such as who is in a specific area or what this person is doing there (eg reading, talking, standing). Finally, a set of actuators Figure 1: The architecture for energy modifies the environmental conditions. management of a smart buildings complex Our system architecture has been con- ceived to guarantee the scalability of the straints, such as those arising from the and Networking Lab (Univ. of Pisa) is proposed solution with respect to the connection with an energy providers investigating the use of advanced number of buildings to be monitored Smart Grid infrastructure. methods for extracting individual con- and the number of different devices to sumption estimates from aggregated be used. In order to efficiently organize The project is currently in progress. We measurements. the system modules, each correspon- have already deployed a minimal proto- ding to a different logical task, we chose typal setup, by equipping one floor of This research is part of the a three-tier architecture as a model. The our department with commonly avail- SmartBuildings Projects funded by physical layer consists of sensors and able sensor nodes for monitoring the the Sicilian regional Government with actuators; the middleware layer defines typical environmental quantities (tem- European funds. a set of AmI components that can be perature, humidity, light) and with the composed to implement intelligent AmI corresponding actuators. Moreover, we Link: http://www.dicgim.unipa.it/ functionalities; the application layer are able to monitor the globally con- ~networks/ndslab/ allows for applying the monitoring and sumed energy through a remotely con- controlling rules in compliance with trollable power meter. The research References: energy constraints. group operating at the Lab of [1] U.S. Energy Information Networking and Distributed Systems Administration: International Energy From the viewpoint of deployment, the (Univ. of Palermo) is currently focusing Outlook 2010 Highlights, report building premises constitute the basic on the design of core intelligent func- DOE/EIA-0484(2010), monitored units of our system, where tionalities, such as user profiling, pre- http://www.eia.doe.gov/oiaf/ieo/highlig the sensor and actuator networks are dicting the occupancy status of the hts.html installed. These networks are heteroge- monitored premises, or detecting the [2] International Energy Agency, 2011: neous both in terms of the adopted tech- activity patterns of users, that will form Technology Roadmap, Energy- nology and of the performed moni- the basis for subsequent intelligent rea- efficient Buildings: Heating and toring/actuating tasks. Several basic soning. For instance, we have devel- Cooling Equipment, monitored units are coordinated by a oped a Bayesian inference system for http://www.iea.org/papers/2011/buildin BuildingAgent, responsible for per- multi-sensor data fusion in order to reli- gs_roadmap.pdf forming reactive control and further ably infer the presence of users from the [3] X. Jiang, et al: Experiences with data aggregation. Small buildings will available sensory information. A High-Fidelity Wireless Building have a single BuildingAgent per Probabilistic reasoning accounts for the Energy Auditing Network, in proc. building, while medium or large build- partial correlation between sensory sig- ACM SenSys 2009. ings could have more. In our vision, an nals and states, and allows the system to individual building is part of a commu- cope with noisy data, while the possi- Please contact: nity coordinated by a central orchestra bility of integrating data coming from Giuseppe Lo Re, Marco Ortolani leader, the AmIBox (see Figure 1). The multiple sensors exploits the redun- DICGIM, University of Palermo, Italy latter ensures coherence of the adopted dancy of such devices deployed E-mail: [email protected], energy saving strategy, besides pro- throughout the environment. In order to [email protected] viding high-level AmI functionalities, reduce the costs of the overall system performing intelligent reasoning and and limit its intrusiveness, the number Giuseppe Anastasi choosing the adopted energy saving of sensors should be kept as low as pos- Dept. of Information Engineering, strategy. The AmiBox could also take sible. To this end, the research group University of Pisa, Italy into account externally imposed con- operating at the Pervasive Computing E-mail: [email protected] ERCIM NEWS 92 January 2013 37

39 Special Theme: Smart Energy Systems Secure Smart Grids or Say Goodnight Vienna! by Florian Skopik, Paul Smith and Thomas Bleier With the increasing use of novel smart grid technologies, a comprehensive Information and Communication Technology (ICT) network will be established in parallel to an electricity grid that, owing to its large size and number of participants and access points, will be exposed to similar threats to those experienced on the current Internet. Whilst there have been a number of guidelines and best practices for securing future smart grids, further work is required in this area to make them readily applicable. In this article, we introduce the (SG)2 project, which aims to address these issues and provide practical advice to smart grid stakeholders in Austria. The smart grid will revolutionize elec- Europe [3], have developed guidelines for an analysis and evaluation of pri- tricity networks, allowing increased use and frameworks that can be used to mary forms of attack and attack sur- of decentralized clean energy sources. It improve the security of Smart Grids. faces. This information can be used to will make use of Information and However, because they do not consider estimate the potential impact of attacks. Communication Technology (ICT) in a factors such as local market conditions, number of ways, for example, to manage requirements and deployments, and Architectural models are examined with decentralized energy sources, such as legal constraints, these guidelines respect to threats and vulnerabilities, in from photovoltaic and the associated cannot be directly applied. order to determine the most effective protective measures against possible attacks. Electricity providers have tradi- tionally focused on ensuring the safety and reliability of their infrastructure. However, in the future, malicious attacks that hinder the increasingly net- ! worked ICT components within their " systems need to be accounted for. An # important outcome of the (SG) project will thus be a taxonomy and catalogue of countermeasures that can be applied $ # to ensure the security of smart grids for % a given threat. For a realistic risk assess- ment, the project also deals with pene- tration tests and security analysis of Figure 1: The (SG) process smart grid components. Because of the model to smart grid cyber security complexity of securing a smart grid, software tools are being developed to support the use of the guidelines and markets. In general, the smart grid will be Furthermore, they have seen limited methodologies produced in the project. more open than current electricity grids, real-world application, making it and will incorporate a greater degree of unclear how suitable they are for their A strong collaboration between measurement-based control deeper intended purpose. Our research aims to industry, research and government within the network, eg, in distribution build on these existing guidelines and In order to attain the ambitious goals of systems, based on data acquired from frameworks, in order to make them the (SG) project and to ensure the wide smart meters. This openness together functional. applicability of developed tools, major with new automatic control loops make stakeholders in important sectors smart grids more vulnerable than current An Austrian perspective: the (SG) related to smart grids in Austria need to power grids to potentially serious attacks project be involved. These stakeholders create a [1]. Personal privacy concerns abound, The goal of the Smart Grid Security well-balanced consortium including for example, issues relating to unautho- Guidance (SG) project is to study security research institutions, compa- rized access to metering data: it will be effective countermeasures to smart grid nies from the security industry sector, possible to determine detailed usage pat- security threats. The project investi- energy utilities, and governmental terns of electrical appliances, including gates and develops methods, concepts organizations. The project is led by the the television channel house occupants and process models, and accompanying AIT Austrian Institute of Technology. are watching, based on metering data [2]. software tools to minimize the risk Other large research partners are the Therefore, it is vital that security and pri- posed by cyber threats and to ensure the University of Technology Vienna, and vacy issues are a primary concern for the security of smart grids in Austria (see the Siemens AG Corporate Technology future smart grid. Figure 1). Novel approaches to the Austria. Practical security expertise in modelling of complex ICT-supported penetration testing is contributed by Various international organizations, smart grid architectures will be defined SECConsult Unternehmensberatung such as NIST in the USA and ETSI in in the project, which will form the basis GmbH. The Energieinstitut an der JKU 38 ERCIM NEWS 92 January 2013

40 Linz GmbH investigates societal impact KIRAS and by the Austrian Ministry for Power Usage Profiles, 2012 of (SG) research results; additionally the Transport, Innovation and Technology International Conference on participation of the Ministry of the (BMVIT). Information and Knowledge Interior (BM.I) and the Ministry of Engineering, Las Vegas Defence and Sports (BMLVS) ensures Links: [3] European Telecommunications the development of applicable solutions http://kwz.me/4l Standards Institute (ETSI): Smart Grid from a governmental perspective. http://kwz.me/4p Technologies; Finally, the developed guidelines will be http://www.etsi.org/website/Technologi evaluated within the context of three References: es/SmartGrids.aspx (2012) energy utilities in Austria, which, owing [1] F. Skopik, et al: A Survey on to their different sizes, require different Threats and Vulnerabilities in Smart Please contact: solutions: LINZ STROM GmbH, Metering Infrastructures, International Florian Skopik, Paul Smith, Thomas Energie AG Obersterreich Data GmbH, Journal of Smart Grid and Clean Bleier and Innsbrucker Kommunalbetriebe AG. Energy, vol 1, issue 1, September AIT Austrian Institute of Technology / 2012, pp 22-28 AARIT, Austria This two year project runs from 2012 to [2] U. Greveler, B. Justus and D. E-mail: [email protected] 2014 and is financially supported by the Loehr: Multimedia Content [email protected], Austrian security-research program Identification through Smart Meter [email protected] Preparing for the Smart Grids: Improving Information Security Management in the Power Industry by Maria Bartnes Line The power industry faces the implementation of smart grids, which will introduce new information security threats to the power automation systems. The ability to appropriately prepare for, and respond to, information security incidents is of utmost importance, as it is unrealistic to assume that one can prevent all possible incidents from occurring. Current trends show that the power industry is an attractive target for hackers. A major challenge for the power industry to overcome are the differences regarding culture and traditions, knowledge and communication, between ICT staff and power automation staff. Two major technological changes electric power engineering and com- make smart grids interesting from puter science. The technology bases an information security point of are different, and so are management view. One is that new technologies routines. Facilitating and achieving are introduced into the power understanding and well-functioning automation systems; commercial collaboration in this intersection off-the-shelf products replace pro- between ICT staff and power prietary hardware and software. The automation staff will be the most other is integration; ICT systems important task on the way to suc- and power automation systems will cessful information security incident be much more tightly connected management for smart grids. than before. Smart grids consist of complex power grids that interact Incident management is the process with equally complex ICT systems. of detecting and responding to inci- This implies that well-known infor- dents, including supplementary mation security threats like com- work such as learning from the inci- puter break-ins, industrial espionage, Figure 1: The complete incident management dents, using lessons learnt as input in malware attacks and denial-of-service process (ISO 27035) the overall risk assessments, and identi- attacks will be highly relevant for the fying improvements to the implemented power industry in the near future, if not Power automation systems and general incident management scheme. ISO/IEC already. ICT security incidents tar- communication oriented ICT systems 27035:2011 Information Security geting power automation systems, or have traditionally been operated sepa- Incident Management [1] describes the other types of SCADA systems, are not rately. There have been limited, if not complete incident management process science fiction - they are already hap- zero, logical connections between them, as consisting of five phases; 1) Plan and pening. We have had Stuxnet, Duqu and and they have served quite different pur- prepare, 2) Detection and reporting, 3) Flame, and we should expect to see poses. The staff operating the two sys- Assessment and decision, 4) Responses, more of the kind in the near future. tems tend to have different backgrounds; and 5) Lessons learnt. ERCIM NEWS 92 January 2013 39

