An Agent-based Approach for Smart Energy Grids

Alba Amato, Beniamino Di Martino, Marco Scialdone and Salvatore Venticinque

Department of Industrial and Information Engineering, Second University of Naples, Via Roma 29, Aversa, Italy

Keywords:

Multi-agent Systems, Smart Grid, Energy Market.

Abstract:

The increasing demand for energy and the availability of several solutions of renewable energy sources has

stimulated the formulation of plans aiming at expanding and upgrading existing power grids in several coun-

tries. According to NIST, smart grid will be one of the greatest achievements of the 21st century. By linking in-

formation technologies with the electric power grid to provide electricity with a brain, the smart grid promises

many beneﬁts, including increased energy efﬁciency, reduced carbon emissions, and improved power reliabil-

ity. In this paper we present an agent based architecture for supporting collection and processing of information

about local energy production and storage resources of neighborhoods of individual houses and to schedule

the energy ﬂows using negotiation protocols.

1 INTRODUCTION

The increasing demand for energy and the availabil-

ity of several solutions of renewable energy sources

has stimulated the formulation of plans aiming at ex-

panding and upgrading existing power grids in sev-

eral countries. Contextually, the big improvement in

ICT (Information e Communications Technology) has

created a convergence of scientiﬁc and industrial in-

terests in the exploitation of these technologies for

implementing a process of structural transformation

at each stage of the energy cycle. This convergence

has given rise to modernized electrical grid that uses

ICT to gather and act on information, such as about

the behaviors of suppliers and consumers, in an auto-

mated way to improve the efﬁciency, reliability, eco-

nomics, and sustainability of the production and dis-

tribution of electricity (U.S. Ofﬁce of Electricity De-

livery & Energy Reliability, 2012). Smart grid will be

one of the greatest achievements of the 21st century.

By linking information technologies with the elec-

tric power grid to provide ”electricity with a brain”

the smart grid promises many beneﬁts, including in-

creased energy efﬁciency, reduced carbon emissions,

and improved power reliability (NIST, 2012). The

main aim here is to reduce the huge economic and

environmental costs deriving from the utilization of

the old paradigms. Due to ICT technologies, the elec-

tricity grid can change from a hierarchical, unidirec-

tional and centralized grid, to a distributed and net-

worked grid. The electricity grid gets power gen-

erated from consumers, who exploit renewable en-

ergy resources, and using ICT network as an orches-

trator to manage the complexity resulting from bidi-

rectional power ﬂows and less predictable demand.

This ambitious vision requires substantial advances in

intelligent decentralized control mechanisms that in-

crease economic efﬁciency, while keeping the phys-

ical properties of the network within tight permissi-

ble bounds(Miller et al., 2012). In fact this paradigm

shift, from hierarchical to distributed, brings with it a

number of difﬁculties due to the bidirectional and de-

centralized ﬂow of both energy and information and

to the heterogeneity and less predictable behaviour of

its components. In this context CoSSMiC, an ICT Eu-

ropean project, aims at fostering a higher rate of self-

consumption of decentralised renewable energy pro-

duction, using innovative autonomic systems for man-

agement and control of power micro-grids on users

behalf. A relevant challenge is the development of

selﬁsh agents that, driven by the user’s preferences,

pursue some goals to improve the utility of their own-

ers. The alignment of agents’ selﬁshness with user’s

interest will allow to win the user’s trust. On the

other hand, behavior patterns, which improve also the

performance of power grids, will provide the desired

smartness. In this paper we present an agent based

architecture for supporting agents to collect informa-

tion about local energy production and storage re-

sources of individual houses, belonging to the same

neighborhood, and to schedule the energy ﬂows us-

ing negotiation protocols. The paper is organized as

164

Amato A., Di Martino B., Scialdone M. and Venticinque S..

An Agent-based Approach for Smart Energy Grids.

DOI: 10.5220/0004820001640171

In Proceedings of the 6th International Conference on Agents and Artiﬁcial Intelligence (ICAART-2014), pages 164-171

ISBN: 978-989-758-016-1

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

follows: Section 2 reviews some related works; Sec-

tion 3 presents an overview of CoSSMiC project; in

Sections 4and 5.1 we describe the requirements of the

solution and a prototype of its implementation; con-

clusions are drawn in Section 6.

