The Management System of INTEGRIS
Extending the Smart Grid to the Web of Energy
Joan Navarro
1
, Andreu Sancho-Asensio
1
, Agust´ın Zaballos
1
, Virginia Jim´enez-Ruano
1
,
David Vernet
1
and Jos´e Enrique Armend´ariz-I˜nigo
2
1
La Salle – Ramon Llull University, 08022 Barcelona, Spain
2
Universidad P´ublica de Navarra, 31006 Pamplona, Spain
Keywords:
Web-based Services, Intelligent Agents, Distributed Databases, Security, Smart Grids.
Abstract:
The recent growth experimented by the Internet has fostered the interaction of many heterogeneous technolo-
gies under a common environment (i.e., the Internet of Things). Smart Grids entail a sound example of such
situation where several devices from different vendors, running different protocols and policies, are integrated
in order to reach a common goal: bring together energy delivery and smart services. Latest advances on
this domain have led to effective architectures that support this idea from a technical perspective, but fail at
providing powerful tools to assist this new business model. Hence, the purpose of this paper is to present a
novel unified and ubiquitous management interface, driven by an intelligent system, that uses the advantages
featured by the Web of Things to manage the Smart Grid. Therefore, this work opens a new path between the
Internet of Things and the Web of Things resulting in a new concept coined as the Web of Energy.
1 INTRODUCTION
The ever growing set of features provided by the In-
ternet has led into a standard communications frame-
work that eases the deployment and development
of distributed and heterogeneous applications. Such
evolution has driven a rising interest in connecting
several gadgets to the Internet, ranging from mobile
phones to home appliances, including special sys-
tems such as traffic lights or embedded devices. This
has led to a new form of distributed system, referred
to as the Internet of Things (IoT), that consists in
(1) uniquely identifying every object in the network,
(2) using the Internet as the communications infras-
tructure, and (3) providing a lightweight interface to
rapidly access everywhere.
Recent advances on Smart Grids have explored
the feasibility of considering the electricity distribu-
tion network as a particular case of the IoT (Guinard
et al., 2011; Zaballos et al., 2011). Certainly, this spe-
cific domain poses appealing challenges in terms of
integration, since several distinct smart devices (also
referred to as Intelligent Electronic Devices or IEDs)
from different vendors—often using proprietary pro-
tocols and running at different layers—must inter-
act to effectively deliver energy and provide a set
of enhanced services and features (also referred to
as smart functions) to both consumers and producers
(prosumers) such as network self-healing, real-time
consumption monitoring, and asset management (IN-
TEGRIS, 2011). Although the latest advances on the
IoT field have definitely contributed to the physical
connection of such an overwhelming number of smart
devices, several issues have arisen when attempting to
provide a common management and monitoring in-
terface for the whole Smart Grid (INTEGRIS, 2011;
Aman et al., 2013).
Indeed, integrating the heterogeneous data gener-
ated by every device on the Smart Grid (e.g., wired
and wireless sensors, smart meters, distributed gen-
erators, dispersed loads, synchrophasors, windmills,
solar panels, communication network devices) into a
single interface has emerged as a hot research topic.
So far, some experimental proposals (Guinard et al.,
2011) have been presented to face this issue by using
the Web of Things (WoT) concept to access a mashup
of smart devices and directly retrieve their infor-
mation using reasonably thin protocols (e.g., HTTP,
SOAP) (Guinard et al., 2010).
However, the specific application of these ap-
proaches to real-world environments is fairly doubt-
ful because (1) they may open new security breaches
(Zeng et al., 2011; Bou-Harb et al., 2013) (i.e., end-
users could gain access to critical equipment), (2)
329
Navarro J., Sancho-Asensio A., Zaballos A., Jiménez-Ruano V., Vernet D. and Armendáriz-Iñigo J..
The Management System of INTEGRIS - Extending the Smart Grid to the Web of Energy.
