A Component-Based Approach For Service Distribution In Sensor Networks

HAL Id: inria-00522709
https://hal.inria.fr/inria-00522709
Submitted on 1 Oct 2010
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entific research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
A Component-based Approach for Service Distribution
in Sensor Networks
Amirhosein Taherkordi, Romain Rouvoy, Frank Eliassen
To cite this version:
Amirhosein Taherkordi, Romain Rouvoy, Frank Eliassen. A Component-based Approach for Service
Distribution in Sensor Networks. 5th International Workshop on Middleware Tools, Services and
Run-Time Support for Sensor Networks, Nov 2010, Bangalore, India. ฀inria-00522709฀

A Component-based Approach for Service Distribution in
Sensor Networks
Amirhosein Taherkordi
University of Oslo
Department of Informatics
N-0314 Oslo
amirhost@ifi.uio.no
Romain Rouvoy
ADAM Project-team
INRIA Lille – Nord Europe,
University Lille 1,
LIFL CNRS UMR 8022
F-59650 Villeneuve d’Ascq
romain.rouvoy@inria.fr
Frank Eliassen
University of Oslo
Department of Informatics
N-0314 Oslo
frank@ifi.uio.no
ABSTRACT
The increasing number of distributed applications over Wire-
less Sensor Networks (WSNs) in ubiquitous environments
raises the need for high-level mechanisms to distribute sen-
sor services and integrate them in modern IT systems. Ex-
isting work in this area mostly focuses on low-level network-
ing issues, and fails to provide high-level and off-the-shelf
programming abstractions for this purpose. In this paper,
we therefore consider WSN programming models and service
distribution as two interrelated factors and we present a new
component-based abstraction for integrating WSNs within
existing IT systems. Our approach emphasizes on reify-
ing distribution strategies at the software architecture level,
thus allowing remote invocation of component services, and
facilitating interoperability of sensor services with the Inter-
net through Web service-enabled components. The latter is
efficiently provided by incorporating the REST architectural
style—emphasizing on abstraction of high-level services as
resources—to our component-based framework. The prelim-
inary evaluation results show that the proposed framework
has an acceptable memory overhead on a TelosB sensor
platform.
Categories and Subject Descriptors
C.2.4 [Computer-Communication Networks]: Distributed
Systems—Distributed Applications
Keywords
Wireless sensor networks, integration, component-based ser-
vice distribution, RESTful Web services
1. INTRODUCTION
WSNs have been recognized as an emerging technology
for monitoring and controlling a variety of phenomena, such
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
MidSens ’10, November 30, 2010 Bangalore, India.
Copyright 2010 ACM 978-1-4503-0454-2/10/11 ...$10.00.
as environmental surveillance, infrastructures, home and of-
fice, and medical environments [3, 17, 25]. Whereas the
early WSN applications were primarily concerned with sens-
ing primitive environmental data and delivering those raw
data to a central node, recent applications consider sensor
nodes as service-enabled devices with the ability of providing
distributed services to other nodes of the network.
The other concern is related to using WSNs in ubiqui-
tous computing environments, where sensor nodes populate
with actuators, RFID readers, and mobile devices for mon-
itoring ambient environments and reacting to the external
stimuli gathered by different devices [1, 8, 14]. As each
device in such a heterogeneous environment has its own re-
quirements in terms of system software and communication
protocol, integrating them at a higher software level brings
the challenges of service orchestration and interoperability.
Furthermore, sensor nodes as one of the technologies driv-
ing the future Internet of Things [10] should be equipped
with protocols that enable them to interoperate with every
IP-enabled node across the Internet.
In contrast to the approach of plain text message stream-
ing widely adopted in traditional“sense and send”models [2,
20], the service-oriented approaches in WSNs rely on high-
level ad-hoc communication protocols to discover and ex-
hibit services across different computing environments. Al-
though there have been a number of significant efforts to
distribute sensor services and integrate them with conven-
tional network platforms [11, 19, 26], the state-of-the-art
mostly focuses on low-level APIs and fails to provide a con-
crete communication abstraction relieving the programmer
from dealing with the tedious and error-prone distribution
tasks in WSNs.