41 Special Theme: Smart Energy Systems Current practice in the power industry Future activities performed, observing the work in prac- In order to study current practice on Qualitative interviews give us a large tice will give invaluable additional how information security incidents are amount of information. There is, how- knowledge. Having knowledge of both being detected and responded to, and to ever, a risk of getting the perfect pic- theory and practice will enable us to identify needs for improvements in ture, how things should be done compare the routines actually imple- order to be prepared for the introduction according to the book, and not just mented, suggest realistic improve- of smart grids, we are now performing actual practice. We would therefore like ments, and hence make a valuable con- qualitative interviews. Our interviewees to follow-up the interviews by running tribution to the industry. represent three different roles in a set of retrospective group interviews at large distribution system operators selected DSOs. If, or when, they experi- The longterm goal is to contribute to an (DSO): ence a high-impact incident, we would efficient incident management process - Chief Information Officer (CIO) like to go through the complete course in smart grid environments. /Head of ICT of events in order to understand how the - Chief of Information Security organization responded to that specific The project is being carried out at Officer(CISO)/Head of ICT Security incident. Questions we are interested in NTNU, in close cooperation with - Head of control room/power exploring include how the incident was SINTEF and the Norwegian Smart Grid automation systems detected, reported and resolved, in Centre. The project period is 2011- which ways they followed their plans, 2015. We chose these particular interviewees and if not, how, and in particular why, with the intention of identifying current they deviated from their plans. Reference: cooperation and possible synergy [1] ISO/IEC 27035:2011 Information effects from future cooperation, and We would also like to study smaller Security Incident Management viewing the overall management DSOs to see whether there are differ- system in general. Both technical meas- ences regarding their current practices Link: ures and human actions are being iden- compared to those of the larger DSOs. http://www.item.ntnu.no/people/person tified, including how the post-incident alpages/phd/maria.b.line/start analysis is performed. Such an after- There are usually quite a few differ- math typically includes information ences between theory and practice. Please contact: sharing, lessons learnt, and how experi- Observation is therefore also included Maria Bartnes Line ences are transferred into the overall in our plans for the near future. While Dep. of Telematics, NTNU, Norway information security routines in the the interviews give much insight in how Tel: +47-45218102 organization. incident management is planned and E-mail: [email protected] Cybersecurity in the Smart Grid by Magnus Almgren, Davide Balzarotti, Marina Papatriantafilou and Valentin Tudor In the past, the easiest way to attack the electrical grid would have been to physically access and destroy components. However, with the introduction of the smart grid and its increased dependence on information and communication technologies (ICT), the future grid may be vulnerable to pernicious cyber attacks performed remotely. In CRISALIS and SysSec, we are studying the properties of the envisioned smart grid to enable us to anticipate and mitigate future attacks against this critical infrastructure. In Europe and elsewhere, the electrical Many of the new security issues in the There are also challenging new prob- grid is being transitioned into the smart smart grid are well-known problems in lems originating from the intersection grid in order to increase flexibility and the information and communication between the electrical engineering and accommodate large scale energy pro- technology (ICT) domain, such as ICT domains, for example where a duction from renewable sources. This buffer overflows in devices and sloppy cyberattack (buffer overflow) in turn transition involves, among other steps, implementations of cryptographic pro- affects properties of the electrical grid the installation of new, advanced equip- tocols. However, the solutions from the (power quality), which in turn may ment for example, the replacement of more mature ICT domain may not be propagate back to the ICT domain (vul- traditional domestic electrical meters directly applicable to the smart grid due nerability of control loop) [2]. An inter- with smart meters - and remote commu- to resource-constrained devices (smart disciplinary approach is required to nication with devices for example, meters), the life cycle of components identify possible solutions to these allowing remote access to an unsuper- (there will always be legacy systems) or problems. vised energy production site. Together the impossibility of immediately shut- with the new functionalities, this transi- ting down and patching a machine that In SysSec, a network of excellence in tion introduces concerns about how the needs to run 24/7. Other issues originate Europe, and CRISALIS, a European technology can be misused by adver- from the electrical and power engi- research project, we are working on saries [1]. neering domain (device tampering). improving the security in critical sys- 40 ERCIM NEWS 92 January 2013

42 tems, in particular the smart grid, plinary nature of this complex through two orthogonal environment. For this reason, approaches. One major problem projects such as SysSec and is the lack of cross-domain CRISALIS, which bring expertise in both ICT security together experts from different and power engineering. Being a domains, are crucial at this network of excellence, SysSec stage. organizes several activities to bring together researchers and CRISALIS may be contacted at practitioners from different [email protected] domains. For example, we SysSec may be contacted at the organized a summer school for corresponding [email protected] students across Europe for a Figure 1: In SysSec, we organize events so that people project.eu, followed in twitter hands-on approach to learn more from different domains can meet and discuss (twitter:syssecproject) and about reverse engineering of interdisciplinary problems related to the smart grid. Facebook (http://www.face- malware targeting critical infra- book.com/SysSec). structures. To our surprise, we hit the ceiling on the number of students we date security claims made by vendors could accept within less than a week of and increase the overall security of the the announcement, forcing us to create a deployed components. One of the first References: waiting list. This points to the need for deliverables will be open-source [1] National Institute of Standards and better education in this area and we will fuzzers for protocols used in this Technology Interagency: Guidelines also include modules for hardware secu- domain. By working closely with indus- for Smart Grid Cyber Security rity and critical infrastructure protection trial partners, the goal is to provide new (NISTIR 7628), vol. 1-3, as part of the effort in SysSec to provide tools to detect intrusions and effective http://csrc.nist.gov/publications/PubsNI a common curriculum on cyber security. techniques to analyse infected systems. STIRs.html#NIST-IR-7628, 2010 [2] Costache et al: Remote control of Another major problem hampering the Even though the smart grid is a neces- smart meters: friend or foe?, EC2ND- analysis of security properties of the sity, it is important to understand the 2011, Gothenburg, Sweden. smart grid is the proprietary nature of security risks before complete systems the technologies and protocols are deployed and interconnected across Please contact: involved: there are few open source Europe. Learning from and avoiding Magnus Almgren tools available to perform an in-depth simple problems that have already been Chalmers University of Technology, analysis of a system. For this reason, we encountered in the ICT domain, we may Sweden are developing a toolset in CRISALIS focus on the new types of threats that Tel: +46 31 772 1702 that can be used by researchers to vali- arise as a consequence of the interdisci- E-mail: [email protected] CoppEnd A Security System for Power Equipment by Dimitrios Serpanos, Athanasios Safacas and Dimitrios Stachoulis Copper theft is emerging as a significant problem in the evolution and operation of critical infrastructures, such as power grids, transportation networks and water facilities. The problem has become acute due to inadequate security measures for infrastructure as well as a lack of a strict legal framework for transportation and trading of metals. CoppEnd (Copper - defEnd) is a security system designed to protect systems that use copper, focusing on power systems and their components, such as transformers. Electricity networks constitute critical modifications, and new routings, while installed equipment. However, the evolu- infrastructures worldwide, since they sup- installing advanced software systems in tion is limited by external factors that can port almost all economic and social activi- control centres. Electricity providers are severely damage, causing them to mal- ties. Importantly, electricity networks are incorporating new tools and processes in function. One of the main problems, scaling to accommodate the increasing order to offer electricity to consumers in especially in Greece, is the unreliable use of renewable energy sources and the an optimal way, based on both demand operation of electrical equipment owing ever-increasing power demand. Efficient and production levels at any given point to the theft of copper in their components. interconnection of new electrical power in time. The evolution of conventional plants with the distribution networks leads power grids to smart grids requires cor- A significant problem in the power dis- stakeholders to introduce extensions, rect and unhindered cooperation of all tribution network is the destruction of ERCIM NEWS 92 January 2013 41

43 Special Theme: Smart Energy Systems Medium Voltage (MV) transformers, that use high-tech security components, the operator, since a movement may not because of their copper content such as CoppEnd, discourage individuals be a malicious attack but another inci- (approximately 150-200 kg in each). from attacking the network and enable dent, eg an intense lightning strike. Their damage has disastrous effects on their tracking when equipment is stolen. When a transformer is removed from its social and economic activities due to This approach differs from the conven- original location, CoppEnd alters its the loss of millions of euros during long tional approaches that use mechanical operation, enabling a self-powered mode power-outs (hours or even days at a means, with little success. via a battery system that is effectively time). These incidents have a very nega- kept charged through the power network tive impact on operators, who need to CoppEnd continuously records the itself; the trigger is enabled when normal replace the transformers and to compen- location of the equipment to which it is power interruption is detected or when sate their customers. Importantly, theft attached. It contains a GPS tracker the included acceleration or pressure incidents sometimes result in serious equipped with a GSM transmission sub- sensors give a command. When on bat- injuries or even loss of life. system. The GPS tracker detects its tery, CoppEnd transmits at longer inter- exact geographical coordinates, while vals, in order to avoid detection and CoppEnd is a security system for the the GSM subsystem transmits the coor- lengthen its battery lifetime. protection of MV transformers in the dinates to the control centre, which network. CoppEnd enables early detec- positions it on the map. Thus, an admin- Overall, CoppEnd is a smart grid com- ponent and not just a security system. Its sophisticated architecture and opera- tion, employing sensors and specialized middleware, discourages copper thieves and enables their tracking, while pro- viding additional information to the operator, such as the exact location of the transformers in the area, evaluation of the quality of the transformers oil and emergency detection, including cases of natural disasters. We envision extending the use of CoppEnd in the future to protect smart grid components for water supply, sew- erage, and railway companies that face similar problems. Link: http://www.isi.gr/en/projects/coppend Figure 1: CoppEnd overview References: [1] T.M. Overman, R.W. Sackman, T.L. Davis, B.S Cohen: High- tion of human intervention to the istrator at a control centre is able to Assurance Smart Grid: A Three-Part system as well as tracking of the trans- identify whether the initial, correct Model for Smart Grid Control portation of components, thus enabling location remains stable. System control Systems, in proc. of the IEEE, 99(6), timely intervention by authorities. Thus, occurs through text messages to pp. 1046 - 1062, June 2011 transformers are protected, even in iso- CoppEnd, which enable operation [2] A.R. Metke, R.L. Ekl: Smart Grid lated areas. modes (on/off/standby) or change the security technology, Innovative Smart coordinate reporting intervals (default Grid Technologies (ISGT), 2010 , vol., The Greek power distribution network, is 30 seconds). Additionally, CoppEnd no., pp.1-7, 19-21 Jan. 2010 operated by the Hellenic Electricity includes a microphone, controlled by Distribution Network Operator S.A., the control centre, for the transmission Please contact: spans more than 230,000 km, mostly in of real-time voice data that may be Dimitrios Serpanos agricultural and isolated areas. The net- useful, depending on the event or emer- ISI/RC ATHENA, Greece work size prohibits its continuous and gency. CoppEnd is attached on a trans- E-mail: [email protected] uninterrupted physical monitoring, thus former so that the existent electromag- enabling the illegal detachment of trans- netic field does cause interference and Athanasios Safakas formers and removal of their copper. The so that it cannot be extracted without Hellenic Electricity Distribution increasing incidents, targeting more than causing damage to the transformer; this Network Operator S.A., Greece 155,000 medium-to-low voltage units, ensures its appropriate activation. E-mail: [email protected] need to be addressed not only by a stricter legal framework, but also Any deviation from the original location through adoption of sophisticated protec- constitutes an event that is reported to the tion systems [1-2]. Protection systems control centre. Events are evaluated by 42 ERCIM NEWS 92 January 2013

44 PowerAPI: A Software Library to Monitor the Energy Consumed at the Process-Level by Aurlien Bourdon, Adel Noureddine, Romain Rouvoy and Lionel Seinturier Energy consumption by information and communication technologies (ICT) has been growing rapidly over recent years. Comparable to the civil aviation domain, the research community now considers ICT energy consumption as a major concern. Several studies report that energy consumption is an issue during all steps of a computers life, from hardware assemblage, to usage, and dismantling. Research in the area of Green IT has proposed various approaches to save energy at the hardware and software levels. In the context of software, this challenge requires identification of new development methodologies that can help reduce the energy footprint. To tackle this challenge, we propose PowerAPI, a tool to quantify this energy consumption, by providing an application programming interface (API) that monitors, in real-time, the energy consumed at the granularity of a system process. Two per cent of the global energy con-