DOI: 10.5220/0004950203290336
In Proceedings of the 4th International Conference on Cloud Computing and Services Science (CLOSER-2014), pages 329-336
ISBN: 978-989-758-019-2
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
there are no mature electric devices implementing
WoT-compliant standards available in the market (IN-
TEGRIS, 2011), and (3) industry is averse to include
foreign modules (i.e., web servers) on their histor-
ically tested and established—but poorly evolved—
proprietary systems (Gungor et al., 2013).
Therefore, the European project INTelligent Elec-
trical GRId Sensor Communications (INTEGRIS)
(INTEGRIS, 2011) has explored a new way to over-
come these issues and, thus, provide a manage-
ment interface for the Smart Grid inspired by the
WoT. More specifically, the aim of INTEGRIS is
twofold: on the one hand, it implements an ICT
infrastructure—based on the IoT paradigm—to han-
dle the Smart Grid security, storage, and commu-
nications requirements (Navarro et al., 2012). On
the other hand, it uses a cognitive-inspired intelli-
gent multi-agent system to manage the whole Smart
Grid and link it with end-users using a WoT-based ap-
proach, which results in a new bridge between the IoT
and WoT.
The purpose of this paper is to present this multi-
agent intelligent system that extends the WoT ap-
proach and implements the management system used
in the INTEGRIS’ real-world domain. This proposal,
which leads in a new form of the WoT coined as the
Web of Energy (WoE), is targeted to provide a secure,
context-aware, and uniform web-based novelenviron-
ment to effectively manage, monitor, and configure
the whole Smart Grid. Moreover, conducted devel-
opments proof the feasibility and reliability of our ap-
proach and encourage practitioners to further research
towards this direction.
The remainder of this paper is organized as fol-
lows. Section 2 reviews the related work and justifies
the proposal. Then, Section 3 details the reference ar-
chitecture for the Smart Grid and Section 4 introduces
how it is integrated into the Web of Things. Finally,
Section 5 concludes the paper.
2 RELATED WORK
Service composition, heterogeneous devices interac-
tion, and a close contact with the real-life demands
are some of the features that have positioned Smart
Grids as an appealing landscape for deploying the
latest advances concerning the IoT and WoT. Over
the last years, practitioners have directed their efforts
on enabling communications between distinct types
of devices spread across the different network facil-
ities that compose the Smart Grid (Zaballos et al.,
2011). Certainly, the IoT and the WoT approaches
have promoted such advances in the sense that IEDs
are no longer considered as isolated entities but re-
ferred to as gears of a complex and heavily cou-
pled distributed system (Guinard et al., 2011; Guinard
et al., 2010). While the IoT has provided a mature
approach to enable hardware communication on the
Smart Grid (Zaballos et al., 2011; INTEGRIS, 2011),
management, monitoring, and grid configuration re-
quirements envisage the need of using either central-
ized Network Management Systems (NMSs) (Gungor
et al., 2013) or WoT-based approaches (Guinard et al.,
2011; Priyantha et al., 2008).
Although centralized NMSs are rapid to deploy
and easy to maintain, their lack of scalability, sin-
gle point of failure exposure, and bottleneck effect
vulnerability make them unfeasible for real-world ap-
plications (Aman et al., 2013). These issues are
driving system designers to explore distributed solu-
tions for the Smart Grid domain such as the Repre-
sentational State Transfer (REST) and Web Services
(Guinard et al., 2010). On the one hand, REST-
ful strategies (Cubo et al., 2012; Luckenbach et al.,
2005) are targeted to ease the development of scal-
able services and applications by (1) uniformly iden-
tifying every smart object through a unique Uniform
Resource Identifier (URI), (2) using HTTP stateless
communications, (3) representing resources through
standard human readable formats (e.g., XML), and (4)
using simple parsing algorithms—typically embed-
ded on tiny web servers—to interact with every IED
(Guinard et al., 2011; Duquennoy et al., 2009). On the
other hand, strategies based on Web Services pursue
the same goal but implement some advanced facilities
(e.g., UDDI server, WSDL, security) rather than using
light weight protocols (Guinard et al., 2010; Priyantha
et al., 2008).