Thus, in this paper we aim at providing a software frame-
work for WSNs that enables programmers to develop dis-
tributed sensor services and integrate them with existing
IT systems at a high-level programming abstraction. To
achieve that, we leverage the concepts of component-based
programming to design, describe and implement distributed
and interoperable WSN services. In particular, we recon-
sider Remora—a lightweight component model for high-
level programming in WSNs [23]—in order to extend it with
the capability of distributing a component’s services across
the network. The second part of the paper is devoted to
integrating the above component-based solution to a uni-
form interaction model in order to facilitate interoperability
of sensor services with the Internet through Web service-

enabled components. This interaction model is inspired by
REST—an architectural style for distributed systems em-
phasizing scalability of component interactions, generality of
interfaces, and independent deployment of components [9].
From another point of view, this paper illustrates the feasi-
bility of utilizing a component framework to develop REST-
ful Web services for resource-constrained sensor nodes.
The remainder of this paper is organized as follows. Sec-
tion 2 gives a survey of existing approaches. In Section 3, we
briefly discuss the Remora component model. The archi-
tecture and specification of our service distribution proposal
are presented in Section 4, while the implementation and the
evaluation result are discussed in Sections 5 and 6, respec-
tively. Finally, Section 7 concludes this paper and identifies
some future work.
2. RELATED WORK
Efforts in providing distribution and integration approaches
for WSNs can be categorized into four groups: i) providing
low-level APIs to enable distribution within sensor networks
for exchanging raw sensed data among nodes, ii) high-level
component-based techniques to enable remote invocation of
sensor services, iii) exploiting Web services standards to
bridge the gap between sensor nodes and the Internet, and
iv) protocols allowing sensor nodes to connect seamlessly to
other devices in pervasive computing environments.
The first group emphasizes on exposing proprietary in-
terfaces in order to unicast or multicast plain text mes-
sages across the network. Most of the works in this field
are inspired by the concepts of message-oriented communi-
cation. TinyOS [12] as the most popular operating sys-
tem for WSNs provides a number of Active Message In-
terfaces to abstract the underlying radio communications
services and Software Components that implement these in-
terfaces [2]. Other sensor operating systems, such as Con-
tiki [6], have also paved the same way [7]. They have also
paid a significant consideration on integrating WSNs with
the Internet by providing the IP protocol stack for low-power
sensor nodes. However, high-level programming APIs and
application-level service distribution facilities are not ad-
dressed by these proposals.
A number of lightweight component models have also been
proposed to provide RPC-like service invocations in sensor
networks. In addition to the fact that none of them consider
Web service-based distribution, they essentially suffer from
making extensive use of sensor resources or lack of generality.
As an example of the former, OpenCOM [4] offers Compo-
nent Frameworks (CFs) to model local and distributed in-
teractions between cooperating components, but OpenCom
is a generic model and basically developed for resource-rich
platforms. LooCI [13], falling in the latter category, is a
loosely-coupled component infrastructure for WSNs, featur-
ing an Event Bus to bind distributed LooCI components.
Nonetheless, the Java-based implementation of LooCI lim-
its its usage to the SunSPOT nodes.
The third group has been motivated by the concept of
Internet of Things—a technological revolution to connect
daily objects and devices to large databases and networks,
and therefore to the Internet. In this model, Web services
standards are used to integrate WSNs and the Internet, e.g.,
in SOCRADES [5] Web services are tailored at the gateway
device where the Device Profile for Web Services (DPSW)
is used to enable messaging, discovery and eventing on de-
vices with resource restrictions. However, since the current
footprint of DPSW for sensor nodes is too large, this solu-
tion is only deployable on gateways. To overcome this issue,
Priyantha et al. [19] propose a SOAP-based Web services,
called Tiny web services, for WSNs. However, apart from its
complexity, this work mainly focuses on low-level issues re-
lated to Web integration in TinyOS-based sensor networks.
