Semantic Interoperability for Web Services based Smart Home Systems
Hannu J¨arvinen and Petri Vuorimaa
Department of Media Technology, Aalto University School of Science, Espoo, Finland
Keywords:
Ambient Intelligence, Building Automation, FIPA, Interoperability, JADE, Multi-agent System, oBIX,
Semantic Web, Smart Home, Web Intelligence, Web of Things, Web Services, WebSocket.
Abstract:
One of the key issues in smart home systems is the lack of interoperability between available solutions. While
data representation and communication formats need to be unified, the goal is to provide intelligence for the
system control. In this paper, we present a solution for Web services based smart home systems to publish
semantic data, and to communicate with agent systems. Communication of browser-based agents with a smart
home system is demonstrated. Efficiency of the system is measured and the results provided. The presented
solution provides a basis for exploiting intelligence from multi-agent systems in smart home system control.
1 INTRODUCTION
The main issue with smart home interoperability is
that products from different manufacturers are incom-
patible with each other, and even products of the same
brand are unlikely to really interoperate. A smart re-
frigerator can, for example, send tweets or commu-
nicate its content to a smartphone application, but is
unable to share data with another smart device. True
interoperability can be achieved when heterogeneous
smart components can utilize data and suggest actions
from and to each other. If manufacturers would agree
on common standards, the problem would be nonexis-
tent as the systems would be interoperable by design.
Unfortunately, while the vision of smart buildings is
nothing new, agreeing on the standards between larger
groups of companies has been unsuccessful, except in
entertainment electronics.
Smart home interoperability can be separated into
four different layers. At the bottom, syntactic inter-
operability layer considers the capability of devices to
be connected into the same system. Functional layer
considers inter-device logic that enables actions to be
executed in one component based on the information
from another component. On semantic layer, data is
presented semantically, categorized using context on-
tologies, and linked to other semantic data. Intelli-
gence layer provides services for controlling the smart
home in more sophisticated fashion, including, e.g.,
adaptive logic for automatizing the smart home be-
havior and programming. The work presented in this
paper concentrates on realizing the semantic layer in
a smart home, and enabling the use of the intelligence
layer with standard agent technology support.
We consider two practices for implementing the
semantic layer. First, we use a standard extensible
smart home data format that allows developers to de-
fine their own abstractions. This creates simple se-
mantics for the data by categorizing different smart
home concepts. Second, we define an ontology based
on this smart home Web language, and transform the
complete data model into a corresponding semantic
representation. The transformation allows represent-
ing smart home data in RDF/OWL. To enable the use
of the intelligence layer, we implement an agent to
handle the communication between smart home and
agent systems. Further, we demonstrate interoper-
ability between the smart home system and browser
agents.
To realize the above contributions, we present de-
tails of the semantic transformation, and the agent
platform integration with the smart home system.
Measurements are made to calculate the system de-
lays caused by the RDF transformation and the agent
platform integration, and to demonstrate the commu-
nication with a browser agent. The smart home sys-
tem is based on open Building Information eXchange
(oBIX) (Considine, 2010) specification and the in-
tegrated agent platform is the Java Agent DEvelop-
ment Framework (JADE) (Bellifemine et al., 1999)
compliant with the Foundation for Intelligent Physi-
cal Agents (FIPA).
141
Järvinen H. and Vuorimaa P..
Semantic Interoperability for Web Services based Smart Home Systems.
DOI: 10.5220/0005201501410148
In Proceedings of the International Conference on Agents and Artificial Intelligence (ICAART-2015), pages 141-148
ISBN: 978-989-758-073-4
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
2 RELATED WORK
Extensive amount of research has focused on smart
home interoperability (Jiang et al., 2004). Many stud-
ies have concentrated on connecting heterogeneous
devices and subsystems together, and providing a uni-
fied interface on top (Perumal et al., 2010; Wang et al.,
2007; Tokunaga et al., 2002; Miori et al., 2006). The
approach is applicable for limited amount of devices,
as protocol converters and integration platforms need
to convert data separately for each protocol. Huge
number of different protocols makes the approach
cumbersome and sometimes impossible due to use of
proprietary protocols.
One of the protocols for the unified high-level
Web services interface is open Building Information
eXchange (oBIX) (Considine, 2010) from the Orga-
nization for the Advancement of Structured Informa-
tion Standards (OASIS). The data format is based on
XML and defines, similarly to any common program-
ming language, a small set of primitive data types for
describing the data. It also allows developers to define
data structures for their own devices and systems.