45 sumption, equivalent to the annual

46 production of eight nuclear plants, a number of studies agree that Information and Communication Technologies (ICT) energy consumption has become a major issue. Several studies aim at reducing this energy footprint at each step of the computer's Life Cycle Assessment (LCA). At the usage stage, this involves not only building new kinds of energy-

47 efficient hardware but also acquiring an understanding of the energy impact of software and how we can influence it. In Figure 1: PowerAPI architecture order to provide a basis for this knowl- edge, the Inria ADAM project team has developed a library, named PowerAPI, 500 milliseconds, displayed on my con- top of PowerAPI, demonstrate the bene- providing an application programming sole., can be transcribed into fits of this software library. interface (API) to monitor in real-time PowerAPI as illustrated by the instruc- the energy consumed at the granularity tion displayed in Figure 2, where What is the energy consumption of of a system process. Process(123) is the process #123, 500 the programming languages? milliseconds the real-time refresh First, we want to compare the imple- Monitor energy spent at the process period and CpuListener a listener mentations of a given algorithm, using level responsible for reporting energy con- several programming languages. The Unlike current state-of-the-art tech- results highlight that the choice of a pro- nology, PowerAPI does not require any gramming language has an impact on external device to measure energy con- the energy footprint of the application sumption. This is a purely software itself. Interpreted languages consume approach where the estimation is based more energy than compiled ones. This on energy analytical models that charac- result can be intuitive as a first insight, terize the consumption of various hard- but raises the question of the rise of ware components (eg CPU, memory, interpreted languages, especially with disk). PowerAPI is based on a highly Figure 2: Example of PowerAPI requirement their recent increase in use on the modular architecture where each server-side by the Web community. module represents a measurement unit sumption on the user console. (2) Details about this experiment are avail- for a specific hardware component (see Efficient, because the library is an able in [1]. Figure 1). actor-based framework in which the user builds the library by choosing Where are energy hotpots located One objective of PowerAPI is to provide modules to consider for the users par- within an application? a simple and efficient way to estimate ticular requirements. PowerAPI is thus Second, we want to delve deeper into the energy consumption of a given limited to the users needs, avoiding any the previous result, by monitoring not process. (1) Simple, because API is extra computational cost. only the process, but also the process' close to the user requirements. For source code itself. The aim of this instance, the question: What is the CPU Use cases experiment is to develop a tool that can energy consumption of my process The following two research contribu- provide developers a real-time cartog- #123. Please give me fresh results, every tions, which have been developed on raphy of energy hotspots in their appli- ERCIM NEWS 92 January 2013 43

48 Special Theme: Smart Energy Systems cation code. This study reports that, PowerAPI. In particular, in collabora- References: when using a Web container like Jetty, tion with several Green IT actors, such [1] A. Noureddine et al: A around 80 % of the global energy con- as ADEME, Green Code Lab, and Preliminary Study of the Impact of sumption under stress conditions is GreenIt.fr, we are currently working on Software Engineering on GreenIT, attributable to a mere seven classes (out a project called Web Energy Archive [email protected] of 146). This information is useful for that aims at monitoring and comparing [2] A. Noureddine et al: Runtime developers who want to know what the energy consumption of Internet Web Monitoring of Software Energy changes could be made to reduce the sites. Hotspots, ASE12, energy footprint of the application. Details about this experiment are avail- Links: Please contact: able in [2]. PowerAPI: Romain Rouvoy http://abourdon.github.com/powerapi- Inria, LIFL, University Lille 1, France On-going work akka ADAM project-team On-going work includes the develop- ADAM project-team, E-Mail: [email protected] ment of additional use cases for http://adam.lille.inria.fr Smart Energy Management for Greener Supercomputing by Mohammed el Mehdi Diouri, Olivier Glck and Laurent Lefvre A supercomputer is a system built from a collection of computers performing tasks in parallel in order to achieve very high performance. An exascale machine is a supercomputer capable of performing more than 1018 floating point operations per second (1 Eflop/s). Such extreme-scale systems are needed by 2018 in order to meet new scientific challenges, such as enabling highly sophisticated genome calculations and proposing individualized patient treatments. As they will gather hundreds of millions of cores, exascale supercomputers are expected to consume enormous amounts of energy (between 25 and 100 MW). In addition to being very large, their power consumption will be very irregular. Furthermore, the applications that will run on such extreme-scale systems will need energy consuming services such as fault tolerance, data collective operations. In order to manage the execution of extreme-scale applications on future supercomputers in a sustainable and energy-efficient way, we propose a framework called SESAMES: Smart and Energy- aware Service-oriented Application Manager at Extreme-Scale [1]. Since the power consumption of these nodes. An energy sensor is plugged to suming the least amount of energy, the large scale systems is enormous and each node and measures the current cleanest energy, at the lowest financial dynamic, SESAMES establishes a per- power consumption. SESAMES col- cost and without exceeding the power manent negotiation with the energy lects these energy logs. In order to capped by the energy provider. If no provider (Figure 1). Through this gather the execution context, solution exists, SESAMES informs the dialog, SESAMES gives the energy SESAMES also establishes a dialog user that the requested reservation is not supplier an agenda of the estimated with users of supercomputers. This possible. If there exists a unique solu- power consumption. It also gathers interaction with the user occurs at the tion that optimizes all the criteria, from the energy supplier the agenda for: moment of reserving computing nodes SESAMES makes the corresponding energy price, energy sources used, and and just before running applications and reservation and informs the user about power capping. The price, energy services (see Figure 1). it. Otherwise, SESAMES computes the source (coal, sun, wind etc) and solutions that optimize each criterion threshold limit will vary at different In order to run their applications, users separately and asks the user to choose times. Supercomputer users may prefer send to SESAMES a reservation request between the solution that minimizes the to consume energy at times when it is to book some of the supercomputer's financial cost or the one that provides the cleanest and least expensive whilst nodes. A reservation request consists of the cleanest energy. energy providers may adapt the supply the number of nodes required, the reser- according to demand and may choose to vation duration, the earliest possible Once the reservation is done, SESAMES disable some production of energy pro- start time and the latest possible start gives the user the opportunity to estimate duced from a polluting source during time. In order to make a reservation, and reduce the energy consumption of times of low power use. SESAMES solves a multi-criteria opti- the different services (like fault toler- mization problem by taking into ance) that he would like to run while exe- Furthermore, in order to reduce global account several constraints. It attempts cuting his applications. For each service, energy consumption, SESAMES is able to allocate the supercomputing nodes at several versions are possible. The least to act directly on the supercomputer the time desired by the user by con- energy consuming version depends on 44 ERCIM NEWS 92 January 2013

49 approach involves dynamically adjusting the performance level of a resource according to the performance level the application and users really need. The green leverages proposed depend on the idle periods predicted and on the rights assigned by the super- computer administrator to the user. The energy consumption of these green solutions is estimated by SESAMES in order to make the user aware of the energy savings generated by the green solutions suggested. Reference: [1] M. Diouri, O. Gluck, and L. Lefevre: Towards a novel Smart and Energy-Aware Service-Oriented Manager for Extreme-Scale Figure 1: Global infrastructure: external interactions with SESAMES applications, First Workshop for Power Grid-Friendly Computing (PGFC'12), San Jose, USA, 2012. the applications features. Hence, the To reduce the energy consumption of first step to consuming less energy is supercomputers, SESAMES proposes Please contact: to choose the least energy consuming to apply some green leverages at the Mohammed El Mehdi Diouri version for each service. Thanks to the component level: shutting down or Inria Avalon Team interaction with the user, SESAMES slowing down an idle resource compo- ENS Lyon, Inria, LIP Laboratory, takes into account the application fea- nent (processor, memory, disk, etc.). University of Lyon, France tures and the user requirements in order The shutdown approach involves Tel : +33472728009 to provide an energy estimation of the dynamically turning off unused E-mail: [email protected] different versions of the services resources and turning them back only http://perso.ens-lyon.fr/mehdi.diouri requested by the user. when they are needed. The slowdown AI4B: Accountable IT Infrastructures for Optimizing Supply Chains in Bioenergy Symbiotic Networks by Theodore Dalamagas and Antonis Kokossis Project AI4B aims to develop a data-driven approach to establish bioenergy networks of biomass feedstock producers and collectors, aiming to remove bottlenecks in the biomass supply pipeline and develop accountable, economically and environmentally sustainable bioenergy practices. Biomass - biological material from three times higher than the EU average, existing bottlenecks in the biomass living or recently living organisms - rep- and thus there are plenty of sources for supply pipeline, and develop economi- resents the largest renewable energy raw biomass materials. cally and environmentally sustainable source and is the only renewable energy bioenergy practices [1]. In AI4B, we source that is based on carbon. Raw The AI4B project, co-financed by EUs focus on symbiotic bioenergy networks. materials, such as wood, agricultural Regional Development Fund and by Symbiotic networks [2] is an innova- residues, food waste, industrial waste national resources, comes at a critical tive environmental practice that brings and co-products have captured the stage for the country, in line with a together companies from all business interest of markets and industries world- major effort to develop renewable sectors through material trading and wide following the uncertainties in fossil energy sources, which is planned to be sharing assets to add value, reduce costs fuel supply and the need to reduce one of the driving forces to turn reces- and benefit the environment. greenhouse gas emissions. Involvement sion into growth. AI4B mobilizes aca- in the biomass business in Greece has demic partners, IT SMEs and regional We focus on symbiotic networks with several advantages, since Greece reports development agencies to develop inno- biomass supply chains involving bio- an agricultural share of the GDP that is vative IT infrastructures to remove mass producers (BP) and biomass col- ERCIM NEWS 92 January 2013 45