Despite the aforementioned benefits featured by
Web Services and RESTful frameworks, the estab-
lished electric industry is still far from widely adopt-
ing them as a standard solution for the Smart Grid
domain (INTEGRIS, 2011). In addition, both solu-
tions may incur in some critical security issues that
are not affordable in this domain (Zeng et al., 2011;
Bou-Harb et al., 2013). Therefore, in the INTEGRIS
project we have split the Smart Grid into two isolated
layers: a lower layer that encompasses IEDs’ com-
munications following the IoT schema—described in
(Navarro et al., 2012)—, and an upper layer—herein
presented—that covers prosumers interactions with
the grid following the WoT approach. To success-
fully bridge both layers and meet the stringent se-
curity and scalability constraints demanded by the
Smart Grid (Aman et al., 2013; Navarro et al., 2012),
we have implemented a multi-agent intelligent system
that (1) is aware of all heterogeneities of the lower
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
330
layer (Zaballos et al., 2011), (2) predicts future situa-
tions (Gama, 2010), (3) builds a comprehensive sys-
tem model (Navarro et al., 2012), and (4) delivers it to
a uniform interface for user interaction using a REST-
ful approach. The following section reviews the key
parameters to be extracted from the lower layer and
presents the internals of this cognitive-inspired intel-
ligent subsystem.
3 A REFERENCE MODEL FOR
SMART GRIDS
Actually, splitting the ICT system that holds the am-
bitious requirements of Smart Grids (Aman et al.,
2013) into the two aforesaid layers enables (1) provid-
ing an extra security barrier by hiding electric-devices
and grid topology from unauthorized end-users, (2)
building scalable mechanisms to collect data from
IEDs, (3) addressing the electric functions (e.g., volt-
age monitoring)separately from user-drivenfunctions
(e.g., power flow management) and communication
issues (i.e., overlay networks (Zaballos et al., 2011)),
(4) integrating heterogeneous protocols and technolo-
gies, and (5) providing a uniform management inter-
face (INTEGRIS, 2011). The purpose of this section
is to present the architecture that supports this idea
and announce the key parameters used to link both
layers.
From our real-world experiences collected dur-
ing the INTEGRIS project, we have found that di-
viding the Smart Grid into these logical layers poses
some critical difficulties arisen from the fact that typi-
cally, IEDs are closed devices that do not allow im-
plementing custom developments (e.g., security or
information-exchange protocols)—as novel experi-
mental devices do (Guinard et al., 2011). Therefore,
we proposed a new device coined as I-Dev (Navarro
et al., 2012) that behaves as a frontier between these
two layers and implements (1) a communications sub-
system that allows heterogeneous network coexis-
tence, (2) a security subsystem that provides a reli-
able and secure low layer communications infrastruc-
ture, (3) a distributed storage subsystem that smartly
stores all data generated by IEDs, and (4) an intelli-
gent cognitive-inspired subsystem that is aware of all
events arisen from any subsystem of the network.
In order to avoid the bottleneck effect and thus
deploying a scalable ICT infrastructure, two distinct
roles have been assigned to I-Devs: the Perception
Action Agent (PAA) and the Domain Management
Agent (DMA). Therefore, the whole Smart Grid is
spread into several regions (e.g., low voltage substa-
tions) referred to as I-Domains (Navarro et al., 2012),
where in each I-Domain a DMA and several PAAs
interact with deployed IEDs using the IoT approach
(Zaballos et al., 2011).
As depicted in Fig. 1, PAAs—placed in the bot-
tom layer—(1) gather information from their associ-
ated IEDs, (2) report their findings to their assigned
DMA, and (3) execute the commands ordered by it as
a result of a collaborative decision between all DMAs
in the Smart Grid. Hence, I-Devs behaving as DMAs
bridge the bottom layer and the top layer while en-
suring scalability and a safe isolation between both
sides. Likewise, the upper layer is committed to in-
terface with the management side of all the smart
functions provided by the Smart Grid (e.g., supervi-
sory control and data acquisition (SCADA), advanced
distributed automation (ADA), distributed energy re-
source (DER), automatic meter reading (AMR), ad-
vanced metering infrastructure (AMI), NMS, or qual-
ity of service) taking benefit from the ubiquitous fea-
tures provided by the WoT approach.