Recently, the use of semantic technologies in
smart environments has been suggested by many au-
thors (Chen et al., 2009; Wang et al., 2004a; Zhang
et al., 2005). It provides a way to represent data
on a higher semantically meaningful level, and share
common understanding of the concepts using ontolo-
gies. As opposed to attempts of standardizing smart
home protocols, semantic community is cooperating
and the standardization of semantic formats has been
successful. In addition, different application domains
can define their own ontologies. In turn, these ontolo-
gies can link concepts and properties to common on-
tologies, for example, in Linking Open Data
1
(LOD)
cloud. Thus, applications can understand data from
different domains and interoperate.
Various studies propose ontology-based context
models for representing smart home context informa-
tion semantically (Gu et al., 2004; Kim and Choi,
2006; Xu et al., 2009). Ontologies, such as Cobra-
Ont (Chen et al., 2003) and CONON (Wang et al.,
2004b) have been defined to model context informa-
tion. Such research interest in using ontologies sug-
gests that adding a semantic layer in smart home can
be useful for developing intelligent applications.
Use of Multi-Agent System (MAS) for the build-
ing control has also been studied by, e.g., (Huberman
and Clearwater, 1995). They present a system using
market-based MAS to providesolution to the problem
of thermal resource distribution. However, to gener-
1
The Linking Open Data (LOD) cloud - http://
lod-cloud.net/
alize the idea, detailed research has to be done on the
integration techniques between the building systems
with the MAS, including data transformation and se-
mantic presentation.
Sashima et al. developed an agent-based context-
mediated Web service coordination framework in
ubiquitous computing (Sashima et al., 2005). In their
system, each agent is a proxy between SOAP Web
services and FIPA Agent Communication Language
(ACL). They suggest solutions to coordination issues
such as context-aware matchmaking and representa-
tion alignment. Similarly to their work, we trans-
form standard smart home Web services language
into FIPA ACL and RDF/OWL. However, in our ap-
proach the transformation is handled by one central-
ized agent. Thus, individual system components only
need to follow standards from their own domain.
Coyle et al. suggested a sensor fusion-based mid-
dleware for smart homes, called ConStruct (Coyle
et al., 2007). Similarly to an integration platform,
the solution interconnects heterogeneous devices and
presents the information with a unified format, in
this case, RDF. In their earlier research (Coyle et al.,
2006), they outlined the idea of transforming XML
information from oBIX via XSLT to RDF and illus-
trated it with the example data of a simple thermostat
from the oBIX specification. However, the transfor-
mation is not presented in detail and generated seman-
tic data seem not to follow any static ontology. In ad-
dition, the transformation is lossy. In Section 4.1, we
compare this with our ontology-based transformation.
3 SEMANTIC
INTEROPERABILITY
Semantic representation of the smart home informa-
tion can be used to facilitate the interoperability. As
stated above, on semantic interoperability layer data
is presented semantically, categorized using context
ontologies, and linked to other semantic data. In ad-
dition to the use of standard Semantic Web technolo-
gies, other techniques are also considered. Low-level
device protocols and syntactic layer Web based data
formats can also provide support for describing se-
mantics.
The semantic support in our smart home system
consists of three layers. Figure 1 illustrates how lay-
ers build from simple to more complex and interoper-
able semantics. Smart Home Web Service Language
Layer: The most basic semantics are defined with
primitive oBIX data types. They can be used to cat-
egorize data into different object types, e.g., boolean
and string value objects. Then, developers can define
ICAART2015-InternationalConferenceonAgentsandArtificialIntelligence
142
!"#$%&'()(*(+,-
!"#$%&'()*+,*---%./
!"#$%.!/*'0.*-
!012%"3)*4567&"87*---%./
1)0'*%2!),
3,4%1,'+(.,%
50/6706,%508,'
9:;
!9:;<$5=7%9:;<9730>987---%./
<35
!0?(<@'47:A#:"&":>'(---%./
1,)0/*(.%3,4
50/6706,%508,'
!"#$%</*!=!68
!01"B<&'(/+,!.01"B<&'(/
;#>?@9:;A
!;"='<C8$"0#---%./
</*!=!68%508,'
Figure 1: Semantic interoperability layers.
concepts with prototype structures, called oBIX con-
tracts. This extensibility makes oBIX a strong can-
didate when choosing a smart home Web language.
Semantic Web Language Layer: On top of the oBIX
primitives and contracts, Semantic Web technologies
are exploited. With RDF we can represent the data as
subject-predicate-object triples, and make statements
about the Web resources as smart home information.