2 RELATED WORK

The scientiﬁc community investigates different pri-

orities in the ﬁeld of smart grids. Some examples

are market deregulation, ICT architecture, IT secu-

rity and data protection, energy efﬁciency, integration

of renewable energies, supply security/Grid bottle-

necks/Grid expansion, decentralised energy produc-

tion, smart meterology, storage devices and load ﬂex-

ibilisation. Much effort has been spent on the in-

vestigation in this ﬁeld of agents technology(Rogers

et al., 2012). In (Rodden et al., 2013) authors con-

sider how consumers might relate to future smart en-

ergy grids, and how exploiting software agents to

help users in engaging with complex energy infras-

tructures. In (Truong et al., 2013) authors deﬁne a

methodology for predicting the usage of home appli-

ances. An agent based prediction algorithm captures

the everyday habits by exploiting their periodic fea-

tures. In addition, the algorithm uses a episode gener-

ation hidden Markov model (EGH) to model the in-

terdependency between appliances. In (Isabel Praa

and Cordeiro, 2005) a Multi-Agent system architec-

ture simulates and analyses Competitive Electricity

Markets combining bilateral trading with power ex-

change mechanisms. Several heterogeneous and au-

tonomous intelligent agents representing the different

independent entities in Electricity Markets are used

and a detailed description of a promising algorithm

for Decision Support is presented and used to im-

prove agents bidding process and counter-proposals

deﬁnition. Agents are endowed with historical infor-

mation about the market including past strategies of

other players, and have strategic behaviour to face the

market. (Jia-hai et al., 2005) presents the architecture

and negotiation strategy of an agent-based negotiation

platform for power generating and power consuming

companies in contract electricity market. An intelli-

gent agent implements the negotiation process by se-

lecting a strategy using learning algorithms. Agent

uses fuzzy logic modiﬁcation of basic Genetic Algo-

rithm to accomplish strategy optimization and rein-

forced learning algorithm to modify the parameter of

negotiation tactics and strategy under different situa-

tions. Protocol Operation Semantics is used as agent

communication mechanism to handle sequential mes-

sage exchange. In (Telesca et al., 2007) an Open Ne-

gotiation Environment (ONE) provides sophisticated

negotiation processes and supports a model of col-

laboration and trust based on the idea of ”collabo-

rative multi-agent systems”, where agents can work

and learn with other trusted agents and develop col-

laborative learning schemes. ONE allows an organ-

isation to dynamically package and compose com-

plex services by negotiating alliances. Users can de-

ﬁne a custom negotiation process taking into account

several speciﬁcation such as negotiation rules, legal

rules, pricing policy, etc using a XML based script-

ing language. The runtime negotiation engine will be

in charge of executing the deﬁned process as a facil-

itator between parties that take into account the de-

ﬁned strategy and rules until the negotiation is suc-

cessfully closed. In (Capodieci et al., 2012b) and

(Capodieci et al., 2012a) an agent-based approach to

manage negotiation among the different parties is pre-

sented. The goal is to propose adaptive negotiation

strategies for energy trading in a deregulated market.