A few works have also been devoted to the use of simple In-
ternet protocols. In fact, these approaches are proposed due
to the high resource needs and complexity of SOAP-based
Web Service protocols for WSNs. TinyREST is one of the
first attempts to integrate WSNs into the Internet [15]. It
uses the HTTP-based REST architecture to retrieve/update
the state of sensors/actuators. The TinyREST gateway
maps a set of HTTP requests to TinyOS messages in order
to link MICA motes to any Internet client. Beside the fact
that in TinyREST only a gateway is able to connect to the
Internet (not any individual sensor node), this approach fails
to follow all standard HTTP methods. The work reported
in [11] also presents a REST-based gateway to bridge the
Web requests to powerful SunSPOT nodes.
Approaches in the last category are based on the Uni-
versal Plug and Play (UPnP) architecture. UPnP is a set
of computer network protocols, promoted by the UPnP Fo-
rum [24], allowing devices to connect seamlessly in the home
and corporate environments. In UPnP, all communications
are peer-to-peer and transferred over TCP/IP, UDP and
HTTP. As most sensor node products are not equipped with
the support of UPnP standards, in [16] a new WSN plat-
form is proposed to support connectivity of sensor nodes
to a UPnP-enabled device. However, UPnP discovery and
control protocols are heavyweight for a typical sensor node
and there is a very limited range of sensor nodes supporting
UPnP-based communications. Furthermore, using UPnP in
this framework imposes many new hardware and new soft-
ware set-up for integration support.
The main difference between our approach and the works
presented above is that we address the service distribution
problem in a top-down manner. Specifically, we first formu-
late high-level service distribution requirements based on a
programming model for WSNs, and then consider how the
programming proposal can simplify and generalize distribu-
tion and integration of sensor services.
3. PROGRAMMING MODEL
We adopt a component-oriented approach to address the
programming needs of service distribution in WSNs. Com-
ponentization provides a high-level programming abstrac-
tion by enforcing interface-based interactions between sys-
tem modules and therefore avoiding any hidden interaction
via direct function call, variable access, or inheritance rela-
tionships [21]. This abstraction instead offers the capabil-
ity of black-box integration of system services. Therefore, it
theoretically becomes a good candidate for developing distri-
bution tasks in WSNs, beside the fact that component-based
software development is extensively used in WSN program-
ming. Thus, in this section we briefly discuss a WSN-specific
component model we have recently proposed—Remora [23].
Then, in the next section, we propose our service distribu-
tion model based on this component framework.
3.1 REMORA in a Nutshell
The main motivation behind proposing Remora is to fa-

cilitate high-level and event-driven programming in WSNs
through a component-based abstraction. Remora achieves
this goal by: i) deploying components within a lightweight
framework executable on every operating system written in
the C language, and ii) reifying the concept of event as
a first-class architectural element simplifying the develop-
ment of event-oriented scenarios. The latter is one of the
key features of Remora since a programming model for em-
bedded systems is expected to support event-driven design.
Reducing software development effort is the other objective
of Remora. A Remora component is composed of two
main artifacts: a component description and a component
implementation.
Component Description. Remora components are de-
scribed in XML as an extension of the Service Component
Architecture (SCA) model [18] in order to make WSN ap-
plications compliant with the state-of-the-art componenti-
zation standards. Based on the SCA Assembly Language,
the component description indicates the specifications of the
component including services, references, interfaces, produc-
ers, consumers, and properties (cf. Figure 1). A service can
expose a Remora interface, which is a separate XML doc-
ument describing the functions provided by the component.
A reference can request a Remora interface, which describes
the operations required by the component. Similarly, a pro-
ducer identifies an event type generated by the component,
while consumer specifies a component’s interest on receiving
a particular type of event.