OWL makes the language more expressive and pro-
vides a standard way to define ontologies. Ontology
Layer: With these tools, we are able to define oBIX
ontology for describing the oBIX smart home data
in Semantic Web. We are also able to transform the
oBIX XML data into RDF/OWL in compliance with
the oBIX ontology. Now, we can exploit other on-
tologies alongside with the oBIX ontology. As we
are integrating an agent platform with the smart home
system, we can use FIPA-RDF0 ontology for the stan-
dard presentation of the ACL message content. We
could also define generic smart home interoperability
ontology for describing the device attributes and pro-
viding device and vendor specific annotations, e.g.,
links to device manuals, software updates, and prod-
uct catalogs.
4 IMPLEMENTATION
The objective of the research was to provide seman-
tic interoperability layer for Web based smart home
system, and enable intelligence layer with support for
standard FIPA compliant agents to communicate with
the smart home system. Additionally, we wanted to
allow browser-based agents to communicate with the
system.
The smart home server used in the implementation
is a further developmentresult of our Java based oBIX
server. Required features were developed to sup-
port semantic technologies and the integration with
an agent platform. JADE agent platform was inte-
grated with the oBIX server using a gateway agent
that communicates with both systems. JADE is FIPA
compliant and one of the most used agent platforms
available.
The main components of the system are illustrated
in Figure 2. Integration of the oBIX Server with the
MAS is realized through the Gateway Agent. RDF
Transformer in the oBIX server is used to convert
the oBIX XML data into corresponding RDF/OWL
representation. The integrated agent platform uti-
lizes WebSocket communication when communicat-
ing with external agents. The Gateway Agent func-
tions as a bridge between the two worlds, oBIX and
agents. Any agent communicating with the Smart
Home System need to have an ontology alignment
with FIPA-RDF0 but also with oBIX ontology.
4.1 Ontology and Transformation
We defined oBIX ontology to represent the device
information semantically. Now, with the ontology
there is an explicit way of describing the oBIX data
in RDF/OWL format. When comparing our method
with the one in (Coyle et al., 2006), the transforma-
tion process and the semantic representation are dif-
ferent. Their example illustrated a transformation of,
e.g., oBIX name facet values into RDF predicates.
Thus, it is impossible to define an ontology, as ev-
ery oBIX XML element with a new name would re-
quire an update to the ontology. Additionally, their
transformation is lossy as only part of the informa-
tion from oBIX XML format is included in the RDF
representation. Our transformation instead generates
RDF/OWL that is based on the oBIX ontology. The
ontology covers every possible oBIX element. It de-
fines classes for all the oBIX primitives, and proper-
ties for their facets. As the ontology covers the whole
oBIX object model, the transformation is also loss-
less. A transformation example is depicted in Figure
3. Each XML element in oBIX is transformed into
OWL NamedIndividual element. While oBIX has el-
ement hierarchy, RDF/OWL lines up all NamedIndi-
vidual elements to the same depth. Hierarchy infor-
mation is stored using hasChild object property.
In practice, RDF Transformer module was devel-
oped for the oBIX server, and two different XSLT
transformation files were created. The first transfor-
SemanticInteroperabilityforWebServicesbasedSmartHomeSystems
143
!"#$%&'(")&!*+%)"
,-.%/012)3%&!*+%)"
4#%)5#*&12)3%
67%)$3#.&,-.%/0
12)3%&!*+%)"
!"#$%&'"()*+,
$"-./&")0/-"&(%-1)2$03
45678")&%99:6/&5(/%6)(%;5-7)5<"6(=
>"(&?)7"./&")75(5
,-%./7")=:#=&-/@(/%6)9"&?56/=9
(89:&;./)3% 8$(5+)$&12)3%
%ABC)C*D
)
E0>FG!D
HIDF>B,HJE0>FG!D
)
<3%(.(2*&&&&&&&&&&&&=)+>$/?%/(3&
%#/K)))))))))))))))))))))))))))%ABC)%#L"&()9%7"8))))
M/@5J-7MN)))))))))))))))))))))>B,H)E0>)I%6("6()D56<:5<")
$4J566%(5(/%6=)))))))))956:M5&(:-"-F7"./&")566%(5(/%6=
!"#$%$&'
(%)&"*+"#
(89:&!)$@)$
E0>)+-56=M%-9"-
4++,)$"-."-
0"./&")05(5
12)3%
!"#$%&'"()*+,
Figure 2: Integration architecture of the smart home system.