50 Special Theme: Smart Energy Systems Figure 1: cessing and supply chain configurations Overview of the based on what-if analysis tools. AI4B infrastructure Accountability in biomass symbiotic networks AI4B will adopt the LOD paradigm to expose: (a) currently available spatial data infrastructures (SDI) as geoLOD data, and (b) process, performance and logistic models as biomass LOD data. LOD is considered worldwide as a driving force to provide transparent and accountable processes. In AI4B, all stages of biomass supply chains will be transparent. Every citizen or public body will have access to supply chain data, and be able to query decisions and demand ramifications. The AI4B project is just about to start, and involves the following partners: Athena Research Center, National Techn. University of Athens, Centre for lectors (BC). BPs are regional players matchmaking and retrieval services so Renewable Energy Sources, that provide storage points or produc- that BCs are able to identify BPs (and Eratosthenes SA, CLMS Ltd, Tero Ltd, tion facilities. BCs are industrial players vice versa) according to their Kenakap SA, Aenol SA. that collect biomass from BPs and supply/production requirements. employ valorization technologies for Acknowledgements: The authors energy production. Our vision is Regional development and planning acknowledge Dr. Ioanna Papamichael, twofold: The dispersed geographical distribution Nontas Tsakonas, Dr. Yannis Zorgios, To set up infrastructures to support of biomass raises difficulties for the Stratos Arampatzis, Panayiotis Patras, networking between BPs and BCs estimation of biomass quantities, the Sevasti Basdeki for their contribution in by providing intelligent matchmak- evaluation of technological pathways this work. ing and retrieval services. for biomass energy and the determina- To leverage societal interaction and tion of the final cost for BCs. BCs need Links: participation as an intrinsic part of the to explore several alternative delivery http://en.wikipedia.org/wiki/Industrial_ network by giving every citizen or models to determine cost-effective bio- symbiosis public administration access to sup- mass supply chains, and to identify the http://linkeddata.org ply chain data. geographic distribution of the economi- http://www.esymbiosis.gr cally exploited biomass potential. http://www.cres.gr/kape/energeia_politi A data-driven symbiotic network Spatial planning can assist biomass s/energeia_politis_biomass_uk.htm We follow a data-driven approach to logistics, and provides a means for maintain biomass symbiotic networks. exploring and evaluating alternative References: BPs register on the network and provide scenarios of biomass supply chains [1] G. Grant et al: Information and information about biomass feedstock, (characteristics, transportation cost, Communication Technologies for Indus- for example: type, quantity and location selection/sizing/costing of biomass pro- trial Symbiosis, Journal of Industrial of production. BCs also register and duction site) to determine optimal con- Ecology, 10(5), 2010 provide information about available figurations. Key components in such an [2] R. Isenmann, K. Chernykh: The biomass processing options and supply infrastructure are: (a) processing tech- role of ICT in Industrial Symbiosis Proj- requirements. We will adopt the Linked nology models that predict the output of ects. Environmental Informatics and Open Data (LOD) paradigm to set up biomass-related process blocks (eg the Industrial Environmental Protection: biomass data infrastructures (BDI) for entire biodiesel process that converts Concepts, Methods and Tools, 2009. BPs and BCs. The LOD paradigm biomass and fossil into new products) involves practices to publish, share, and given input information (eg material Please contact: connect data on the Web. Linked, and energy streams), (b) performance Theodore Dalamagas means that data are interlinked under models that evaluate the economical Athena Research Center, Greece diachronic naming schemes, so that performance, energy consumption, and E-mail: anyone can easily combine them and greenhouse gas emissions, and (c) [email protected] extract knowledge in multiple ways. logistic models that address logistics Open, means that data are fully dis- and supply chain considerations. AI4B Antonis Kokossis closed under open formats, open will provide services to set up cost- National Technical University of licenses, and machine readable formats. effective biomass supply chains, and Athens, Greece AI4B will set up a LOD base providing explore alternatives in biomass pro- E-mail: [email protected] 46 ERCIM NEWS 92 January 2013

51 Hydrodynamics-Biology Coupling for Algae Culture and Biofuel Production by Olivier Bernard, Jacques Sainte-Marie, Bruno Sialve and Jean-Philippe Steyer Biofuel production from microalgae represents an acute optimization problem for industry. There is a wide range of parameters that must be taken into account in the development of this technology. Here, mathematical modelling has a vital role to play. The potential of microalgae as a lated velocity fields can provide source of biofuel and as a techno- lagrangian trajectories of the logical solution for CO2 fixation is algae. The resulting light pattern the subject of intense academic to which each cell is submitted and industrial research. Large- when travelling from light (sur- scale production of microalgae has face) to dark (bottom) can then be potential for biofuel applications derived. It will then be repro- owing to the high productivity that duced in lab experiments to study can be attained in high-rate photosynthesis under realistic raceway ponds [1]. light patterns. Based in France, Green Stars is a It is clear, however, that many large research and development Figure 1: A typical raceway for cultivating microalgae. Notice complex physical phenomena project involving scientists and the paddlewheel which mixes the culture suspension. have to be added to our model, industry whose aim is to explore Photo: INRA such as the effect of sunlight on the use of micro-algae, particularly water temperature/density, evap- in the form of third generation biofuels. vided by the paddlewheel; consequently oration and external forcing (wind). The program has enormous promise, with our strategy is to develop efficient Moreover, some microalgae species do this resource potentially offering a models and numerical tools to repro- not only swim in the water (advection tremendous solution for the major eco- duce the flow induced by the paddle- plus diffusion effects) but also deposit. nomic developments of the coming wheel and the evolution of the biolog- decade. Some microalgal species have far ical species within this flow. Here, more efficient growth, by photosynthesis, mathematical models can greatly help References: than terrestrial plants. Moreover, they can us to reduce experimental costs. [1] B. Sialve, N. Bernet and O. accumulate oils or sugars, which can be Bernard: Anaerobic digestion of turned into biodiesel or bioethanol. Owing to the high heterogeneity of microalgae as a necessary step to make Twenty to 30 tons of oil per hectare per raceways due to gradients of tempera- microalgal biodiesel sustainable, year are expected to be extracted from ture, light intensity and nutrient avail- Biotechnology Advances, 27:4, p. 409- micro-algae, compared with six tons ability through water height, we cannot 416, 2009 from the palm trees and a little more than use depth-averaged models (Shallow [2] E. Audusse, M.-O. Bristeau, M. one ton from rapeseed. Water type models). We adopt instead Pelanti and J. Sainte-Marie: Approxi- more accurate models that have recently mation of the hydrostatic Navier- The objective of Green Stars is to lay the been proposed [2]. These models are Stokes system for density stratified foundations for the entire sector, from particularly appropriate for representa- flows by a multilayer model, Kinetic energy generation to waste recycling and tion of these free surface systems. For interpretation and numerical solution, the production of compounds of interest. hydrodynamics, we use the incompress- J. Comp. Phys., 230, p. 3453-3478, Green Stars also plans to play a long-term ible hydrostatic Navier-Stokes equa- 2011. role in this field by training technicians, tions, forced by a paddlewheel-like engineers and researchers. move. The biological dynamics are rep- Please contact: resented by an improved and distributed Olivier Bernard, Inria team Biocore, The role of mathematical modelling (in space) model that includes light Sophia-Antipolis, France and simulations effect on algae growth and carbon [email protected] One of the key challenges in the produc- storage depending on nitrogen limita- tion of microalgae is to maximize algae tion. Jacques Sainte-Marie, Inria team growth with respect to the used exoge- ANGE, Paris-Rocquencourt, France nous energy (paddlewheels, pumps, We show, through 3D numerical simu- [email protected] etc.). lations, that our approach is capable of discriminating between situations of Bruno Sialve , INRA Narbonne, France There are a large number of parameters rapidly moving water or slow agitation, [email protected] that need to be optimized, including: the choosing an optimal water height or characteristics of the biological species, proposing initial conditions for the bio- Jean-Philippe Steyer, INRA Narbonne the raceway shape and the stirring pro- logical variables. Moreover, the simu- [email protected] ERCIM NEWS 92 January 2013 47

52 Research and Innovation A Projector as Mobile Visualization device on an Assistive Robot by Paul Panek, Christian Beck, Georg Edelmayer, Peter Mayer and Wolfgang L. Zagler Even small robots have great potential to support the elderly. We have created a prototype of a LED projector module that enables a small humanoid robot to project text, graphics and video to a surface next to the user. Figure 1: Prototype of LED projector unit (left) mounted on the humanoid NAO robot (right) The research project KSERA (Knowledgeable SErvice Robots for Aging) develops a social assistive robot that sup- ports older persons, especially those with Chronic Obstructive Pulmonary Disease (COPD), in their daily activ- ities and care needs. Independence and overall quality of life can be enhanced when individuals are able to engage in self- management of their disease. The small NAO humanoid, from French company Aldebaran, is used for the robotic platform. It serves as an Figure 2: NAO robot with a LED projector unit on its back during interface between the user and the system, which is video communication between KSERA user (sitting on a couch) and embedded in a smart home environment enabling ubiquitous a call centre operator. monitoring of the users activities and health status and of the environmental conditions. tion regarding the actual state and emotional situation of the As the humanoid robot we used is comparatively small (57 user than can be achieved with an audio only connection. The cm tall) it is not able to carry HCI devices such as a tablet PC video connection is particularly useful in emergency sce- and to present it to a sitting or standing user. To overcome narios, when the operator of the emergency centre needs to this restriction an innovative LED projector module mounted quickly assess the severity of the situation. on the robots back was developed in the KSERA project. Discussion and conclusions The projector unit on the robots back (Figure 1) projects A mobile video solution is much appreciated by experts and text, graphics and video information towards a wall next to potential end users. Despite the promising results, some limi- the user. With video phone communication a camera in the tations of the current prototype system exist, namely: The robots head is used to transmit the users video stream low brightness of the LED projector used (30 ANSI lumen). towards the communication partner (Figure 2). This is By using blinds on the windows and artificial light in the test intended for social communication (friends, family mem- room, a realistic environment could be set up but it was at the bers), video communication to medical services and in the lower limit of ambient brightness recommended for living case of emergency (eg a fall). areas. It is expected that brighter projectors will be available in the near future overcoming this drawback. The main innovation is considered to be the mobility of the solution, as the assistive robot with its projector equipment Further work will involve validating the integrated KSERA can come to the user anywhere in the apartment. The text and prototype with older users in Austria and Israel in near to real graphics to be projected are provided by the KSERA intelli- life environments. gent server; similarly, the timing of the beginning and end of projections is controlled by the KSERA state machine. The Acknowledgement projector component is responsible for the visualization of This work was partly funded by EC in the FP7 KSERA the information to be shown to the user. For initial laboratory project. evaluation, a workshop with a small group of experts from the care domain was organized following a qualitative Links: approach. The experts agreed that the quality of audio and http://www.ksera-project.eu video was good and would be adequate for video communi- http://www.aldebaran-robotics.com/ cation in the given context. http://www.aat.tuwien.ac.at/ksera/ Added value was found to be in the persistence of projected Please contact: text and graphic for somewhat more complex messages (eg Paul Panek, Vienna University of Technology, Austria daily agenda, air quality information) and in the video phone Tel: +43 1 58801 187713 capability. The latter provides more comprehensive informa- E-mail: [email protected] 48 ERCIM NEWS 92 January 2013