As the lower layer has been broadly addressed in
(Navarro et al., 2012; Zaballos et al., 2011), the re-
mainder of this section is devoted to (1) summarize
the functionalities of every subsystem running in any
I-Dev and (2) specify the key monitoring and config-
uration parameters to be delivered to the herein pro-
posed upper layer.
3.1 Distributed Storage Subsystem
Smart Grids intrinsically generate vast amounts of
heterogeneous information (i.e., every IED may have
its own proprietary data format) that need to be ef-
fectively stored in order to be later processed by
the aforementioned smart functions. In this context,
data are generated at different points of the grid and
need to be reliably replicated in order to afford site
failures and boost their availability through a dis-
tributed storage system. Typically, classical relational
databases are unable to handle neither the dynamic
nature, nor the scalability requirements, nor the strin-
gent computing and storage capabilities of smart de-
vices (Aman et al., 2013). Hence, we developed our
own distributed storage architecture running a custom
replication protocol that smartly placed data to opti-
mize the performance of smart functions being exe-
cuted at the I-Devs layer. This replication protocol is
aimed to provide a consistent view of any datum
stale but consistent versions are useful for non critical
functions such as average consumption monitoring—
at any situation. To achieve this commitment data
is replicated and partitioned following a hierarchi-
cal hybrid approach between state-machine and syn-
chronous primary copy replication that attempts to
TheManagementSystemofINTEGRIS-ExtendingtheSmartGridtotheWebofEnergy
331
Figure 1: Deployed reference architecture for Smart Grids.
keep system scalability by epidemically updating data
across I-Devs (Navarro et al., 2012). This strategy
converts the storage infrastructure composed by I-
Devs as a dynamic set of nested onion layers that store
eventually consistent versions of data.
Manually configuring and monitoring this dis-
tributed system is unfeasible because (1) there is an
overwhelming number of devices involved on the
replication process, (2) there are several interdepen-
dencies with other subsystems, and (3) the system
conditions may change abruptly. Therefore, we pro-
pose to deliver the following key performance metrics
and configuration parameters to an intelligent system
who is aware of the status of the whole Smart Grid, in
order to get a reliable set of actions to be performed on
the storage domain. Specifically, we need to track the
amount of read/write operations, response time, repli-
cation depth, required consistency degree, and repli-
cation hierarchy layout (Navarro et al., 2012). The
following subsection describes the communications
and security subsystem and details which metrics are
delivered to the intelligent system.
3.2 Communications and Security
Subsystem
ICTs, trust management, and technological integra-
tion play an essential role when coordinating all IEDs
to solve every smart function. However, they have to
be carefully addressed since the electrical distribution
infrastructure encompasses aerial and underground
areas that current communications technologies are
unable to reach (Zaballos et al., 2011). Current ap-
proaches in this domain rely on the Internet network
to boost their performance but inherit the same threats
and critical risks in terms of cyber-security(Bou-Harb
et al., 2013). Therefore, communications and security
in Smart Grids are crucial for the survival and fea-
sibility of the global electricity distribution concept
(Navarro et al., 2012). So far, very few standards con-
cerning security (e.g., IEC62351, NISTIR7268) have
been proposed to address the specific issues posed by
Smart Grids, which still leaves the network and its
links open to cyber-security attacks that may produce
Denial of Service (DoS) and eavesdropping of critical
network management messages. Cyber attacks in the
communications network are aimed to bring the max-
imum damage, exploit the greatest benefits and take
advantage of the network structure or from protocols
vulnerabilities for malicious purposes (e.g., extended
power outages and destruction of power equipment
(Bou-Harb et al., 2013)). Latest advances on network
resilience protocols (Chen et al., 2012) implemented
on I-Devs permit them to sustain from temporal node
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
332
failures or disconnections as long as countermeasures
can be activated.