Figure 1. The XML template for describing R
<?xml version="1.0" encoding="UTF-8"?>
<componentType name="COMPONENT_NAME”>
<service name="SERVICE1_NAME">
<interface.remora name="INTERFACE1_NAME"/>
</service> ...
<reference name="REFERENCE1_NAME">
<interface.remora name="INTERFACE2_NAME"/>
</reference> ...
<property name="PROP1_NAME" type="PROP1_TYPE">
PROP1_DEFAULT_VALUE
</property> ...
<producer>
<event.remora type="EVENT1_TYPE"
name="EVENT1_NAME"/>
</producer> ...
<consumer operation="CONSUMER_OPERATION">
<event.remora type="EVENT2_TYPE"
name="EVENT2_NAME"/>
</consumer> ...
</componentType>
Figure 1: The XML template for describing Remora
components.
Component Implementation. The component imple-
mentation contains operations implementing: i) the compo-
nent’s service interfaces, ii) event handlers, and iii) private
utilities of the component. Remora components are im-
plemented by using the C programming language extended
with a set of new commands. This extension is essentially
proposed to support the main features of Remora, namely,
component instantiation, event processing, and property ma-
nipulation.
Remora Development Process. A Remora application
consists of a set of Remora components, containing descrip-
tions and implementations of software modules (cf. Fig-
ure 2). The Remora engine processes the component de-
scriptions and generates standard C code deployable within
the Remora framework. The framework is an OS-independent
C module supporting the specification of the Remora com-
ponent model. Finally, the Remora framework is deployed
on the target sensor node through the Remora runtime,
which is an OS-abstraction layer integrating the application
with the system software.
Application
REMORA Framework
Description
<XML>
Implementation
C-like
REMORA Components
REMORA
Engine
Deployable
Components
Development Machine Sensor Node
REMORA Framework
REMORA Runtime
Sensor Hardware
Operating System
nts
l
Figure 2: Development process of Remora applica-
tions.
4. COMPONENT-BASED SERVICE DISTRI-
BUTION
Remora provides a complete component-based abstrac-
tion for node-level programming in the sense that the ac-
cessibility scope of Remora services is limited to the run-
time process of a sensor node. Services deployed on two
different nodes can communicate through an indirect ad-
hoc interaction model, which is message-based and handled
by low-level data transmission protocols provided by sensor
operating systems. It means that for network-level program-
ming, the developer requires to switch from component-
based programming to system-level messaging techniques,
a cumbersome and error-prone task. It is also the case when
a Remora-based application is integrated with heteroge-
neous systems with a different set of software and hardware
settings.
4.1 Basic Concepts
We propose a new key concept, which is considered as
a basis for our distribution proposal. This concept, called
Remora Distribution, refers to a new paradigm in component-
based software design for resource-constrained networks in
which every system functionality is encapsulated in a com-
ponent and distributed for the use of both homogenous and
heterogeneous systems. This indicates the capability of defin-
ing platform-specific bindings for Remora components in
order to transparently handle the communication issues in
WSNs. In this way, Remora will automatically generate
part of the code which is required for sending data over
the network or invoking methods, instead of time-consuming
and error-prone programming with low-level APIs.
Generally, the Remora bindings are categorized into two
classes: remote binding and interoperable binding. The for-
mer specifies a type of remote service call occurring within
a sensor network between two homogenous sensor nodes in
order to send the required service data from one node to the
other node (like Java RMI), while the latter refers to the
invocations happening beyond a sensor network between a
sensor node and a node in an existing IT system, such as the
Internet. To concretely describe the Remora distribution,
we first discuss the principles underpinning this framework,
including:
Access Transparency. From the programmer’s point of
view, local and distributed services should be accessed us-
ing identical operations. Additionally, she/he should not be
forced to write distribution-related code that is out of scope
of application logic.

Location Transparency. The distribution mechanism should
enable Remora components to access remote services with-
out any knowledge of their location in the network.