53 able to easily express their needs, and an efficient query reso- ATLAAS-P2P: lution mechanism should efficiently find relevant resources and limit the number of messages exchanged. Common tech- A Two-Layer Architecture niques for searching resources in P2P systems are based on range queries over a set of attributes. However, the volume for Approximated Search of resources in a P2P network may be very large and hetero- geneous, and users rarely have the appropriate knowledge in Peer to Peer about the available resources to allow them to properly for- mulate their queries. A user may, however, be able to define by Ranieri Baraglia, Patrizio Dazzi, Matteo Mordacchini their ideal resource and ask the search system to find and Laura Ricci resources close to such an entity. Thus, instead of having to specify precise ranges on all attributes, the user simply has to ATLAAS-P2P is a two-layered peer-to-peer (P2P) provide an example of what is needed. architecture for developing systems, providing resource aggregation and approximated discovery in P2P This mechanism would simplify the work for users and lead networks. It gives users a flexible and easy means of to a more efficient exploitation of the search system. searching for resources and also benefits resource Moreover, it would provide an effective infrastructure for providers by assisting users to find them. advertising for resource providers, facilitating their dis- covery by users. The process of identifying useful resources in a P2P network is highly dependent on query formulation. Users should be ATLAAS-P2P consists of a P2P system that provides flexi- bility in the way that users express their requirements and an effective solution for enabling users requests to reach resource providers. It is based on a two-layer architecture, where peers in the network represent the resources of providers. The lower layer is an unstructured, gossip-based, P2P network allowing peers to efficiently gather in logical G ET GET groups of nodes representing similar resources. The role of G ET GET this layer is to automatically capture the affinities existing G GET ET between resources belonging to different providers and to group them in common communities. Those communities distributively elect their own representatives. The profiles of PUT PU these representatives are used as the descriptors of such com- DHT DHT w with ith L LSH SH PUT PUT support su pport munities. Once elected, each representative registers itself on L the higher layer, a structured, DHT-based network. L The structured network has been extended to support approx- imated searches over the community representatives. Users PUT PUT can submit the queries to this network by providing sample resources consisting of prototypes of the resources they are L searching for. Gossip-based protocols are used to find and select similar representatives to forward the query within Client C lient their community. This means that when none of the resources R esource Resource available in the system matches a users request, the user is offered suitable alternatives. Figure 1: The overall architecture of ATLAAS-P2P The overall architecture of ATLAAS-P2P is sketched in Figure 1. Peers (circles) form distinct communities built on a F-Measure Comparison with ERGOT similarity basis in the unstructured gossip-based layer. Each Our Solution ERGOT community elects a representative, denoted with an L in the 1 figure. Each representative is in charge of registering itself on the higher structured layer. Users of the systems 0.8 (rhombus) can query the structured network searching for the 0.6 resources they need. Results will consist of the most signi- ficative community profiles and their representatives. The 0.4 representatives will act as entry points to further forward queries to the peers of the represented community. 0.2 Instead of searching for peers whose profile is similar to that 0 specified by the user this architecture searches for communi- query 1 query 2 query 3 query 4 query 5 query 6 query 7 query 8 query 9 query 10 query 11 query 12 query 13 query 14 query 15 query 16 query 17 query 18 query 19 query 20 ties. This reduces both the number of comparisons to per- form and the number of peers to contact. As a consequence, Figure 2: ATLAAS-P2P performances the amount of generated network traffic also decreases. ERCIM NEWS 92 January 2013 49

54 Research and Innovation The ability of ATLASS-P2P to return significant resources ances. Recently, P2P protocols have also been exploited for has been tested using a dataset of 200 word domain labels information diffusion and aggregation, including resource organized in a hierarchical structure built by exploiting the discovery and system monitoring and community-based WordNet domain [1]. The content of this dataset has been information dissemination. These applications introduce used to generate textual descriptions. Such descriptions have new requirements to P2P protocols, since information fresh- been assigned to 5000 peers according to a Zipf distribution ness, rather than information precision, is their main focus. for building the peer profiles. Epidemic-based (also known as gossip-based) P2P protocols ATLAAS-P2P performances are presented in Figure 2, in are unstructured communication approaches that disseminate which they are compared with those provided by ERGOT information in a manner similar to the spread of viruses in a [2], a solution for this task, which is based on semantic biological community. They are often used to solve problems overlay networks. that might be difficult to tackle in other ways owing to the complex structure and dimension of the network and the fast References: rate of information change. [1] B. Magnini, G. Cavaglia: Integrating subject field codes into wordnet, in proc. of LREC 2000 These developments have fostered an increasing interest [2] G.Pirr, D.Talia, P.Trunfio: A DHT-based semantic within the research community in the conception and design overlay network for service discovery, Future Generation of novel epidemic protocols. A typical issue is the need to Computer Systems 28 (4), 2012 consider classical non-functional requirements, such as scale and performance as foundational aspects of protocol design. Please contact: Indeed, these protocols fit networks comprising hundreds of Patrizio Dazzi, ISTI-CNR, Italy thousands of nodes characterized by frequent changes in E-mail: [email protected] shared data and affected by considerable churn rates. Since it is unfeasible to obtain access to thousands of machines worldwide, the performances and limits of epi- demic protocols are normally studied through simulations. In Epeerdemics: this sense, well-designed simulators facilitate the develop- ment of new protocols allowing for the simulation of many A Peer-to-Peer Simulator nodes within limited computational units. Properly designed simulators also ease the deployment of the protocols on a real Targeting Epidemic-Based infrastructure with minimal disruption to the code. Protocols In the last decade, several P2P simulators have been pro- posed. Essentially, they differ in the level of abstraction pro- by Patrizio Dazzi and Emanuele Carlini vided, the programming language used and flexibility in developing protocols. To ease the evaluation and comparison Since the late nineties, peer-to-peer (P2P) protocols of protocols, these simulators are often bundled with several have become increasingly popular. Traditionally, these well-known protocols. Unfortunately, only a few P2P simu- systems have been used to implement widely lators are specific to unstructured protocols and only a subset distributed applications, such as file-sharing services, as of them provide a bundle of epidemic-based protocols as they provide efficient support for the discovery and baselines for testing. distribution of information. We developed Epeerdemics with the aim of filling this void. Several different structured P2P protocols have been pro- Epeerdemics is an extension to Overlay Weaver [1], an posed for distributed networks. Structured, in this context, overlay construction toolkit widely diffused in the P2P com- refers to the protocols ability to organize network links and munity, mainly used for developing structured protocols. data to provide specific guarantees and bounds on perform- Epeerdemics is specifically designed to ease the develop- Figure 1: Epeerdemics protocol architecture. 50 ERCIM NEWS 92 January 2013

55 ment and testing of epidemic-based protocols. Protocols can either be developed from scratch or by extending the proto- gRecs: Exploiting the cols provided with Epeerdemics. To this end, Epeerdemics provides two of the most used epidemic-based peer sampling Power of data Mining protocols: Cyclon and Vicinity. Cyclon realizes an inexpen- sive peer sampling to build an overlay characterized by a Techniques for Efficient random-graph structure. Vicinity is a semantic peer-sam- pling based protocol built on top of Cyclon that realizes an Computation of Group overlay network in which the links express similarity between peers. Recommendations From an operative point of view, each epidemic protocol is by Kostas Stefanidis and Kjetil Nrvg based on two threads, one active and one passive. The active thread is activated by a timer. When activated it selects one gRecs is a research prototype system designed for or more peers to communicate with and sends them (a subset providing suggestions to groups of users about items of of) its own knowledge. These interactions awaken, in the potential interest. In particular, gRecs proposes an selected peers, the respective passive threads. Each passive extensive model for group recommendations based on thread answers by sending back to the sender (a subset of) its recommendations for items liked by similar users to the knowledge. group members. This is achieved with the use of data mining techniques. More specifically, since the main The amount and kind of information exchanged as well as bottleneck is to identify the most similar users to a given the information retention strategy is protocol dependent. one, we model the user-item interactions in terms of Epeerdemics supports programmers in implementing these clustering and use the extracted clusters for predictions. choices and strategies. To implement an epidemic protocol with Epeerdemics, a programmer has simply to extend two Recommendation systems provide suggestions to users Java classes: One for implementing the selection strategy to about a variety of items, such as movies and restaurants. The decide which peers to communicate with and one to specify large majority of these systems are designed to make recom- the information retainment strategy, ie which information to mendations for individual users. However, there are contexts store or discard. in which the items to be suggested are intended for a group of people, rather than an individual; for instance, recommenda- Here at ISTI-CNR, Epeerdemics has been used successfully tions for restaurants, tourist attractions, movies, TV pro- for developing several different epidemic-based protocols. grams and holiday destinations. Recent approaches try to sat- These range from simple enhancements to existing protocols isfy the preferences of all group members either by creating a to the definition of completely new protocols, including joint profile for the group and suggesting items with respect GoDel [2] a protocol for building Delaunay overlays and [3] to this profile or by aggregating the single user recommenda- a gossip-based overlay construction for large scale online tions into group recommendations. gRecs opts for the second games. approach owing to its greater flexibility and potential for efficiency improvements. References: [1] Kazuyuki Shudo, Yoshio Tanaka, and Satoshi gRecs proposes a framework for group recommendations fol- Sekiguchi: OverlayWeaver: An Overlay Construction lowing the collaborative filtering approach. The most promi- Toolkit, Computer Communications, 31(2):402412, 2007 nent items for each user of the group are identified based on items that similar users liked in the past. Users are considered [2] R. Baraglia, P. Dazzi, B. Guidi, and L. Ricci, GoDel: similar if there is an overlap in the items consumed. In partic- Delaunay Overlays in P2P Networks via Gossip, IEEE ular, the two types of entity that are dealt with in recommenda- International Conference on Peer-to-Peer Computing, 2012 tion systems, ie, users and items, are represented as sets of rat- ings, preferences or features. Users initially rate a (typically [3] E. Carlini, L. Ricci, and M. Coppola: Reducing Server small) subset of items and ratings are expressed in the form of Load in MMOG via P2P Gossip, in proc. ACM Workshop preference scores. A recommendation engine estimates prefer- on Network and System Support for Games, Venice, Italy. ence scores for the unrated items and offers appropriate rec- ommendations. Once the unknown scores are computed, the k Please contact: items with the highest scores are recommended to users. Patrizio Dazzi and Emanuele Carlini, ISTI-CNR, Italy E-mail: [email protected], To efficiently aggregate the single user recommendations [email protected] into group recommendations, we leverage the power of a top-k algorithm. We employ three different aggregation designs: (i) the least misery design, capturing cases where strong user preferences act as a veto, (ii) the most optimistic design, capturing cases where the most satisfied member is the most influential one, and (iii) the fair design, capturing more democratic cases. To deal with reliability issues, we introduce the notion of support in recommendations to model how confident the recommendation of an item for a user is. ERCIM NEWS 92 January 2013 51