However, these countermeasures may prevent
other subsystems to behave properly (e.g., the dis-
tributed storage system may be unable to collect data
from a given PAA) and thus drive the whole Smart
Grid to a panic situation. Therefore, our proposal
sends the following key communications and security
metrics of every I-Dev to an intelligent system that
owns a global view of the Smart Grid: connectivity—
the number of links to the Smart Grid control center—
, security performance—number of packet errors,
number of encryption errors, number of unsuccess-
ful access attempts, number of I-Dev reboots, num-
ber of network overload situations and other DoS at-
tacks, and number of packet retransmissions—, qual-
ity of service performance—amountof flow rates, dis-
carding probabilities and queue lengths—, and de-
lay performance—average link delays. Then, the in-
telligent subsystem examines these parameters and
selects the best security and communications action
(e.g., isolating an I-Dev) in order to achieve the best
Smart Grid overall performance.
3.3 Intelligent Subsystem
Although designing and modeling every subsystem of
the Smart Grid separately eases the development pro-
cess, we have observed that it is unfeasible to allow
them running on their own (i.e., without considering
the whole grid status) (Navarro et al., 2012). Con-
sidering the vast amount of data and distinct events
continuously arisen from the grid, the management of
such a system cannot fully rely on an expert. There-
fore, we have chosen an intelligent multi-agent sys-
tem to interprete the aforesaid key parameters of ev-
ery subsystem in the grid in order to build a com-
prehensive model of the whole system. Specifically,
this context requires an online learning scheme able
to (1) put together the parameters of each subsystem,
(2) rapidly adapt to the constant changes of the Smart
Grid, (3) provide an accurate estimation concerning
the whole grid status, and (4) come up with the best
configuration at every subsystem to reach the optimal
overall performance. To this regard, we have selected
an enhanced version of the eXtended Classifier Sys-
tem (XCS) (INTEGRIS, 2011) that diligently meets
these requirements as shown in what follows.
In fact, XCS, a cognitive-inspired algorithm, is
targeted to evolve a population [P] of classifiers. At
the end of the learning process, the population is ex-
pected to acquire a high quality model using these
classifiers. Each one consists of a production rule—
composed by an antecedent and a consequent—and
Figure 2: Intelligent subsystem learning architecture.
DMAs use the Web of Energy to share their experiences
collected at every I-Domain and build a global Smart Grid
knowledge model.
a set of parameters that evaluate the quality of the
rule. The antecedent part of the rule contains the input
variables collected by PAAs from every subsystem
and related to the environment (i.e., what the algo-
rithm “senses”). Likewise, the consequent part con-
tains the action predicted for the future status of the
Smart Grid. The main parameters that evaluate the
quality of every rule are (1) an estimation of the re-
ward that will be received if the predicted status is
triggered, (2) the expected error, (3) the experience of
the classifier, and (4) the fitness of the classifier.
Thoroughly, XCS follows the online learning pro-
cess depicted in Fig. 2 at the DMA: it starts with an
empty population and it learns by sampling new train-
ing examples. This learning process creates the match
set [M] containing all the classifiers that match with
the current example (INTEGRIS, 2011). If [M] does
not contain a minimum user-defined threshold of dif-
ferent actions, the system generates new ones arbitrar-
ily. Then, the action to be proposed to the actuators
(i.e., PAAs) is selected via a fitness-weighted average
of all matching classifiers in [M], forming the action
set [A]. All classifiers in [A] predict the same action
and hence they share the evaluation payoff in a nich-
ing scheme. At the end of the iteration, a genetic al-
gorithm is applied at [A] to discover new promising
TheManagementSystemofINTEGRIS-ExtendingtheSmartGridtotheWebofEnergy
333
rules, for instance if data response time [10, 100]
ms and ... and reconnections [40, 60] then block
I-Dev3.