Synchronisms. The calling mode can be either synchronous
or asynchronous corresponding to the mode supported by
the network protocol, e.g., RESTful calls are synchronous,
while remote calls over RIME—a WSN-specific communi-
cation protocol implemented by Contiki [7]—are handled
asynchronously.
Pluggability of Bindings. Remora bindings should be
maintained in a well-structured way in the sense that ev-
ery binding-related library for a particular network protocol
should be easily pluggable to or unpluggable from services
provided by Remora components without affecting other
parts of the system.
Figure 3 depicts the overall architecture of the Remora
distribution model. Each component of the system exhibits
a set of services, which can interact in two different ways: lo-
cally or remotely. In the local mode, service calls are carried
out through a simple and lightweight node-level invocation
plan, while the distributed calls occur either remotely (be-
tween two nodes with the same set of configurations), or in
an interoperable manner (between a Remora-enabled node
and an IP-enabled node).
!
!
!
Remora Runtime
OS
Remora Runtime
OS
OS
Sensor Node
Sensor Node
IP-enabled Node
Remora
Component
Non-Remora
Component
Service
Reference
Local Call
Remote Call
Figure 3: Architecture of the Remora distribution
model.
While the development of each main type of Remora
binding would require consideration of many kinds of chal-
lenges, such as programming constructs, streaming, net-
working, and runtime support, the contribution of this pa-
per is limited to providing service-level integration of sensor
nodes with IP-enabled systems. We therefore leave the is-
sues of the Remora remote binding as our future work and
in this paper we study the Remora interoperable binding
in the context of Web integration. Remora Web binding is
basically concerned with integrating sensor services with IP-
enabled networks by translating the service requests to the
equivalent RESTful web services. In this way, the distribu-
tion mechanism of Remora follows a standard and widely-
adopted protocol to link WSN services to any Web-enabled
device in ubiquitous environments.
4.2 REMORA Web Services
In this section, we first describe the principles of the REST
architectural style and RESTful Web services, and then present
our component-based approach for integrating Remora ser-
vices with the Web.
REST Principles. The REpresentational State Transfer
(REST), as coined by Fielding [9], is an architectural style
for distributed systems emphasizing scalability of compo-
nent interactions, generality of interfaces, and independent
deployment of components. The “resource-oriented” prin-
ciples of REST are described through the REST triangle
defining the principles for addressing, accessing and encod-
ing a collection of resources using the Internet standards. A
distributed system, which follows REST principles, is called
RESTful, e.g., the Web. In a RESTful system, a component
can interact with other distributed components by knowing
two things: i) the unique identifier of the representative re-
source of component, and ii) the predefined standard oper-
ations to invoke (GET, POST, PUT and DELETE). In this
model of interaction, the client-server separation of concerns
can simplify component implementation, reduce the com-
plexity of connector semantics, and increase the scalability
of server components.
RESTful Web services. Web services are a set of stan-
dards and techniques for developing interoperable distributed
applications that are accessed via HTTP and executed on
a remote system hosting the requested services. Nowadays,
Web services are generally categorized into two groups: SOAP-
based Web services and RESTful Web services. The former
follows the Simple Object Access Protocol (SOAP), which is
a heavyweight protocol specification for exchanging struc-
tured information in the implementation of Web services,
while the latter is a lightweight model based on REST, which
does not impose SOAP or XML.
We believe that the simplicity and uniform interfaces of
RESTful Web services is an efficient approach to integrate
any individual sensor node to any RESTful system in perva-
sive environments. Additionally, there have been reported
a number of valuable works focusing on the integration of
REST and WSNs and providing the primitives required to
RESTful programming in sensor systems [26]. Therefore, we
adopt this approach and study how to enable RESTful Web
service development in WSNs by the Remora component
model.