!"#$%$!&
'()*+,
'-./,,,/0/1/',,,,"0$ 2'2(/21
/,,,,,"0#,1//,,,,"0*--342/2$$%1
//,,,,"0$5672
,)'8'''/1
'(',,,/0/19
/
4''/1
,)'8''
,)'8'''/1
'(',,,/0/1-
'()*+'()
,,
!"#$%$!&
,)'8''
'-.
Figure 3: An example transformation from oBIX XML to RDF/OWL.
mation generates pure RDF/OWL compliant with the
oBIX ontology, and the second transformation addi-
tionally takes into account the FIPA-RDF0 ontology
as it is used to generate data for the agent systems.
Contracts in oBIX are not represented as classes in the
oBIX ontology because that would require dynamic
updates to the ontology itself. However, that could
be useful in some application scenarios, and could be
implemented as a separate oBIX contract ontology.
The oBIX ontology provides a common vocabu-
lary for the agents to handle the smart home informa-
tion. Additionally, it can be exploited on the intelli-
gence layer for making inferences from the oBIX data
to provide reasoning and find inconsistencies. Thus,
while it here functions as a common semantic data
representation format with shared vocabulary, it also
enables further applications to be built on top.
4.2 Data Interfaces
The implemented system provides three different data
interfaces for the communication with external ap-
ICAART2015-InternationalConferenceonAgentsandArtificialIntelligence
144
!"#$"%&'$()
)*+,-.,/)!)01,-234.,-2454,/)*-06,)$/78+,/91,-2:;/78+,/)*0../,++,+)!+,<=,->,)8+*??;/78+,/91,-2*@@@A?0>>BB
)*/,>,4C,/)!+,2)!)01,-234.,-2454,/)*-06,)7;4D01,-2:EFGHIFFHJKHEGJ*EGAA?L9M()*0../,++,+)!+,<=,->,)8+*??EFGHIFFHJKHEGJ*@@@N?0>>BB)B
)*>7-2,-2)OPQD6R)C,/+47-STEHGTQU)
P/.5*&MV)D6R-+*/.5STW22X*??888H8FH7/1?EAAA?GI?II3/.53+Y-20D3-+ZT
)))))))))D6R-+*54X0STW22X*??888H54X0H7/1?+>W,60+ZT
)))))))))D6R-+*78RSTW22X*??888H8FH7/1?IGGI?G@?78RZT
)))))))))D6R-+*7;4DSTW22X*??888H7;4DH7/1?/,+7=/>,?7;4DZTU
)))P/.5*M,+>/4X247-)/.5*0;7=2STW22X*??EFGHIFFHJKHEGJ*NGNG?2,6X+,-+7/?C0R=,Z/,0RT?U
P?/.5*&MVUOB
!'[V\&]
)*+,-.,/))!)01,-234.,-2454,/)*-06,)7;4D01,-2:EFGHIFFHJKHEGJ*EGAA?L9M())*0../,++,+)!+,<=,->,)8+*??EFGHIFFHJKHEGJ*@@@N?0>>)BB
)*/,>,4C,/))!+,2)!)01,-234.,-2454,/)*-06,)$/78+,/91,-2:;/78+,/))*0../,++,+)!+,<=,->,)8+*??;/78+,/91,-2*@@@A?0>>)BB)B
)*>7-2,-2))OPQD6R)C,/+47-S^OEHG^O),->7.4-1S^O#_V3N^OQU
P/.5*&MV)D6R-+*/.5S^OW22X*??888H8FH7/1?EAAA?GI?II3/.53+Y-20D3-+Z^O
)))))))))D6R-+*54X0S^OW22X*??888H54X0H7/1?+>W,60+Z^O
)))))))))D6R-+*78RS^OW22X*??888H8FH7/1?IGGI?G@?78RZ^O
)))))))))D6R-+*7;4DS^OW22X*??888H7;4DH7/1?/,+7=/>,?7;4DZ^O
)))))))))D6R-+*D+S^OW22X*??888H8FH7/1?_&?IGGN?&(%3D6R3IGGNEEIKZ^O
)))))))))D6R-+*D+4S^OW22X*??888H8FH7/1?IGGE?`]a">W,6034-+20->,^OU
)))P78R*[06,.'-.4C4.=0R)/.5*0;7=2S^OW22X*??EFGHIFFHJKHEGJ*NGNG?2,6X+,-+7/?C0R=,Z/,0R^OU
))))))P/.5*2YX,)/.5*/,+7=/>,S^OW22X*??888H7;4DH7/1?/,+7=/>,?7;4DZ&,0R^O?U
))))))P7;4D*-06,U+,-+7/C0R=,P?7;4D*-06,U
))))))P7;4D*W/,5UW22X*??EFGHIFFHJKHEGJ*NGNG?2,6X+,-+7/?C0R=,P?7;4D*W/,5U
))))))P7;4D*.4+XR0Y[06,U_,6X,/02=/,)",-+7/)b0R=,P?7;4D*.4+XR0Y[06,U
))))))P7;4D*8/420;R,U2/=,P?7;4D*8/420;R,U
))))))P7;4D*C0RUIGHANc@AKFNGGNIKc@P?7;4D*C0RU
)))P?78R*[06,.'-.4C4.=0RU
P?/.5*&MVUO)
)*/,XRY3842W))$/78+,/91,-2:;/78+,/EJG@EKEIJANNE)B
!"#$"%&'
!"%()*%"'
Figure 4: A subscription request from a browser agent to the gateway agent and a response sent by the gateway agent.