In particular, strategies derived from game theory are

used, in order to optimize energy production and sup-

ply costs by means of negotiation and adaptation. To

manage negotiation between agents the El Farol Bar

Game is used and the equilibrium model is proposed

in (Whitehead, 2008). Agents act on behalf of end

users, thus implying the necessity of being aware of

multiple aspects connected to the distribution of elec-

tricity. These aspects refer to outside world variables

like weather, stock market trends, location of the users

etc. A web service integration in which agents con-

tracting energy will automatically retrieve data to be

used in adaptive and collaborative aspects. More-

over, a MAS (Multi Agent System) is used to sim-

ulate a new paradigm for collaboration among var-

ious actors and it is used the adaptive collaboration

strategies to calculate the energy that the big compa-

nies must provide. In particular, this energy is cal-

culated by subtracting the amount of energy required

and the amount of energy that the producers can sup-

ply. In (Peleg and Rosenschein, 2012) multiagent re-

source allocation in a competitive peer-to-peer envi-

ronment is addressed making use of micro-payment

techniques, along with concepts from random graph

theory and game theory. It provides an analytical

characterization of protocol, and speciﬁes how an

agent should choose optimal values for the protocol

parameters. In (Brazier et al., 2012) authors claim

that agent and peer-to-peer based decentralized self-

management can change the future of energy markets

in which the power grid plays a core role. Both con-

sumers and providers of energy are autonomous sys-

tems, represented by software agents or peers capable

of self-management, virtual organizations of systems

AnAgent-basedApproachforSmartEnergyGrids

165

can emerge and adapt when necessary. Overlay struc-

tures (as deﬁned within p2p research) deﬁne adap-

tive communication structures, multi-agent research

provides interaction patterns. Our contribution, and

in particular the CoSSMic project, is going beyond

the state of art by supporting negotiation among ﬁnal

users on real power grid, that to the authors’ knowl-

edge has not been implemented before.The frame-

work that will be validated on real infrastructures by

trials that involve inhabitants of three different Eu-

ropean countries. Both software and hardware will

be integrated and customized ad hoc to be compli-

ant with existing installations. Finally we have ex-

perience in building network of agents both in smart

cities applications (Amato et al., 2012a) and for ne-

gotiation and brokering of computational resources in

Cloud markets(Venticinque et al., 2012; Amato et al.,

2012b).

3 THE CoSSMic SCENARIO

CoSSMic (Collaborating Smart Solar-powered

Micro-grids - FP7-SMARTCITIES-2013) is an ICT

European project that aims at fostering a higher

rate for self-consumption (< 50%) of decentralised

renewable energy production by innovative au-

tonomic systems for management and control of

power micro-grids on users’ behalf. This will allow

household and neighborhood power optimisation

and sales to the network. In addition CoSSMic will

provide a higher degree of predictability of power

deliveries for the large power companies, and it will

satisfy the requirements and achieve the beneﬁts

discussed above. CoSSMic research partners are

Stiftelsen Sintef International Solar Energy Research

Center Konstanz, Second University of Naples,

Norges Teknisk-Naturvitenskapelige Universitet,

Sunny Solartechnik, Boukje.com Consulting. City

of Konstanz in Germany and province of Caserta

in Italy are project partners that provide trial sites

for experimental activities and validation of results.

In CoSSMIc a micro-grid is typically conﬁned in a

smart home or an ofﬁce building, and embeds local

generation and storage of solar power, and a number

of power consuming devices. In addition electric

vehicles will connect and disconnect dynamically,

thus representing a dynamically varying storage

capacity. In Figure 1 an overview of the reference

scenario is shown. On the left side, micro-grids,

embedded with renewable energy production, storage

capacity and consumption, are combined with an

intelligent ICT platform. Such a framework will

allow for both peer-to-peer collaboration between

Figure 1: CoSSMIc reference scenario.

micro-grids in a neighborhood, forming a cluster.

All the cluster collaborates for the reduction of

variation of decentralised renewable energy transfer

to the grid and a higher rate for self-consumption.

At the same time the central power grid enables the

provision of energy for contingent requirements or

to complement the neighborhood resources. Agents

are used to manage the Home Area Network (HAN)

with the aim to optimize self-consumption rates

using renewable energy sources. Households with

renewable energy facilities can sell excess energy

to the neighborhood while households that needs

more energy can buy it. Ideally, the pool is at the

zero level when the energy production within the

household matches the consumption, and there is not

external exchange of energy. This agents pool also

participates to two different market places: for the

energy exchange within the neighborhood, and for

the exchange with the outside world. The design of

an ICT architecture is necessary to support sharing

of information, scheduling of exchanges between

power producers and storages in accordance with

policies deﬁned by owners, collection of data from

weather stations, weather forecasts, and habits and

plans of participants. Cloud services will be provided

to connect distributed installations, to allows for

power monitoring and updating policies by users.