Since an application built from REST principles is trans-
formed from operation-centric into a data-centric model, ev-
ery entity that offers a service becomes a resource (e.g., a
temperature sensor) that can be identified unambiguously
using a Uniform Resource Identifier (URI) [22]. Every re-
source then defines a uniform interface just including four
main operations provided by REST (GET, POST, PUT
and DELETE). Therefore, incorporating REST principles
into Remora requires applying the REST triangle of nouns,
verbs, and content types to the specification of Remora
component model.
Remora Service Identifier. The service identifier is a
unique noun described using the URI format. Therefore,
service identifiers include a server address, a service path,
and a sequence of request parameters:
/server-address/service-path?request-params
Since Remora components in a sensor application are ba-
sically organized in a hierarchical model (like Java packages),
the URI of a distributed service is corresponding to the hier-
archical organization of Remora components within the ap-
plication. For instance, assume that Sensors is a component
providing a service for temperature sensor (myTemperature)
and another service for light sensor (myLight), thus the ser-
vices of the Sensors component, located under /app/peripheral,
can be addressed as follow:
/node-id/app/peripheral/Sensors/myTemperature
/node-id/app/peripheral/Sensors/myLight
As the Remora engine has a complete knowledge about
the structure of applications (including hierarchical organi-

zation of system components), service identifiers are auto-
matically obtained by the Remora engine and therefore the
programmer does not need to specify any identifier for dis-
tributed services during the application development. Nev-
ertheless, we have provided a special REST binding tag
(binding.rest) through which the programmer can change
the URI of a component, as well as its services. Figure 4
shows the description of the Sensors component, where the
URI of the myTemperature service is changed. Therefore,
the new URI of the myTemperature service becomes:
/node-id/app/myTemp
Figure 4. RESTful Service identification in R
<componentType name="app.peripheral.Sensors”>
<service name="myTemperature">
<interface.remora name="api.ITemperature"/>
<binding.rest uri="/app/myTemp"/>
</service>
<service name="myLight">
<interface.remora name="api.ILight"/>
</service>
...
</componentType>
Figure 4: RESTful service identification in Remora.
Remora Service API. The four main operations of REST
can be mapped to the operations implemented by a particu-
lar Remora service. Figure 5 shows an excerpt of the inter-
face ITemperature. Similar to the Remora services, opera-
tions of an interface can be considered as a REST resource
and have their own URI. Depending on the functionality of
a service and variety of its operations, we may need to de-
fine a new URI for a collection of service operations. The
other important part of the REST binding tag is that we
need to provide the name of the equivalent REST operation
for each operation. For example, the current temperature
of environment can be retrieved from a temperature sensor
node using the HTTP request:
GET http://device.uio.no:8080/node-id/app/myTemp
which returns the current temperature. Temperature con-
figuration information can also be pushed into a sensor node
by using an HTTP PUT request as follow:
PUT http://device.uio.no:8080/node-id/app/
myTemp/threshold
In this case, the HTTP request is sent to the operation
setThreshold, along with the required parameters, then the
Remora framework will call the setThreshold operation.
Note that in the sample code of Figure 5, threshold is con-
sidered as a new separate REST resource with its own URI.
Content Type. Finally, for a REST binding we can specify
the format of data delivered to the client. For instance,
according to the REST annotation defined for getCurrent
in the ITemperature interface, this operation provides the
current temperature in the JSON format (cf. Figure 5).
4.3 A Concrete Use Case
An example use case that can benefit greatly from our dis-
tribution framework is home monitoring systems. Such ap-
plications are characterized as being filled with sensor nodes
Figure 5. Mapping REST verbs to the equivalent R
<interface.remora name="api.ITemperature">
<operation name="getCurrent" return="xsd:short">
<annotation.rest method="GET"
type="application/json"/>
</operation>
<operation name="setThreshold" return="void">
<in name="threshold" type="xsd:short"/>
<annotation.rest method="PUT" path="/threshold"/>
</operation>
...