Table 1: Characteristics of the smart home data interfaces.
!"#$%$&' !"#$%()* +,)-%()*
./!0!1!2
*++& *++& ,#-!./0#$
'3456357
.123 .123 4567.123
)303%&!872
.123 89: 89:
97/:32:;30:!4
3;6 3;6 !$<=>()7)3;6
plications. Main characteristics of the interfaces are
summarized in Table 1.
The oBIX specification (Considine, 2010) de-
fines both Simple Object Access Protocol (SOAP)
and Hypertext Transfer Protocol (HTTP) bindings
for the communication. In our oBIX server, the
HTTP interface is supported. The interface is sim-
ple with direct mappings from read, write, and in-
voke operations to corresponding HTTP GET, PUT,
and POST request methods. Testing the GET method
is thus as easy as writing the URI of the object,
e.g., http://obixserver/temperaturesensor, to the navi-
gation bar of any browser.
The system also supports requesting RDF/OWL
data in XML serialization over HTTP. This is im-
plemented by adding a parameter support for the
URIs. For example, the example URI mentioned
above can be requested in RDF format using the URI
http://obixserver/temperaturesensor?format=rdf.
The additional delay caused by the transformation
process is discussed in Section 5.
The integrated agent platform provides a standard
interface for FIPA compliant agents to communicate
with the gateway agent. The gateway agent in turn
provides access to the oBIX information in the smart
home system. Now, intelligent agents can communi-
cate with the system using ACL and understand the
content described in FIPA-RDF content language.
4.3 Gateway Agent
Smart home information is provided to the agents
through the integrated JADE agent platform. A spe-
cial gateway agent in the platform was developed to
handle the communication with the oBIX server. Ex-
ternal agents can contact the gateway agent to access
the information on the oBIX server.
The gateway agent interprets the FIPA-RDF mes-
sages coming from other agents and forwards the re-
quests to the oBIX server. At the moment, query
(FIPA, 2002b), request (FIPA, 2002c), and subscribe
(FIPA, 2002d) interaction protocols are supported.
Query-ref communicative act equals to oBIX read re-
quest (HTTP GET) and is used for querying for an
identified object. Request communicative act equals
to oBIX write request (HTTP PUT) and is used to re-
quest to perform a write action. Subscribe commu-
nicative act is used to request a subscription on an
object and subsequent updates when the referenced
object changes. Figure 4 presents a subscribe request
message sent to the gateway agent and one of the re-
sponses from the gateway agent.
4.4 Real-time Data Update
The oBIX specification describes the oBIX data
model, XML presentation and the HTTP REST and
SOAP communication interfaces. However, it does
not define any details on how the devices are actually
connected to the server. One possibility is to integrate
oBIX server into device or subsystem controllers that
have direct access to the device data. The idea of the
specification is that enterprise applications can access
the smart home information through the defined oBIX
SemanticInteroperabilityforWebServicesbasedSmartHomeSystems
145
interface. Unfortunately, there is no support for real-
time data updates. In practice, while oBIX provides
a feature called watches for reducing the traffic load
when polling for changes in the device data, it does
not offer any true real-time notification method to im-
plement server-push communication.
The agent platform integration provides a definite
improvement to the real-time communication abilities
of the smart home system. The subscribe interaction
protocol (FIPA, 2002d) can be used to receive up-
dates on specified data from the server immediately
when available. It provides the server-push method
that the oBIX specification lacks. Automatic notifi-
cations followed by the subscriptions also prevent the
possibility of occasionally not registering some fre-
quently changing data values.
4.5 WebSocket and Browser Support
FIPA standards define a reference model for agent
Message Transport Service (MTS) (FIPA, 2002a).