Control agents can query the service to update their

knowledge and plan new schedules by negotiating

at eventually changed conditions. In the same way

agents can use information retrieved by weather

forecasting services to make projections about energy

needs or to make market forecasts. The behaviour of

the smart micro-grids will be governed business mod-

els based on credits. This should ensure rewards to

the users willing to share resources and collaborate to

optimise the overall working of the power grid. The

design and development of such a framework should

support the communication among agents over a

ICAART2014-InternationalConferenceonAgentsandArtificialIntelligence

166

peer-to-peer overlay to negotiate the scheduling of

power sources to energy storages. Agents will act

autonomously guided by rules and policies set by the

users and agree on a coordinated behaviour towards

the central power grid. The distributed agents net-

work will also implement a distributed infrastructure

for metering and monitoring the real production,

consumption and energy level of those smart grids

that accept to use Cloud services to provide relevant

information about their status. This information will

be exploited for collecting and studying experimental

activities conducted at trial sites and to validate

developed techniques and technologies. NoSQL data

stores will be used with this purpose because they

are designed to scale well horizontally and run on

commodity hardware. NoSQL data stores come up

with a number of key features (Cattell, 2012), such

as the ability to horizontally scale simple operation

throughput over many servers, the ability to replicate

and to distribute (partition) data over many servers, a

simple call level interface or protocol (in contrast to a

SQL binding), a weaker concurrency model than the

ACID transactions of most relational (SQL) database

systems, efﬁcient use of distributed indexes and

RAM for data storage, and the ability to dynamically

add new attributes to data records. At the trial

locations, which are rather different in terms of

population, sun, and available equipments, CoSSMic

will investigate how to motivate people to participate

in acquiring (more) renewable energy and the sharing

of renewable energy in the neighborhood, and test

methods for making money with these schemes. The

Second University of Naples will develop the agents

based software platform. The University of Oslo will

deﬁne some behavior patterns of micro-intelligent

networks, governed by selﬁsh agents that seek to

maximize the satisfaction of its users, participating in

a market by mechanisms that will improve the overall

performance of the electricity grid.

4 REQUIREMENTS ANALYSIS

AND DESIGN

The basic functionality of the smart grid focuses on

the integration of GENCO and Distributed Energy

Resources (DER) in the current system, where DER

refers to generators, accumulators and controllable

loads connected to the electrical distribution system.

4.1 High Level Requirements

The CoSSMic Framework is composed of devices,

ICT platforms and power Grid. COSSMic Frame-

work will support sharing of information, scheduling

of exchanges between power producers, consumers

and energy storages in accordance with policies de-

ﬁned by owners, collection of data from weather sta-

tions, weather forecasts, and habits and plans of par-

ticipants. At this level stakeholders include CoSSMic

Users, Devices, power grid operators (GenCO). The

COSSMic User will participate in two different sce-

narios, interacting by a GUI according to three high

level use cases (UCs). The Management UC allows

to conﬁgure and manage the available devices and to

manage and control at a higher level through rules and

policies the energy ﬂows. Monitoring provides facili-

ties to supervise and to potentially reconﬁgure the de-

vices and the energy ﬂows. Reporting and statistics

integrate information from several sources, including

power companies, weather reporting and forecasting

and to encourage the growth of the neighborhood net-

work. They facilitate the interaction with more and

other sources of information in the future. The COSS-

Mic Devices will use the Platform providing metering

and management services. COSSMic will exchange

power with GenCO when the MicroGrid cannot sat-

isfy its requirements in the case of over- or under-

production of energy within the CoSSMic neighbor-

hoods. In Figure 2 a component diagram of the CoSS-

Mic framework is shown. The CoSSMic platform

will run on embedded computer systems which will

be provided to ﬁnal users as a black box, to be plugged

into the power network and connected to Internet. The

Platform will be installed in every household and will

join a community of other instances within the neigh-

borhood. Instances of the platform communicate by

a P2P overlay and with the Cloud to eventually ex-

ploit advanced services. User’s information will be

bounded to the private network of each household and

will be forwarded outside only with the user’s agree-

ment and for debugging purpose. Each platform in-

stance will communicate with other households only

for the energy negotiation.

4.2 The CoSSMicICT Platform

As it is shown in Figure 3 COSSMic Platform is com-

posed of :

• a Graphical User Interface (GUI) that allows users

to interact with electronic devices through graph-

ical icons and visual indicators;

• a Knowledge Base (KB) that is an information

repository that provides a means for information

to be collected, organized, shared, searched and

utilized;

• a Multi Agent System (MAS) to allow for the de-

ployment of agents of consumers and producers

AnAgent-basedApproachforSmartEnergyGrids

167

Figure 2: The CoSSMic Framework.

that will participate in the energy distribution;

• a Market for the energy negotiation within the

neighborhood, and eventually with GenCO.

In particular agents will be the main actors of the

CoSSMic framework. They need to connect and dis-

connect to the ICT network as they control the con-

nection and disconnection of devices with the grid.