56 Research and Innovation Group recommendations are presented to users along with explanations about the reasons that the particular items are Utility-Theoretic Ranking being suggested. Explanations are given as text using a tem- plate mechanism. for Semi-Automated Text A main problem of this approach is to identify the most similar Classification users for each user in the group. A solution that involves no pre-computation requires computing the similarity measures by Giacomo Berardi, Andrea Esuli and Fabrizio between each user in the group and each user in the database. Sebastiani To avoid exhaustively searching for similar users, we perform some pre-processing steps offline. In particular, we propose Researchers from ISTI-CNR, Pisa, have addressed the building clusters of similar users, considering as similar those problem of optimizing the work of human editors who users that have similar preferences. To partition users into proofcheck the results of an automatic text classifier clusters we employ a bottom-up hierarchical agglomerative with the goal of improving the accuracy of the clustering algorithm. Initially, our algorithm places each user automatically classified document set. in a cluster of his own. Then, at each step, it merges the two most similar clusters. The similarity between two clusters is Suppose an organization needs to classify a set of texts under defined as the minimum similarity between any two users that a given classification scheme, and suppose that this set is too belong to these clusters (max linkage). The algorithm termi- large to be classified manually, so that resorting to some form nates when the similarity of the closest pair of clusters violates of automated text classification (TC) is the only viable option. a user similarity threshold . Ideally, the most similar users to a Suppose also that the organization has strict accuracy stan- specific user are the members of the cluster that the user dards, so that the level of accuracy that can be obtained via belongs to. Recommendations are computed based on the state-of-the-art TC technology is not sufficient. In this case, preferences of these cluster members. Figure 1 shows a high the most plausible strategy to follow is to classify the texts by level representation of the architecture of our system. means of an automatic classifier (which we assume here to be generated via supervised learning), and then to have a human editor proofcheck the results of the automatic classification, Figure 1: correcting misclassifications where appropriate. gRecs architecture - -%&,'7/#&,8"3&' %&,'7/#&,8"3&' overview The human editor will obviously inspect only a subset of the automatically classified texts, since it would otherwise make )*)+",'-%&,' (()*)+",'-%&,' Clusters Clusters o off Users Users no sense to have an initial automated classification phase. A .&/ .&/&,"#0, &,"#0,! software system could actively support the human editor by !$# !"# ranking, after the classification phase has ended and before 6' 1&,%0/"+'2&30**&/4"50/' 1&,%0/"+'2&30**&/4"50/' !%# the inspection begins, the automatically classified documents ! "#"$"%&' !"#"$"%&' .&/ &,"#0,' .&/&,"#0,' in a such a way that, if the human editor inspects the docu- ments starting from the top of the ranking and working down the list, the expected increase in classification accuracy that .,09:'2&30**&/4"50/' .,09:'2&30**&/4"50/' .&/&,"#0,' .&/&,"#0,' derives from this inspection is maximized. We call this sce- nario semi-automated text classification (SATC). Our results show that employing user clustering consider- A common-sense ranking method for SATC could consist in ably improves the execution time, while preserving a satis- ranking the automatically classified texts in ascending order factory quality of recommendations [1]. To deal with the of the confidence scores generated by the classifier, so that high dimensionality and sparsity of ratings, we envision sub- the top-ranked documents are the ones that the classifier has space clustering to find clusters of similar users and subsets classified with the lowest confidence [1]. The rationale is of items for which these users have similar ratings. that an increase in accuracy can derive only by inspecting misclassified documents, and that a good ranking method is We designed and developed the gRecs system at the simply the one that top-ranks the documents with the highest Norwegian University of Science and Technology in probability of misclassification, which (in the absence of Trondheim, Norway, funded by the ERCIM Alain other information) we may take to be the texts which the Bensoussan Fellowship Programme, in collaboration with classifier has classified with the lowest confidence. Irini Ntoutsi and Hans-Peter Kriegel from the Ludwig Maximilian University of Munich, Germany. We have recently shown [2] that this strategy is, in general, suboptimal. Simply stated, the reason is that, when we deal Reference: with imbalanced TC problems (as most TC problems indeed [1] I. Ntoutsi, K. Stefanidis, K. Nrvg and H-P. Kriegel: are [3]) and, as a consequence, choose an evaluation measure Fast Group Recommendations by Applying User Cluster- - such as F1 - that caters for this imbalance, the improve- ing, in proc. of ER 2012. ments in effectiveness that derive from correcting a false positive or a false negative may not be the same. Please contact: Kostas Stefanidis, Kjetil Nrvg, NTNU, Norway We have devised a ranking method for SATC that combines, E-mail: [email protected], [email protected] via utility theory, (i) information on the probability that the 52 ERCIM NEWS 92 January 2013

57 document is misclassified, and (ii) information on the gain in overall accuracy that would derive by proofchecking it. A Radio Telescope We have also proposed a new evaluation measure for SATC, of the Superlative called Expected Normalized Error Reduction (ENER). Since different users will inspect the ranked list down to a certain by Ton Engbersen inspection depth, ENER uses a probability distribution over inspection depths as a parameter. ENER measures then The worldwide community of Radio-Astronomy has the expected value (over this probability distribution) of the envisioned building a very large, highly sensitive radio reduction in error that inspecting a ranked list down to the telescope partly in South Africa and partly in Australia specified depth would bring about. by 2020. The total effective area of this radio telescope should approach one square kilometer and therefore it We have used ENER as the evaluation measure for our is called the Square Kilometre Array (SKA). The SKA experiments, which we have run on a standard text classifica- instrument is expected to generate Exabytes of data per tion dataset. The results show that, with respect to the day which need to be processed and reduced, such that common-sense baseline method mentioned above, our approximately 1 Petabyte per day is left to be stored for utility-theoretic ranking method is substantially more effec- later use by Radio Astronomers. tive, with computed improvements ranging from +16% to +138%. Current expectations for the SKA are that the low frequency array (70 450 MHz) and the initial mid frequency ( 450 The approach we present is extremely general, since it 3000 MHz) will each comprise about 500,000 antenna ele- applies straightforwardly to cases in which evaluation meas- ments while the high frequency array ( 3 10 GHz) will con- ures different from F1 are used; multivariate and non-linear sist of approximately 3000 dishes. A quick calculation evaluation measures can be handled too, provided they can assuming no beamforming before Nyquist sampling results be computed from a standard contingency table. By using in 3.5 1015 samples/s or 300 ExaSamples per day (assuming our method, it is also easy to dynamically provide the human 24 hour operation). Processing this is clearly beyond the editor with an estimate of how accurate the classified set has capabilities of even the fastest supercomputers one can envi- become as a result of the proofchecking activity. sion by 2020. The streaming and real-time nature of the SKA makes it unlikely that supercomputers are ideally suited for References: this application, like in LOFAR [1]. A significant research [1] A. Esuli and F. Sebastiani: Active Learning Strategies and development effort is therefore needed. For IBM, with for Multi-Label Text Classification, in proc. of ECIR our focus on future Big Data and Big Data analytics, this is a 2009, Toulouse, FR, 2009, pp. 102-113 highly interesting field of research: it promises to make ana- lytics low cost and energy efficient. We have named the [2] G. Berardi, A. Esuli, and F. Sebastiani: A Utility-Theo- project DOME after the protective astronomical telescope retic Ranking Method for Semi-Automated Text Classifica- covering. tion, in proc. of ACM SIGIR 2012, Portland, US, pp. 961- 970 DOME A five-year, 33 million Euro project has been defined [3] H. He and E. Garcia: Learning from imbalanced data. between IBM Research Zrich and ASTRON, funded by IEEE Transactions on Knowledge and Data Engineering, the Dutch Ministry of Economic Affairs, Agriculture and 21(9), 1263-1284. Innovation and the Province of Drenthe, The Netherlands. The objective is to investigate novel exascale computing Link: technologies and concepts, with a focus on energy-efficient http://nmis.isti.cnr.it/sebastiani/Publications/SIGIR12.pdf data processing, data storage, and nano-photonics at a funda- mental level. In addition, the DOME project will collaborate Please contact: with Small and Medium Enterprises and other academic Fabrizio Sebastiani, ISTI-CNR, Italy partners in the Netherlands to stimulate economic activity Tel: +39 050 3152 892 through supporting the development and testing of new high- E-mail: [email protected] performance computing applications. Research Projects In DOME, seven research tracks are defined: 1. Algorithms and Machines: The goal is to design a whole- system bounds framework enabling system-design space exploration in the early phases of the SKA implementation and thus guide the design decisions for platforms which will hold future exascale systems. A methodology already in development in the IBM Laboratory in Zrich forms the basis: analytical models and equations tie application properties, device technology and compute architecture trends together to arrive at predictions of performance[2], power and hardware cost. ERCIM NEWS 92 January 2013 53

58 Research and Innovation 2. Access Patterns: The SKA will generate approximately These seven work-streams will be performed in close coop- one Petabyte per day, data which will need to be kept on eration between ASTRON and IBM in the ASTRON & IBM storage media and made available for future analysis and Center for Exascale Technology, Dwingeloo, The distribution. New storage technologies as well as very low Netherlands and the IBM Research Laboratory, Zrich, power storage technologies (magnetic tape) will be inves- Switzerland. We expect to achieve exciting results in the area tigated and through the hopefully automatic learning of exascale computing, applicable not only to SKA and radio of the system about typical usage patterns of this radio astronomy but also to Big Data analytics. After all, isnt the astronomy data, the system can autonomously decide on SKA the ultimate Big Data analytics challenge? which storage tier the data will be stored, and moved when its access is anticipated. References: 3. Nano-photonics: Transport of data will remain a major [1] M. de Vos, A.W. Gunst, R. Nijboer: The LOFAR Tele- cost factor in the SKA system. A particular focus will be scope: System Architecture and Signal Processing, in put on the processing of signals in the optical domain. proc. IEEE, vol. 97, Issue 8, pp 1431- 1437, DOI 4. Micro servers: Through carefully selecting the appropriate 10.1109/JPROC.2009.2020509, 2009, IEEE Journals & computing hardware and energy-efficient peripheral hard- Magazines ware, this work-stream tries to pack as much computing power in as small an area as possible under severe ener- [2] P. Stanley-Marbell, V. Caparrs Cabezas, R. P. Luijten: gy limitations. Pinned to the walls - impact of packaging and application 5. Accelerators: This work-stream will address questions properties on the memory and power walls, in IEEE Inter- around what makes an architecture energy-efficient, and national Symposium on Low Power Electronics and Design easily programmable. (ISLPED11), pp 5156, Fukuoka, Japan, 2011 6. Compressive Sampling: capture and processing of analog signals is traditionally done in 2 steps: sampling and com- [3] J. W. Romein: An Efficient Work-Distribution Strategy pression. Usually sampling is done at the Nyquist frequen- for Gridding Radio-Telescope Data on GPUs, in ACM cy, followed by often lossy compression. Why sample at International Conference on Supercomputing (ICS12), this high frequency to then discard samples? Venice, Italy, 2012 7. Real-Time Communications: The objective of this work- stream is to create a computing architecture able to real- Please contact: time process high-bandwidth data motion and compute Ton Engbersen, IBM Research GmbH, Switzerland intensive workloads on an Exascale-class system. E-mail: [email protected] Figure 1: The Astronomical Data Deluge 54 ERCIM NEWS 92 January 2013

59 The Green-Wake Project Targets both Air Traffic Security and Airport Throughput by Sbastien Lugan and Benoit Michel The Green-Wake project has developed a new LIDAR (Light Detection And Ranging) scanner able to detect vortices generated by airplanes taking off and landing, Figure1: Hazard detection and feedback loop thereby increasing airport security while decreasing the time between two consecutive flights. Scanning mirrors Wake vortices and wind shear are potential causes of acci- Oscillating mirrors provide the required volume scanning in dents and injuries to passengers and crews on aircraft of all front of the laser. Moving mirrors back and forth ten times types. They cannot be detected by sight and result in sudden per second may seem a trivial task compared with setting up disruption to the aircrafts trajectory, potentially affecting lasers, light amplifiers, and interferometers. However, in safety. There are currently few options for protection from order to give useful results, the scanning system required the these phenomena, and the main way to ensure safety with development of a new, stiff, lightweight composite mirror regard to wake vortices is to impose mandatory separation with a beryllium face skin mounted on a honeycomb core. between aircraft. The resulting delays can affect the airports operating performance. Green-Wake has developed and Airfield trials tested a Doppler LIDAR system able to detect both wake Airfield trials were conducted over a two-week period at vortices and wind shear in front of an aircraft that may soon Charleroi-Brussels South (Belgium) airport in October 2012. be inserted into the noses of commercial aircrafts or on the With over 60 commercial airplanes taking off each day in side of the runways and will help avoid the related hazards. front of the Green-Wake LIDAR prototype over the trial period, a lot of data were acquired to validate the concept and LIDAR challenges to give enough information to the team that is now facing the Wake vortex and wind shear detection has been the focus of a last step of this research effort: its implementation in a com- few research programmes funded by Europe and the US, and mercially deployable system. the LIDAR technique offers a technical solution to this problem. Since LIDAR requires the use of a laser, there is a A complementary European consortium considerable challenge to designing and building a system The Green-Wake project was funded by the European that has the performance required to detect hazards, is suit- Commission from 2008 to the end of 2012. The project con- able for installation on aircraft, and also meets the safety and sortium included EADS Deutschland GmbH, Universit cost requirements of the aerospace industry. catholique de Louvain (Belgium) and Technical University of Sofia (Bulgaria), the Aeronautical Research and Test Institute The LIDAR system may be installed in the airplanes nose or VZLU (Czech Republic), and the German Aerospace Center on the ground on the side of an airport runway. Fixed locations DLR, and a number of European high-tech SMEs. offer several advantages, such as minimizing the weight and complexity of on-board equipment and providing measure- Link: http://www.greenwake.org/ ments and warnings to all approaching and departing aircraft. Please contact: The Doppler LIDAR Sbastien Lugan, Benot Michel, UCL-ELEN, Belgium The Doppler LIDAR system uses an ultraviolet laser that E-mail: [email protected], emits a beam into the air. The beam is slightly decolimated [email protected] and diverted 10 times per second by a pair of orthogonal oscillating mirrors in order to scan a volume rather than a Figure 2: False colour representation of wind shear and wake vortex single line. When the beam encounters aerosols in the in front of a departing airplane. scanned volume, it is reflected back to its source, where a semi-transparent mirror lets the received photons pass through to a detector. Interference fringes between the orig- inal laser beam and reflected beam are used to measure the aerosol particles radial velocities. The useful air volume scanned by the system covers distances from 50 to 200m in front of the LIDAR system and the scanning area is approxi- mately 120m wide and 50m high. As the beam is invisible and decolimated to present a minimum 50 mm width, it poses no safety threat to any person in the measurement zone. ERCIM NEWS 92 January 2013 55