As the XCS is a reinforcement learning approach,
training and scalability need to be properly consid-
ered: First, as it is unfeasible to manually label all
examples of the training set, we have used an on-
line clustering approach based on k-Nearest Neigh-
bors (Navarro et al., 2012) to label them and thus let
the system autonomously learn from the environment.
Last, in order to meet the intrinsic scalability require-
ments of the Smart Grid, a unique centralized thinking
unit is avoided by deploying a DMA and its associated
PAAs on every I-Domian.
However, the distributed nature of this cognitive
subsystem entails further concerns on how integrat-
ing all the knowledge to an accurate set of human-
readable rules, which enables to think globally and
act locally. The following section is devoted to detail
how these rules are delivered across the upper layer of
the herein presented WoT approach.
4 FROM THE IoT TO THE WoE
In fact, distributing the knowledge building layer of
the cognitivesubsystem (i.e., DMAs) to meet the scal-
ability constraints posed by the Smart Grid defini-
tively hampers the learning process. As shown in
Fig. 2, refusing a centralized architecture forces lo-
cally collected rules at every I-Domain to be shared
among all DMAs of the Smart Grid—the underlying
hypothesis in rule sharing is that similar structures
(i.e., I-Domains) require similar configurations (i.e.,
knowledge model)—, which may lead to some con-
flicting situations arisen from the fact that the cogni-
tive subsystem is unable to reach 100% of accuracy
(Navarro et al., 2012). Note that rules shared be-
tween I-Domains are those with a classification accu-
racy greater than a user-defined minimum threshold.
Although the dynamics of this on-line learning archi-
tecture allow removing these conflicting rules without
shutting down or reseting the affected DMA, a reli-
able decision process to conduct this action is manda-
tory.
Therefore, as depicted in Fig. 2, we have intro-
duced a new role on the Smart Grid: the expert sys-
tem; which is the combination of the aforesaid soft-
ware entities and a specialist that owns enough expe-
rience related to the electricity domain. To this regard,
this specialist is continuously analyzing the results
provided by the intelligent system and thus, learning
from the Smart Grid in order to supplement software
suggestions by deciding which rules must be deleted
from every DMA (also referred to as conflict resolu-
tion). Likewise, this expert can introduce new knowl-
edge based on his expertise to the cognitive subsystem
by forcing the usage of new rules. We have found (IN-
TEGRIS, 2011; Navarro et al., 2012) that this expert
system greatly enhances the performance of the intel-
ligent system because it ensures that critical decisions
are taken consistently.
However, the feasibility of this approach relies
on a uniform interface that fits with the intrinsic
distributed nature of the Smart Grid and enables
its effective management. So far, existing prelimi-
nary solutions (INTEGRIS, 2011) achieve such com-
mitment by using centralized management systems
(e.g., SCADA) that are unable to fully integrate the
long-term requirements and applications posed by the
Smart Grid domain (Gungor et al., 2013). Hence, we
have deployed a WoT-inspired infrastructure—coined
as WoE—on top of the Smart Grid that links all I-
Domains and permits a bidirectional communication
between electricity domain (i.e., bottom layer in Fig.
1) and the application domain (i.e., top layer in Fig.
1).
To this concern, every I-Dev in the Smart Grid has
been labelled with an URI, which enables specialists
and all entities residing at the top layer of Fig. 1 using
asynchronous communications to (1) incorporate new
knowledge or erase existing rules that enter in conflict
with previously discovered ones, (2) reset and manip-
ulate the configuration parameters that control every
subsystem—especially the cognitive one—, and (3)
obtain accurate statistics from every DMA, and thus
from the whole Smart Grid as depicted in Fig. 3.
More specifically, Fig. 3 depicts some of the web
interfaces that integrate the heterogeneous framework
found on the bottom layer of the Smart Grid—and
linked using a IoT approach—to a uniform environ-
ment. The top left screenshot in Fig. 3 shows the
configuration interface of the distributed subsystem
(i.e., replication depth, cache size, number of layers
(Navarro et al., 2012)) running at I-Domain 6. The
top right screenshot in Fig. 3 shows the configuration
of the communications and security subsystem run-
ning at I-Domain 3. Finally, the bottom screenshots
in Fig. 3 are devoted to depict the two layers of the
cognitive system.