Moreover they will sell or buy energy as they are

capable of behave as power producer or consumer

according to the device they manage. In substance,

agents will be can be classiﬁed according to three cat-

egories:

Figure 3: The CoSSMic ICT Platform.

• Consumers: they buy energy for passive devices.

E.g. they will run in houses to manage objects

that absorb energy: electric car, computers, ovens,

washing machines, etc.

• Producers: they can sell energy. In this category

there are, for example, power generators, solar

panels, wind turbines.

Those devices, which are able both to produce and

consume energy will be deﬁned Prosumers. In this

category there will be also storages, which are rep-

resented by a couple of agents belonging to the two

different classes. In Figure 4 the MAS Use Case is

depicted. Agents implement use cases to manage de-

vices. At the same time they get information from

the KB for performing the required reasoning to set

Figure 4: The MAS Use Case.

up the best negotiation strategy. An important char-

acteristic of the agents would be the ability to handle

generators and active loads in the grid using proto-

cols and information ﬂows, coordinating to perform

certain functions in real time as, for example, to man-

age a peak,for dynamic load balancing, to meet a sud-

den drop in voltage by using energy from districts in

which there is a surplus. Programming models that al-

low to implement selﬁsh strategies for achieving the

user’s goals should be enabled. The allocation prob-

lem of mapping the consumption demand to the pro-

ducible or produced energy, can be modeled as a ne-

gotiation of micro power grids agents in a market-

place over a network overlay, where an energy bro-

ker, is responsible to handle local energy ﬂow and

to exchange energy with big companies. Market use

case enable agents to communicate with other agents

from different households. Communication and co-

ordination mechanisms are necessary to implement

negotiation protocols and strategies which lead the

system to an equilibrium that is the best compromise

between the global optimum and the user’s interest.

The Market will be implemented over a P2P overlay.

Agent and P2P concepts are closely related to selﬁsh-

ness, distributed system programming and P2P sys-

tems. Agents are able to improve functionality of a

P2P system and also P2P architecture can be an envi-

ICAART2014-InternationalConferenceonAgentsandArtificialIntelligence

168

Figure 5: Structure.

ronment in which abilities of agents are fully utilized.

Negotiation and brokering models among agents have

been widely investigated in literature in many ﬁeld

and the complexity of an automated negotiation de-

pends on several factors: the number of negotiated

issues, dependencies between these issues, number of

agents, representation of the utility, negotiation pro-

tocol, constraints, etc. Services are necessary to ex-

change any kind of information among heterogeneous

entities, such as agents, smart meters, user interfaces

and web services providers and requestors.

5 IMPLEMENTATION

In Figure 5 a conceptual model of the current proto-

type is shown. Consumer and Producer agents repre-

sent smart meters which notify their measures by an

Info Queue. Control agents of each category of de-

vice subscribe to the Info Queue to be notiﬁed about

changes in the household, on whose occurrence they

react and plan the optimal schedule inside the house-

hold. The residual energy to be sold or to be bought

are notiﬁed to the Market Queue. The Agent Nego-

tiator is updated about the quota to be negotiated with

the neighborhood or with the Genco, according to the

contingency level and availability on the P2P overlay.

We can observe that the message exchange in the ICT

platform can occur along the same patterns used by

the energy ﬂow in the power grid.

In Micro-grids the energy is exchanged within

household. The passive elements receive energy from

solar panels. Also electric cars, or smart-phones and

laptops are charged in the household, even if they are

mobile and they consume outside(Stein et al., 2012).

For this reason the smart-meters of all these devices

deliver their messages to the Info Queue, directly or

using Cloud services. The peer to peer network is

used both if the energy produced by solar panels is

not sufﬁcient, to allows the user for contacting neigh-

bors to have more energy, and if the produced energy

is greater than the required amount. In this case the

information retrieved in the Market Queue is used to

look for vendors or customers by the P2P overlay. An

offer is published in the P2P overlay if the required

resources are currently unavailable in the neighbor-

hood. Gencos: in case of unavailable energy offer

or requirement in the neighborhood the Gencos are

contacted. At the beginning of the project and in or-

der to set-up and test the technologies we are eval-

uating for building the ICT platform, we have im-

plemented a preliminary software demonstrator. For

communication between agents and devices, we used

a queues system in order to make their interaction

asynchronous. As shown in Figure 6, two queues

have been conﬁgured. Each device of the system

(consumer and producer) is associated with a differ-

ent agent. Consumer and producer agents communi-

cate the amount of energy required/produced by the

ﬁrst queue (Info Queue). Control agents use the Mar-

ket queue to agree about the internal energy schedule.