60 Events International Workshop Building Automation and Facilities that could be understood by industry Management. participants. In turn, industry practi- on Information tioners were invited to have demonstra- The event counted the attendance of the tion stands displaying their solutions Technology for Energy successful figure of 60 participants, and to discuss relevant issues and col- nine high quality full papers (out of 20 laboration opportunities with special- Applications submissions) and nine short papers. The ists. Moreover, student presenters had first day of the event was dedicated to an opportunity to collect positive and by Paulo Carreira and Vasco Amaral eight tutorial sessions, and the second constructive comments from a diversi- day to research papers organized into fied panel. The National Engineers Association in four presentation sessions on the sub- Lisbon, Portugal hosted the jects of Smart Grids, Energy The goal of this workshop was to estab- International Workshop on Information Consumption Profiling, Energy Data lish itself as an impacting discussion Technology for Energy Applications Management and Intelligent Load forum on the topic of Information (IT4ENERGY 2012) organized in con- junction CITI and INESC-ID research laboratories held on September 6-7th, 2012 and sponsored by IEEE Portugal Section. This was the first event on the topic that counted more than 60 partici- pants including senior researchers, PhD and Msc students as well as industry practitioners. Information technology (IT) for energy applications is growing in relevance Figure 1: Professor Kumar Figure 2: One of the stands displaying ISAs mostly due to the need of IT solutions to Venayagamoorthys keynote address Cloogy , a home energy management support the growing dynamics of energy system markets as well as to an upsurge of interest in intelligent systems to opti- Control. In his keynote address entitled Technology for Energy Applications. mize energy usage within homes, build- Intelligent Scalable Monitoring and An expanded and revised selection of ings as well as in commercial and indus- Control Technologies for Smart Micro- the best papers is now being organized trial facilities. These IT tools will have Grids and Grids, Professor G. Kumar to be published as a post-proceedings the ability to (i) coordinate production Venayagamoorthy, a world renowned volume. with consumption in scenarios of specialist in Smart Grids, highlighted demand variability, (ii) integrate and the multi-disciplinary nature of his own Links: analyze data from multiple sources and research, which underscore the reason http://it4energy.com (iii) assist users in the decision-making of this event. http://www.inesc-id.pt process as well as (iv) to intelligently http://citi.di.fct.unl.pt manage equipment and devices on The papers presented covered a number http://www.ordemengenheiros.pt/ behalf of the user. of very important topics, namely: Software for energy applications; Data Please contact: Developing IT for energy applications is mining and decision support techniques Paulo Carreira, INESC-ID and a challenging multi-disciplinary effort for energy data; Models and techniques IST/UTL, Portugal that often requires bringing together dis- for energy consumption forecasting; Tel: +351 214 233 287 (ext 5087) tinct engineering disciplines (e.g., Civil, Descriptions and characterizations of E-mail: [email protected] Mechanical, Electric and Computer energy consumption patterns; Engineering) along with specialists from Integration of energy data; Energy data Vasco Amaral, CITI and FCT/UNL, other fields such as Architecture and visualization; Sensor networks, Portugal Management. We believe that Computer metering and energy data acquisition; Tel: +351 212 948 536 (ext. 10712) Science will play a critical role not only Interoperability solutions including E-mail: [email protected] as a catalyst toward creating a homoge- middleware and protocols for energy neous body of knowledge regarding applications; Demand-side manage- information technology for Energy ment; Home and building automation Management, but also as a vehicle for applications to energy; Energy-efficient creating disruptive new solutions for control techniques; and Intelligent load energy problems. Therefore, this work- control. shop aimed at bringing together special- ists from academia with different back- This workshop also aimed at estab- grounds spanning Mechanical, lishing links between industry and aca- Electrical Engineering as well as demia. Therefore, researchers were Computer Science; Industry experts invited to present and discuss the fore- active in the fields of Energy, IT, seeable impact of their work in ways 56 ERCIM NEWS 92 January 2013

61 Systems focussed on a novel joint ERCIM/EWICS/ safety and security architecture for dependable time-triggered systems, Embedded (Cyber- adding the security aspects to the already well-studied and proven time- Physical) Systems triggered system architecture (by AIT, TU Vienna, Austrian Academy of Workshop 2012 Sciences and TTTech). by Erwin Schoitsch Todays and evolving cyber-physical Workshop participants systems (CPS) have as typical feature Following a tradition since 2006, the wide-spread distribution of nodes. In ERCIM Working Group on Dependable the presentation of the work of Embedded Systems and EWICS organ- pSHIELD, another ARTEMIS project, ized again a full day workshop during by SESM (Italy) and the Polytechnic the annual SAFECOMconference. Although there were five workshops Institute of Coimbra (Portugal), an and one tutorial in parallel, 18 regis- architecture framework was demon- SAFECOMP is nowadays an estab- tered and two partially attending partici- strated supporting security, privacy and lished international conference in the pants listened to the Cyber-physical dependability as a built-in feature in a field of Computer Safety, Reliability Systems Workshop and took part in the network of embedded nodes, improving and Security. The 31st International intensive discussions. also re-use of already verified SAFECOMP Conference took place in embedded components and systems. Magdeburg, Germany, on 25-28 The workshp was composed of six ses- The Cyber-Physical Attacker, devel- September, 2012. About 150 partici- sions: oped by Technical University of pants attended this years conference. Introduction: ERCIM, EWICS, Denmark (DTU), models attacker sce- SAFECOMP has contributed to the ARTEMIS: Embedded Systems Safe- narios addressing the peculiarities of a progress of the state-of-the-art in ty, Security and European Strategy cyber-physical adversary, which allows dependable application of computers in (providing an overview over ERCIM, the security properties of a CPS to be safety-related and safety-critical sys- EWICS, MBAT, SafeCer and R3- studied. tems since it was established in 1979 by COP) EWICS TC7, the European Workshop Dependable Embedded Systems Validation, verification and qualifica- on Industrial Computer Systems, TC7, Applications tion are issues of great importance when Reliability, Safety and Security. Secure Systems Systems Security trying to prove trust in CPS. The Validation, Verification and Qualifi- NuSMV model checker is well known Partners from ERCIM, EWICS cation in the formal methods community. The (European Workshop on Industrial Systems Safety and Trust first paper in this session presented an Computer Systems Reliability, Safety Ambient Assisted Living. interesting extension to NuSMV, and Security), from several EU- Parallel NuSMV, which was presented Framework and ARTEMIS (European Three embedded systems applications by ALES S.r.l. from Italy as part of the Embedded Computing Systems were presented in the first session of the FormalSpecs Verifier Framework for Initiative) projects atttended the confer- workshop: one paper discussed the the formal verification of complex ence and the associated workshops, and elaboration of safety requirements in embedded systems, using Simulink/ reported on related subjects. the avionic domain (by EADS, an Stateflow models. industrial paper), the second paper pre- TheERCIM / EWICS / Embedded sented the ARTEMIS robotics/ One of the ideas to considerably (Cyber-Physical) Systems Workshop autonomous systems research project improve and speed up development of workshop was jointly co-organised by R3-COP, focussing on the knowledge- safety-critical embedded systems is the the ARTEMIS projects MBAT based approach to compose robotic use of tool chains, which implies seam- (Combined Model-based Analysis and applications and tool chains for V&V less integration of different tools to Testing of Embedded System), SafeCer from a collection of building blocks in cover significant parts of the develop- (Safety Certification of Software- ontology-driven data bases, which is ment life cycle. Safety standards require Intensive Systems with Reusable considered as basis for a reference tech- qualification of tools, but are not Components) and R3-COP (Resilient nology platform for robotics and looking in-depth into the issue of inte- Reasoning Robotic Co-operating autonomous systems, as developed by gration of pre-qualified tools into tool- Systems). To be distinct from the AIT Austrian Institute of Technology, chains. The paper on Automated SAFECOMP conference mainstream, DTI Danish Technology Institute, tec- Qualification of Tool Chain Design the workshop accepted reports on nalia Spain, TU Brno, Czech Republic from KTH (Sweden) presented a prom- work in progress aiming at fruitful and other R3-COP partners. ising approach to reduce effort in quali- discussions and experience exchange. fying tool chains by automatically Reports on European or national The session on system security included analysing a tool chain model for safety research projects (as part of the required three presentations looking at different issues acknowledging the MBAT dissemination) as well as industrial aspects of secure systems. On the project. The last paper of this session experience reports were welcome. Design of Secure Time-Triggered was on a model-based development ERCIM NEWS 92 January 2013 57