The model defines the modules in agent platforms
that are responsible for delivering messages. Specifi-
cally, Message Transport Protocols (MTPs) are proto-
col level implementations used for message transport
between platforms. A platform can have any number
of MTPs for inter-platform communication.
Web browsers are traditionally HTTP clients. This
has complicated the development of web applications
for the Web 2.0. WebSocket is a fairly new technol-
ogy that provides full-duplex communication chan-
nels between web browsersand web servers. Once the
client (browser) sends a handshake, the HTTP server
interprets it as an upgrade request and establishes a
WebSocket. Now, the client can receive data from the
server at any time without specifically requesting it.
We run the integrated agent platform with the
WebSocket MTP (J¨arvinen et al., 2014). While Web-
Socket communication is more appropriate for the
agent communication paradigm than, e.g., HTTP,
there is another benefit that motivated us to utilize
it. When using a WebSocket, client always initiates
the communication. However, after the bi-directional
pipe has once been established, both sides can send
and receive data at any time. This provides a highly
useful server-push feature for the systems that do not
want to implement server side functionalities.
Using WebSocket MTP in the integrated agent
platform allows browsers to communicate with the
agents. This provides a standard way for desktop
and mobile device browsers to communicate with the
smart home system and receive updates in real-time.
Mobile devices can now take part in distributed com-
puting of a smart home. This can result in interesting
applications in the future as mobile devices have mul-
tiple functionalities that can benefit the smart home
system. In addition to the possible sensor and actua-
tor services, smart phones can provide different kind
of context information, e.g., location and user avail-
ability, and function as a remote user interface and
computational entity for several purposes.
To facilitate the development of browser-based
agents, we have implemented an ACL JavaScript li-
brary. It handles the creation of the FIPA message
envelope and the ACL representation. The client only
needs to define the agent name, address, and the actual
message content. The library was used in the mea-
surements described in the Section 5.
5 PERFORMANCE
Real-time requirements of smart home systems set
limits for system delays. We evaluated the presented
system by conducting performance measurements.
The objective of the measurements was to find out
additional system delays caused by the data conver-
sion and the agent platform integration. Addition-
ally, we measured the network traffic load when us-
ing different data interfaces. As we also wanted to
demonstrate the interoperability with browser agents,
the tests were conducted with JavaScript client pro-
grams running in a browser. Further, we developed a
browser agent for visualizing temperature data.
To measure the network traffic, and to calculate
the additional delays caused by the data conversion
and the use of an agent platform, we developed four
JavaScript test programs. In each test, the client pro-
gram and the server program were run in separate
computers connected with a hub. Tcpdump packet
analyzer
2
program was used to record the network
traffic for the post test analysis. Also, for the three
first test programs the Round-Trip Time (RTT) of
each request-response cycle was recorded. Wireshark
packet analyzer software
3
was used for the network
traffic analysis.
Each test program fetched a piece of smart home
information repeatedly 10 000 times from the server.
First test program sent HTTP GET requests to the
oBIX server fetching an element in oBIX XML for-
mat. Second test program fetched the same infor-
mation in RDF format over HTTP. Both third and
fourth test programs again fetched the same informa-
tion, but used WebSocket communication, ACL, and
the integrated JADE agent platform. The difference in
2
Tcpdump packet analyzer - http://www.tcpdump.org/
3
Wireshark packet analyzer - http://www.wireshark.org/
ICAART2015-InternationalConferenceonAgentsandArtificialIntelligence
146
*++&),-./
*++&)012
3#4!,56#$)&,7789(
3#4!,56#$)!:4;5<84#
>
C>>>
%>>>>
%C>>>
B>>>>
BC>>>
D>>>>
DC>>>
A>>>>
AC>>>
%>%?>
%?%@%
A%A@B
BBB?@
1'$').9$#<E'5#
+<'EE85)F6-G
Figure 5: Comparison of traffic load produced using differ-
ent data interfaces.
*++&),-./ *++&)012 3#4!,56#$)&,7789(
G
D
F
J
H
%G
%D
%F
%J
BCBDD
%EC%FE
%GCHII
1'$').9$#<K'5#
:;#<'(#)0,=9>?+<8@)+8A#)LAMN
Figure 6: Comparison of RTT (ms) using different data in-
terfaces.
these tests is that the third program was polling the in-
formation with Query-ref messages, while the fourth
program subscribed to the information once and got
an update every time the information changed on the
server. The sample data on the server was updated in
100 ms time intervals and the clients were polling that
information in 300 ms time intervals.