The residual energy to be bought or to be sold in the

neighborhood is ﬁnally scheduled and notiﬁed to the

negotiator by the same queue. It look for offers in the

P2P overlay and publish its offer to be contacted in

turn. Labels are used to route class of messages to

agents. The sequence diagram in Figure 7 shows a

particular scenario where there are one consumer and

one producer device. Messages with related amount

of consumed/produced energy are posted in the info

queue. Agents read data and report the results in a

market queue. Assuming the amount of energy to be

consumed is greater than that produced, the negotiator

does not ﬁnd any customers in the P2P and publishes

Figure 6: Prototype’s Structure.

AnAgent-basedApproachforSmartEnergyGrids

169

Figure 7: Sequence diagram.

its own offer.

5.1 Technological Choices

For the development of a prototype, the Jade agent

platform has been used (Bellifemine et al., 2008) to

develop and deploy agents. Jade is a software frame-

work that implements an efﬁcient agent platform in

compliance with the FIPA(FIPA, 2002) speciﬁcations.

As message queue Apache ActiveMQ(Snyder et al.,

2011) has been chosen. It is a popular and powerful

open source messaging and integration patterns. With

ActiveMQ it is possible to use pure queues or topics.

The substantial difference between queues and top-

ics is that for the latter only who is subscribed to that

particular topic receives the messages. Every smart

home appliance is built by an embedded device Rasp-

berry Pi running both JADE and ActiveMQ. More-

over it hosts a web based user interface. Raspberry

Pi is a single board RISC (Reduced Instruction Set

Computer) computer that allows to install operative

system based on Linux kernel. All energy requests

(both consumption and production) go through active

queues on the appliance where all agents are running

locally. Each producer and consumer element of the

house is associated with an agent that has the main

task of optimizing its use of energy and to avoid any

wastage; another agent’s task is ensuring that the con-

sumer has always electricity. For simulating a con-

sumer we developed some tests using a mobile de-

vice with Android operative system. The Android

software app extends the MQTT Client that is devel-

oped by Jason Sherman and based on the FuseSource

MQTT client library. This application uses MQTT

protocol that is a machine to machine connectivity

protocol. Through the latter it is possible to connect

to a server where there is ActiveMQ and send/receive

messages. In particular, as shown in ﬁgure 8, provid-

ing address, username and password to this app, it is

possible to connect to the Info Queue. For experimen-

tal purpose the battery level, position, local time are

Figure 8: App Screenshot.

read and sent interactively, but also asynchronously

on the occurrence of some events such as the battery

full, battery low, adapter connection or disconnection.

An agent receives all the messages and stores them in

a MySql database, by which we can draw graphics

dynamically. We made this to simulate the metering

and monitoring of smart power grids.

6 CONCLUSIONS

The paper presents software architecture for support-

ing agents to collect information about local energy

production and storage resources of neighborhoods of

individual houses and to schedule the energy ﬂows

using negotiation protocols. Besides the implemen-

tation of a preliminary prototype is described. Fu-

ture works will aim at replacing smartphone with

smart-meters and smart-plugs in a household inter-

connected which will part of a sensor network that re-

ceive commands and route information to an embed-

ding computing device hosting the described MAS.

Performance bottlenecks will be identiﬁed and off the

shelf software will be replaced by custom implemen-

ICAART2014-InternationalConferenceonAgentsandArtificialIntelligence

170

tations. Negotiation strategies and protocols will be

implemented. We also plan to use a NoSQL database

for metering and monitoring all smart power grids

participating in the trials of the project.

ACKNOWLEDGEMENTS

This work has been supported by CoSSMic (Col-

laborating Smart Solar-powered Micro-grids - FP7-

SMARTCITIES-2013).

REFERENCES

Amato, A., Di Martino, B., and Venticinque, S. (2012a).

Semantically augmented exploitation of pervasive en-

vironments by intelligent agents. pages 807–814.

Amato, A., Liccardo, L., Rak, M., and Venticinque, S.

(2012b). Sla negotiation and brokering for sky com-

puting. pages 611–620.