62 Events approach for the design and validation researchers, developers, content of electronic control systems by simu- providers and practitioners in the field of lation, using a Data Time Flow digital libraries. Started as ECDL in Simulator, developed by AIT in the 1997, in conjunction with the activities context of the ARTEMIS project of the ERCIM-coordinated DELOS POLLUX which tackles problems on Network of Excellence on Digital the design of the next generation of Supported by ERCIM Libraries, TPDL has progressed and is electric cars. now acknowledged as one of the state- 17th International of-art research conference not only in A topic always crucial in the context of the field of digital libraries, but also in safety-critical systems is how to Conference on Theory the fields of computer and information achieve and prove trust in such sys- sciences. tems. One issue in these systems is pre- and Practice of Digital dictability, essentially in the time Under the general theme Sharing domain. Compiling for time pre- Libraries meaningful information, TPDL invites dictability is one approach to generate submissions for the proliferation of sci- code which has a predictable timing Valletta, Malta, 22-26 September 2013 entific and research osmosis in the cate- behaviour even in the case of complex gories Full Papers, Short Papers, processors. Within the T-CREST The International Conference on Theory Posters and Demonstrations. project, the University of Technology and Practice of Digital Libraries (TPDL) of Vienna and the University of constitutes a leading scientific forum on More information: Hertfordshire (and others) worked on digital libraries that brings together http://www.tpdl2013.info HW/SW architectures and code-gener- ation strategies to achieve time-pre- dictability, explaining the single-path code generation process in their paper. In the NOR-STA project, the University of Gdansk addressed devel- opment, maintenance and assessment of structured, evidence-based argu- ments to support trust assurance in CPS, using the TRUST-IT method- ology and presenting the adequate tool Call For Papers support in the NOR-STA platform of The International Conference on Network and Service Management (CNSM) is the premier annual conference of novel software services available on internet. results and experience reports in all aspects of management of networks, pervasive systems, enterprise and cloud computing environments. It provides an excellent venue for presenting and discussing the latest innovations and developments. CNSM 2013 will take place in Zrich, Switzerland. The conference is organized around a single track of paper presentations where fundamental research results are discussed. In the last session NTNU (Norwegian This core track is accompanied by series of workshops, tutorials and poster sessions where more focus discussions and progress reports are presented. CNSM invites academic, government, and industry researchers to submit papers to be University of Technology) presented a presented in the core track of the conference. All areas of Network and Service management are welcome. third application and implementation: Topics of interest include but are not limited to: an experience with a low-cost AAL General Management Issues monitoring system to enable elderly 0RQLWRULQJFRUUHODWLRQDQGGLDJQRVLV 3HUIRUPDQFHPDQDJHPHQW people a longer and safer stay at home. fault management and reconfiguration 5HVRXUFHDOORFDWLRQDQGRSWLPL]DWLRQ ,7VHUYLFHVEXVLQHVVLPSDFWDQDO\VLVDQG business process management The SAFECOMP proceedings are &RQILJXUDWLRQDQGDFFRXQWLQJ 6HFurity, trust and privacy published by SPRINGER in the LNCS (QHUJ\ management and optimization 4R6PDQDJHPHQW series no. 7612. Management Techniques and New Paradigms 6HOI-Management and adaptation (FRQRPLFPRGHOVIRUPDQDJHPHQW Link: http://www-e.uni- 3ROLF\-based and declarative models of management 9LUWXDOL]DWLRQ magdeburg.de/safecomp/about-sc- &RQWUROWKHRU\DQGRSWLPL]DWLRQ 2012/workshops/103-ercim-cps &RQWHQWFHQWULFQHWZRUNLQJ 6WRFKDVWLFPRGHOVDQGPDFKLQHOHDUQLQJLQPDQDJHPHQW Authors are invited to submit original contributions (written in not Please contact: published or submitted for publication elsewhere can be submitted. Papers s can be of two types: fulll or short papers. Submissions will be limited to 8 pages for full papers and 4 pages for short papers, in IEEE 2-column style. Papers Erwin Schoitsch exceeding these limits, multiple submissions, and self-plagiarized papers will be rejected without further review. chair of the ERCIM Working Group The workshops of CNSM 2013 invite researchers and practitioners to share and discuss their ongoing work, research and on Dependable Software-Intensive practical development experiences, and original results on specific new challenges and emerging trends in relation to the management of networks and services. Systems AIT Austrian Institute of Technology We welcome your contributions and look forward to your participation at CNSM 2013. / AARIT, Austria General Chair Burkhard Stiller, University of Zrich, Switzerland E-mail: [email protected] TPC Co-Chairs Gabi Dreo Rodosek, Universitt der Bundeswehr Mnchen, Germany Metin Feridun, IBM Research - Zurich, Switzerland Workshop Chair Rui L. Aguiar, Universidade de Aveiro, Portugal 58 ERCIM NEWS 92 January 2013

63 In Brief Latest ISTAG report identifies Marco Conti nominated Head of Software Technologies as the DIITET, a new department of CNR missing key enabling technology Following a restructuring and streamlining of the Italian The IST Advisory Group (ISTAG) - the advisory body to the National Research Council European Commission in the field of Information and (CNR), seven departments have Communication Technology (ICT or IST) stresses in its latest been created to represent the report that missing the strategic importance of software tech- main thematic areas of research. nology as a key enabling technology will lead to a significant The departments are responsible drawback for global competitiveness. The report mainly rec- for coordinating and promoting ommends: A Strategic Agenda for Software Technologies in the scientific and technology Europe should be created in cooperation with Industry, transfer activities of CNR. Academia and Public sector. The agenda should outline the DIITET will focus on the strength from a European perspective and how we can renew domains of manufacturing, ICT, energy and transportation and strengthen it. Common goals and needs in the short, and will supervise the activities of 22 institutes. The main medium and long term should be described. It should also objective of DIITET will be to promote interdisciplinary describe what actions should be implemented to achieve these research in key sectors of the European digital agenda and goals and needs and how we can, in a smarter way, make use Horizon 2020 with particular attention to the development of of existing efforts, resources and facilities. The agenda should those technologies that will ensure secure cities and commu- also draw up proposals on how the strategic work should be nities, smart energy and intelligent mobility and transporta- organized, run and quality assured. The first version of this tion systems. Areas of importance will include ICT, energy, Agenda should be delivered before mid-2013. advanced materials, sustainable mobility and advanced man- The full report is available at: http://kwz.me/SD ufacturing and processing. Dr Conti is currently Head of the Ubiquitous Internet Lab of the CNR Institute for Informatics and Telematics (IIT). He has published widely in the fields of The Web We Live In design modelling and performance evaluation of computer networks pervasive systems and social networks. He is The last-of-the-year issue of the journal Computer Networks, Editor-in-Chief of Elsevier Computer Communications and entitled The Web We Live In, has just been published. The Associate Editor-in-Chief of Elsevier Pervasive and Mobile issue reprints the first two papers published on Google Computing. He is the founder of successful conference and (written by Googles founders in 1998), and has additional workshop series such as ACM RealMAN, IEEE AOC, ACM papers on the evolution of the Web and Web Science, a paper MobiOpp, and IFIP/IEEE SustainIT. on economic aspects of the Web, comprehensive surveys on information privacy and social networks, a description of how the Web combines the virtual with the real, and what the Informatics Europe offers Web has done for education. A paper on security raises the question: Do we already have more security technology than Department Evaluation we are willing to live with? http://www.sciencedirect.com/science/journal/13891286/56/18 Informatics Europe through the Department Evaluation ini- tiative introduces a new service aimed at the assessment of research quality in the fields of Informatics, Computer Book Science and IT. The service is offered to Informatics Europe Michal Haindl, Ji Filip members and to all other Departments, Faculties, Schools and Research Institutes in Informatics, Computing, etc in Visual Texture Europe and beyond. It is characterized by an exclusive, peer- Accurate Material Appearance review driven evaluation process based on the standards and Measurement, Representation experiences of Informatics Europe. and Modeling This comprehensive book presents a survey of the state of On 20 November at the 2012 European Computer Science the art in multidimensional, physically-correct visual tex- Summit, held in Barcelona, the first Research Evaluation ture modeling. It is the first book to provide a detailed treat- Certificate was awarded to the Department of Informatics of ment of texture synthesis covering all known aspects of the the University of Zurich. The evaluation at the University of most advanced visual surface representation the Zurich was performed in July by a committee of highly Bidirectional Texture Function. From basic principles and respected experts in the field, chaired by Prof. John building upon the fundamentals to the latest advanced Mylopoulos. The evaluation report and results were already methods, the book brings together research from computer presented and the outcomes and advices thoroughly dis- vision, pattern recognition, computer graphics, virtual and cussed between the Department of Informatics of the augmented reality. This book is intended for researchers, University of Zurich and the Department Evaluation lecturers, students and practitioners. Committee of Informatics Europe. Springer, Series: Advances in Computer Vision and Pattern http://www.informatics-europe.org/services/department- Recognition, Hardcover 300 pages, ISBN 978-1-4471-4901-9 evaluation.html ERCIM NEWS 92 January 2013 59

64 ERCIM the European Research Consortium for Informatics and Mathematics is an organisa- tion dedicated to the advancement of European research and development, in information technology and applied mathematics. Its member institutions aim to foster collaborative work within the European research community and to increase co-operation with European industry. ERCIM is the European Host of the World Wide Web Consortium. Austrian Association for Research in IT I.S.I. - Industrial Systems Institute c/o sterreichische Computer Gesellschaft Patras Science Park building Wollzeile 1-3, A-1010 Wien, Austria Platani, PATRAS, Greece, 265 04 http://www.aarit.at/ http://www.isi.gr Consiglio Nazionale delle Ricerche, ISTI-CNR Portuguese ERCIM Grouping Area della Ricerca CNR di Pisa, c/o INESC Porto, Campus da FEUP, Via G. Moruzzi 1, 56124 Pisa, Italy Rua Dr. Roberto Frias, n 378, http://www.isti.cnr.it/ 4200-465 Porto, Portugal Czech Research Consortium Polish Research Consortium for Informatics and Mathematics for Informatics and Mathematics Wydzia Matematyki, Informatyki i Mechaniki, FI MU, Botanicka 68a, CZ-602 00 Brno, Czech Republic Uniwersytetu Warszawskiego, ul. Banacha 2, 02-097 Warszawa, Poland http://www.utia.cas.cz/CRCIM/home.html http://www.plercim.pl/ Centrum Wiskunde & Informatica Science and Technology Facilities Council, Science Park 123, Rutherford Appleton Laboratory NL-1098 XG Amsterdam, The Netherlands Harwell Science and Innovation Campus http://www.cwi.nl/ Chilton, Didcot, Oxfordshire OX11 0QX, United Kingdom http://www.scitech.ac.uk/ Fonds National de la Recherche Spanish Research Consortium for Informatics and Mathematics, 6, rue Antoine de Saint-Exupry, B.P. 1777 D3301, Facultad de Informtica, Universidad Politcnica de Madrid, L-1017 Luxembourg-Kirchberg Campus de Montegancedo s/n, http://www.fnr.lu/ 28660 Boadilla del Monte, Madrid, Spain, http://www.sparcim.es/ Swedish Institute of Computer Science FWO FNRS Box 1263, Egmontstraat 5 rue dEgmont 5 SE-164 29 Kista, Sweden B-1000 Brussels, Belgium B-1000 Brussels, Belgium http://www.sics.se/ http://www.fwo.be/ http://www.fnrs.be/ Foundation for Research and Technology Hellas Swiss Association for Research in Information Technology Institute of Computer Science SIRA Swiss c/o Professor Abraham Bernstein, Ph.D., Department of P.O. Box 1385, GR-71110 Heraklion, Crete, Greece Informatics Research Informatics, University of Zurich, Binzmhlestrasse 14, Association http://www.ics.forth.gr/ CH-8050 Zrich FORTH http://www.sarit.ch Fraunhofer ICT Group Magyar Tudomnyos Akadmia Anna-Louisa-Karsch-Str. 2 Szmtstechnikai s Automatizlsi Kutat Intzet 10178 Berlin, Germany P.O. Box 63, H-1518 Budapest, Hungary http://www.iuk.fraunhofer.de/ http://www.sztaki.hu/ Institut National de Recherche en Informatique University of Cyprus et en Automatique P.O. Box 20537 B.P. 105, F-78153 Le Chesnay, France 1678 Nicosia, Cyprus http://www.inria.fr/ http://www.cs.ucy.ac.cy/ Norwegian University of Science and Technology Technical Research Centre of Finland Faculty of Information Technology, Mathematics and PO Box 1000 Electrical Engineering, N 7491 Trondheim, Norway FIN-02044 VTT, Finland http://www.ntnu.no/ http://www.vtt.fi/ ! %! ! " ! " " ! ! $ " % # ! !" "#" ! & " " ! # ' %! # You can subscribe to ERCIM News and order back copies by filling out the form at the ERCIM Web site at http://ercim-news.ercim.eu/

Load More