Comparison of traffic load produced using differ-
ent data interfaces is depicted in Figure 5. More traffic
is produced using the integrated agent platform with
WebSocket interface compared to the oBIX HTTP in-
terface. This can be explained with the differences in
the message format. While the overhead of the HTTP
protocol frame is many times larger compared to the
WebSocket protocol frame (J¨arvinen et al., 2014), the
message body in our application overturns that ad-
vantage. As seen in Figure 3, the RDF/OWL XML
format is clearly more space-consuming than the cor-
responding oBIX XML format. In addition, agent
communication requires a comparatively large mes-
sage envelope and ACL message body, in addition to
the actual FIPA-RDF content. It is notable that with
no need to poll for changes in the data, agent sub-
scription method gains significantly in the compari-
son and ends up with almost equal traffic load with
HTTP RDF method. Larger packet size is compen-
sated with the reduced number of packets.
Comparison of the round-trip times is depicted in
Figure 6. It can be seen that the additional delay
caused by the RDF transformation is only approxi-
mately 7 ms, which is acceptable in most use cases.
Surprisingly, using the integrated agent platform in-
terface results in less delay than using HTTP RDF in-
terface. From the results we can conclude that the
additional delays are not significant. The subscrip-
tion test was not included in these results, as it does
not have separate request for each update, and is thus
not applicable. However, it is notable that the sub-
scribe method is naturally the best option when im-
plementing applications with strict real-time require-
ments. Subscription method also ensures that every
update is actually communicated to the client, while
in oBIX polling it is possible to miss rapid changes
that happen between poll cycles.
In addition to the measurements, we developed a
temperature view browser agent. The application sub-
scribed to the temperature information on the smart
home server and received sensor value updates every
100 ms. The application visualized the received in-
formation in real-time with HTML5 Canvas 2D API.
6 CONCLUSIONS
We presented a solution for realizing the semantic in-
teroperability layer in Web services based smart home
systems. The system provides smart home informa-
tion semantically in RDF/OWL format using both an
HTTP interface in oBIX server and a WebSocket in-
terface in JADE agent platform. The system enables
the use of MASs to provide intelligence for smart
home control.
To represent oBIX information semantically, we
defined oBIX ontology. The ontology consists of
classes for all the oBIX primitives, and properties
for their facets. Every possible oBIX element can be
aligned with the ontology. Additionally, FIPA-RDF0
ontology was used for the agent platform integration
to support standard FIPA agent communication.
We conducted performance measurements to find
out the additional delays caused by the data trans-
formation and the agent platform integration, and to
measure the traffic load when using different inter-
faces. Standard agent communication envelope and
ACL format cause more traffic load when using the
agent platform interface. However, the use of sub-
scription method can even out the difference signifi-
cantly. The additional delays are not significant.
In addition to semantic interoperability, subscrip-
tion support enabled real-time data updates that do not
normally exist in oBIX solutions. Real-time commu-
nication abilities can be of high importance in vari-
SemanticInteroperabilityforWebServicesbasedSmartHomeSystems
147
ous smart home scenarios. Automatic updates also
prevent the possibility of occasionally not registering
some frequently changing data values.
We developed an ACL JavaScript library, and
demonstrated interoperability between browser
agents and the smart home system. The use of
WebSocket MTP enabled the communication. Using
the default HTTP MTP in JADE it would be impos-
sible for browser agents to receive messages from
the smart home system. Providing a standard Web
interface for mobile devices, e.g., smart phones can
result in interesting applications in the future. We
consider browser agents as important technology to
participate in realizing the ambient intelligence.
Both the idea of transforming smart home Web
services data into RDF and to represent it in stan-
dard FIPA-RDF agent language have been proposed
by earlier research in the field. However, detailed
and evaluated generalizable approach similar to ours
to map Web services based smart home information
from oBIX into RDF has not been presented before.
Future work includes better data mapping with
known ontologies, implementation of real world use
cases, and development of intelligent agent services.
REFERENCES
Bellifemine, F., Poggi, A., and Rimassa, G. (1999). Jade–
a fipa-compliant agent framework. In Proceedings of
PAAM, volume 99, page 33. London.
Chen, H., Finin, T., and Joshi, A. (2003). An ontology
for context-aware pervasive computing environments.
The Knowledge Engineering Review, 18(3):197–207.
Chen, L., Nugent, C., Mulvenna, M., Finlay, D., and Hong,
X. (2009). Semantic smart homes: towards knowl-
edge rich assisted living environments. In Intelligent
Patient Management, pages 279–296. Springer.