Bellifemine, F., Caire, G., Poggi, A., and Rimassa, G.

(2008). Jade: a software framework for developing

multi-agent applications. lessons learned. Information

and Software Technology Journal, 50:10–21.

Brazier, F., Ogston, E., and Warnier, M. (2012). The future

of energy markets and the challenge of decentralized

self-management. In Agents and Peer-to-Peer Com-

puting, volume 6573 of Lecture Notes in Computer

Science, pages 95–103. Springer Berlin Heidelberg.

Capodieci, N., Alsina, E. F., and Cabri, G. (2012a). A

context-aware agent-based approach for deregulated

energy market. In Reddy, S. and Drira, K., editors,

WETICE, pages 16–21. IEEE Computer Society.

Capodieci, N., Cabri, G., Pagani, G. A., and Aiello, M.

(2012b). Adaptive game-based agent negotiation in

deregulated energy markets. In Smari, W. W. and Fox,

G. C., editors, CTS, pages 300–307. IEEE.

Cattell, R. (2012). Scalable sql and nosql data stores. Tech-

nical report.

FIPA (2002). Fipa contract net interaction protocol. Tech-

nical report.

Isabel Praa, Carlos Ramos, Z. V. and Cordeiro, M. (2005).

Intelligent agents for negotiation and game-based de-

cision support in electricity markets. In Interna-

tional Journal of Engineering Intelligent Systems, vol-

ume 13, pages 147–154.

Jia-hai, Y., Shun-kun, Y., and Zhao-guang, H. (2005). A

multi-agent trading platform for electricity contract

market. In Power Engineering Conference, 2005.

IPEC 2005. The 7th International, pages 1024–1029

Vol. 2.

Miller, S., Ramchurn, S. D., and Rogers, A. (2012). Opti-

mal decentralised dispatch of embedded generation in

the smart grid. In AAMAS, pages 281–288.

NIST (2012). Nist smart grid. Technical report.

Peleg, Y. and Rosenschein, J. (2012). Agents and peer-to-

peer computing: Towards p2p-based resource alloca-

tion in competitive environments. In Agents and Peer-

to-Peer Computing, volume 6573 of Lecture Notes in

Computer Science, pages 129–140. Springer Berlin

Heidelberg.

Rodden, T. A., Fischer, J. E., Pantidi, N., Bachour, K., and

Moran, S. (2013). At home with agents: Exploring

attitudes towards future smart energy infrastructures.

In Proceedings of the SIGCHI Conference on Human

Factors in Computing Systems, CHI ’13, pages 1173–

1182, New York, NY, USA. ACM.

Rogers, A., Ramchurn, S. D., and Jennings, N. R. (2012).

Delivering the smart grid: Challenges for autonomous

agents and multi-agent systems research. In AAAI.

Snyder, B., Bosanac, D., and Davies, R. (2011). ActiveMQ

in Action. Manning Publications Co., Greenwich, CT,

USA.

Stein, S., Gerding, E., Robu, V., and Jennings, N. R.

(2012). A model-based online mechanism with pre-

commitment and its application to electric vehicle

charging. In Proceedings of the 11th International

Conference on Autonomous Agents and Multiagent

Systems - Volume 2, AAMAS ’12, pages 669–676.

Telesca, L., Finnegan, J., Ferronato, P., Malone, P., Ricci,

F., and Stanoevska-Slabeva, K. (2007). Open negoti-

ation environment: An open source self-learning de-

centralised negotiation framework for digital ecosys-

tems. Proceedings of IEEE International Conference

on Digital Ecosystems and Technologies.

Truong, N. C., Tran-Thanh, L., Costanza, E., and Ram-

churn, D. S. (2013). Activity prediction for agent-

based home energy management. In Agent Technolo-

gies for Energy Systems (ATES 2013).

U.S. Ofﬁce of Electricity Delivery & Energy Reliability

(2012). Smart grid / department of energy. Techni-

cal report.

Venticinque, S., Tasquier, L., and Di Martino, B. (2012).

Agents based cloud computing interface for resource

provisioning and management. pages 249–256.

Whitehead, D. (2008). The el farol bar problem revisited:

Reinforcement learning in a potential game. Ese dis-

cussion papers, Edinburgh School of Economics, Uni-

versity of Edinburgh.

AnAgent-basedApproachforSmartEnergyGrids

171