The bottom left screenshot in Fig. 3 depicts the
management interface of the intelligent subsystem
running at I-Domain 9; first, it shows a list of three
PAAs, below it depicts the predicted I-Domain status
and the selected action in [P]. Additionally, the bot-
tom right screenshot in Fig. 3 depicts the collected
data at PAA 2 from I-Domain 7. In this way, special-
ists can monitor the (1) effects of applying a given ac-
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
334
Figure 3: The Web of Energy. A uniform interface to support smart functions and manage the associated distributed, commu-
nications, security, and intelligent subsystems.
tion, (2) set of locally and globally collected rules, (3)
parameters perceived at every PAA, and (4) training
accuracy at a glance.
Thoroughly, the herein presented approach pro-
vides a uniform web-based interface referred to as
WoE that safely isolates the electricity grid domain
from its applications. More specifically, this sys-
tem includes an ad-hoc distributed storage architec-
ture to support the massive amount of data gener-
ated by the grid, that delivers these data through a se-
cured communication network to a multi-agent—able
to face the intrinsic dynamic nature of Smart Grids—
intelligent subsystem that processes the events and
changes arisen in this domain.
5 CONCLUSION
This paper presents a particular application of the
WoT to the concrete scenario of power networks.
We have conducted our experimentation over the sys-
tem presented in (Navarro et al., 2012), where an
IoT-based infrastructure enabled machine-to-machine
interactions between small and resource-constrained
devices on the Smart Grid domain. Thus, we have
extended the IoT concept by providing a bidirec-
tional human-to-machine interface—inspired by the
WoT—that results in a ubiquitous energy control and
management system coined as Web of Energy. This
proposal combines the web-based visualization and
tracking tools with the Internet protocols, which en-
ables a uniform access to all devices of the Smart
Grid.
In order to provide such an effective and reliable
management interface aimed to address the hetero-
geneous nature of devices residing on the grid, we
have deployed an intelligent subsystem devoted to (1)
learn from the real-world events, (2) predict future sit-
uations, and (3) assist on the decision making pro-
cess. This intelligence layer is composed by means of
a multi agent system to meet the scalability require-
ments of the Smart Grid. Moreover,it builds a knowl-
edge model in terms of production rules, implements
its own apportionment of credit mechanism (i.e., uses
reinforcement learning), and has an ad-hoc rule dis-
covery technique based on a genetic algorithm. How-
ever, we have seen that the intelligent system is un-
able to cope with the complexities of the Smart Grid
that hamper the optimal learning performance (i.e.,
100% accuracy). Therefore, we have successfully
TheManagementSystemofINTEGRIS-ExtendingtheSmartGridtotheWebofEnergy
335
included the role of the specialist in the presented
WoE approach. This entity is targeted to dynami-
cally rectify the intelligent subsystem outcomes and
improve the global grid performance. Note that the
aforesaid WoE framework eases the specialists com-
mitment considerably, in the sense that they are able
to interact with the different modules that control the
Smart Grid through a pervasive and user friendly web-
based interface rather than traditional roughly com-
mand lines.
Finally, we have demonstrated the feasibility of
our proposal by running it on the real-world sce-
nario defined by the INTEGRIS project (INTEGRIS,
2011). Our collected experiences show that this uni-
form management interface—depicted in Fig. 3—
plays a key role in the process of development and
standardization of current and new smart functions.
Certainly, this paper encourages practitioners to con-
duct future work in this direction by defining new
electric applications following this layered scheme.
ACKNOWLEDGEMENTS
The research leading to these results has received
funding from the European Union European Atomic
Energy Community Seventh Framework Programme
(FP7/2007-2013 FP7/2007-2011) under grant agree-
ment 247938 for Joan Navarro, Agust´ın Zaballos by
Generalitat de Catalunya for its support under grant
2013FI
B2 00089 for Andreu Sancho-Asensio, and
by the Spanish National Science Foundation (MEC)
(grant TIN2012-37719-C03-03) for Jos´e Enrique Ar-
mend´ariz-I˜nigo.