Considine, T. (2010). Open Building Information Exchange
(oBIX) version 1.1. OASIS Working Draft 06, June
2010.
Coyle, L., Neely, S., Rey, G., Stevenson, G., Sullivan, M.,
Dobson, S., and Nixon, P. (2006). Sensor fusion-based
middleware for assisted living. In Smart Homes and
Beyond: ICOST 2006: 4th International Conference
on Smart Homes and Health Telematics, volume 19,
pages 281–288. IOS Press.
Coyle, L., Neely, S., Stevenson, G., Sullivan, M., Dobson,
S., and Nixon, P. (2007). Sensor fusion-based middle-
ware for smart homes. International Journal of Assis-
tive Robotics and Mechatronics, 8(2):53–60.
FIPA (2002a). FIPA Agent Message Transport Service
Specification. Foundation for Intelligent Physical
Agents.
FIPA (2002b). FIPA Query Interaction Protocol Specifica-
tion. Foundation for Intelligent Physical Agents.
FIPA (2002c). FIPA Request Interaction Protocol Specifi-
cation. Foundation for Intelligent Physical Agents.
FIPA (2002d). FIPA Subscribe Interaction Protocol Speci-
fication. Foundation for Intelligent Physical Agents.
Gu, T., Wang, X. H., Pung, H. K., and Zhang, D. Q. (2004).
An ontology-based context model in intelligent envi-
ronments. In Proceedings of communication networks
and distributed systems modeling and simulation con-
ference, volume 2004, pages 270–275.
Huberman, B. A. and Clearwater, S. H. (1995). A multi-
agent system for controlling building environments. In
ICMAS, pages 171–176.
J¨arvinen, H., Garcia-Gasulla, D., and Cort´es, U. (2014). A
push-based agent communication model empowering
assistive technologies. International Journal on Arti-
ficial Intelligence Tools, 23(1).
Jiang, L., Liu, D.-Y., Yang, B., et al. (2004). Smart home
research. In Proceedings of the Third Conference on
Machine Learning and Cybernetics, pages 659–664.
Kim, E. and Choi, J. (2006). An ontology-based context
model in a smart home. In Computational Science and
Its Applications-ICCSA 2006, pages 11–20. Springer.
Miori, V., Tarrini, L., Manca, M., and Tolomei, G. (2006).
An open standard solution for domotic interoperabil-
ity. 52(1):97–103.
Perumal, T., Ramli, A. R., Leong, C. Y., Samsudin, K.,
and Mansor, S. (2010). Middleware for heteroge-
neous subsystems interoperability in intelligent build-
ings. Automation in Construction, 19(2):160 – 168.
Sashima, A., Izumi, N., and Kurumatani, K. (2005). Agents
that coordinate web services in ubiquitous comput-
ing. In Ubiquitous Computing Systems, pages 131–
145. Springer.
Tokunaga, E., Ishikawa, H., Kurahashi, M., Morimoto, Y.,
and Nakajima, T. (2002). A framework for connect-
ing home computing middleware. In ICDCSW, pages
765–770, Washington, DC, USA. IEEE Computer So-
ciety.
Wang, S., Xu, Z., Cao, J., and Zhang, J. (2007). A middle-
ware for web service-enabled integration and interop-
eration of intelligent building systems. Automation in
Construction, 16(1):112 – 121. CAAD Futures, 2005.
Wang, X., Dong, J. S., Chin, C., Hettiarachchi, S. R., and
Zhang, D. (2004a). Semantic space: an infrastruc-
ture for smart spaces. Pervasive Computing, IEEE,
3(3):32–39.
Wang, X. H., Zhang, D. Q., Gu, T., and Pung, H. K.
(2004b). Ontology based context modeling and rea-
soning using owl. In Pervasive Computing and Com-
munications Workshops, 2004. Proceedings of the
Second IEEE Annual Conference on, pages 18–22.
Ieee.
Xu, J., Lee, Y.-H., Tsai, W.-T., Li, W., Son, Y.-S., Park, J.-
H., and Moon, K.-D. (2009). Ontology-based smart
home solution and service composition. In Embedded
Software and Systems, 2009. ICESS’09. International
Conference on, pages 297–304. IEEE.
Zhang, D., Gu, T., and Wang, X. (2005). Enabling
context-aware smart home with semantic web tech-
nologies. International Journal of Human-friendly
Welfare Robotic Systems, 6(4):12–20.
ICAART2015-InternationalConferenceonAgentsandArtificialIntelligence
148