REFERENCES
Aman, S., Simmhan, Y., and Prasanna, V. K. (2013). Energy
management systems: State of the art and emerging
trends. IEEE Communications Magazine, 51(1):114–
119.
Bou-Harb, E., Fachkha, C., Pourzandi, M., Debbabi, M.,
and Assi, C. (2013). Communication security for
Smart Grid distribution networks. IEEE Communi-
cations Magazine, 51(1):42–49.
Chen, P.-Y., Cheng, S.-M., and Chen, K.-C. (2012). Smart
attacks in Smart Grid communication networks. IEEE
Communications Magazine, 50(8):24–29.
Cubo, J., Brogi, A., and Pimentel, E. (2012). Towards
behaviour-aware compositions of things in the Fu-
ture Internet. In Proceedings of the 2nd International
Workshop on Adaptive Services for the Future Inter-
net and 6th International Workshop on Web APIs and
Service Mashups, WAS4FI-Mashups ’12, pages 28–
35, New York, NY, USA. ACM.
Duquennoy, S., Grimaud, G., and Vandewalle, J.-J. (2009).
The Web of Things: Interconnecting devices with
high usability and performance. In International Con-
ference on Embedded Software and Systems, ICESS
’09, Hangzhou, Zhejiang, P. R. China, May 25-27,
2009, pages 323–330, Hangzhou, Zhejiang, P. R.
China. IEEE.
Gama, J., editor (2010). Knowledge Discovery from Data
Streams. Advances in Database Systems. Chapman
and Hall/CRC, Florida, USA, first edition.
Guinard, D., Trifa, V., Karnouskos, S., Spiess, P., and Savio,
D. (2010). Interacting with the SOA-Based Internet of
Things: Discovery, query, selection, and on-demand
provisioning of Web Services. Services Computing,
IEEE Transactions on,, 3(3):223–235.
Guinard, D., Trifa, V., Mattern, F., and Wilde, E. (2011).
From the Internet of Things to the Web of Things:
Resource-oriented architecture and best practices. In
Architecting the Internet of Things, pages 97–129.
Springer Berlin Heidelberg.
Gungor, V. C., Sahin, D., Kocak, T., Erg¨ut, S., Buccella,
C., Cecati, C., and Hancke, G. P. (2013). A survey
on smart grid potential applications and communica-
tion requirements. IEEE Trans. Industrial Informat-
ics, 9(1):28–42.
INTEGRIS (2011). INTEGRIS FP7 Project INTelligent
Electrical Grid Sensor communications. ICT-Energy-
2009 call (number 247938).
Luckenbach, T., Gober, P., Arbanowski, S., Kotsopoulos,
A., and Kim, K. (2005). TinyREST-A protocol for
integrating sensor networks into the Internet. In in
Proc. of REALWSN.
Navarro, J., Zaballos, A., Sancho-Asensio, A., Ravera, G.,
and Armendariz-I˜nigo, J. E. (2012). The information
system of INTEGRIS: INTelligent Electrical GRId
sensor communications. Industrial Informatics, IEEE
Transactions on, PP(99):1.
Priyantha, N. B., Kansal, A., Goraczko, M., and Zhao, F.
(2008). Tiny Web Services: design and implementa-
tion of interoperable and evolvable sensor networks.
In Proceedings of the 6th International Conference on
Embedded Networked Sensor Systems, SenSys 2008,
Raleigh, NC, USA, November 5-7, 2008, pages 253–
266, St. Louis, Missouri, USA. ACM.
Zaballos, A., Vallejo, A., and Selga, J. M. (2011). Heteroge-
neous communication architecture for the Smart Grid.
IEEE Network, 25(5):30–37.
Zeng, D., Guo, S., and Cheng, Z. (2011). The Web of
Things: A survey (invited paper). JCM, 6(6):424–438.
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
336