</interface.remora>
Figure 5: Mapping the REST verbs to equivalent
Remora operations and identifying content types.
to observe various types of ambient context elements (tem-
perature, smoke, occupancy, and health conditions of in-
habitant), actuators to physically control home appliances
(lights, TV, and air conditioning), and smart phones to pro-
vide information about the preferences of owners.
Integration of multi-scale entities is one of the main chal-
lenges in such an environment. Mobile devices and sen-
sors have different hardware and software capabilities, which
make some devices more powerful than others. Therefore,
this heterogeneity requires a flexible and simple solution that
supports multiple interaction mechanisms and considers the
restricted capabilities of some devices. In particular, regard-
ing sensor nodes, the immaturity of high-level communica-
tion protocols, as well as the inherent resource scarceness,
bring a critical challenge to the system: how to connect sen-
sor nodes to mobile devices and actuators through a stan-
dard high-level communication protocol. We believe that
the RESTful framework presented above can significantly
smooth the way to integrate a sensor network with actuators
and existing infrastructure networks in the home monitoring
systems.
5. IMPLEMENTATION
The implementation of the proposed model consists of
two main parts according to the model’s architecture. The
first part is concerned with modifying the implementation
of Remora specifications in order to support the new com-
mands related to Web services programming, such as the
REST binding tag. The second part is dedicated to devel-
oping a middleware framework taking care of REST commu-
nication issues. The former is performed within the imple-
mentation of the Remora engine. In addition to enhancing
the engine to support the new commands, it should stati-
cally create the data structures required for maintaining the
REST-related service data, such as resource identifiers and
Service API information.
The middleware framework, supporting Web service-based
communications, is a lightweight module integrated with the
Remora runtime and the Remora framework to support
runtime requirements of the proposed model. To this end,
we first need to choose and exploit an existing library that
is flexible and completely support the REST principles. We
adopt the REST framework presented by Yazar et al. [26]
for two reasons. Firstly, they have used the Contiki oper-
ating system to develop the RESTful architecture and our
current Remora runtime is also available on Contiki. Sec-
ondly, this framework’s architecture along with Contiki’s
extensive work on the IP protocol stack provide a set of
complete and easy-to-use low-level libraries for RESTful en-

hancements.
Figure 6 depicts the overall implementation architecture of
Remora Web services, where Contiki’s core and TCP/IP
stack lie in the bottom. The gray box represents the REST
framework and contains core modules required for REST-
ful Web services development over Contiki. The Remora
runtime is integrated with the REST framework through
the REST Wrapper API. It should be noted that Wrapper
API is one of the main features of Remora, providing a
well-described method for integrating a Remora applica-
tion with underlying system software [23].
Distributed Remora Application
Remora Runtime
REST Framework
HTTP Server REST
Engine
XML
Parser
uIP (TCP/IP Stack)
REST Wrapper API
REST Broker
Contiki Core
Figure 6: Overall implementation architecture of
Remora Web services.
Figure 7 shows an excerpt of the REST Wrapper API,
including operations for initializing REST engine, setting
the representation type for a HTTP connection through the
Remora component HTTPConnection, sending GET data,
receiving POST data, and updating the status of a connec-
tion. The implementer of this interface is a Remora com-
ponent in which the low-level APIs of the REST framework
are invoked.
Figure 8. An excerpt of the REST Wrapper API
<interface.remora name="api.wrapper.IREST">
<operation name="initialize"/>
<operation name="setReperType">
<in name="httpConnection"
type="rest.comm.HTTPConnection"/>
<in name="type" type="xsd:int"/>
</operation>
<operation name="setGETData">
<in name="data" type="xsd:string"/>
</operation>
<operation name="getPOSTData" return="xsd:string"/>
<operation name="setHTTPStatus">
<in name="type" type="xsd:int"/>
</operation>
...
</interface.remora>
Figure 7: An excerpt of the REST Wrapper API.
REST Broker contains a set of Remora components pro-
cessing REST requests received from Remora runtime. Specif-
ically, it is an intermediate module for handling the REST
requests received from a Web client or sent from the sensor
node to a node hosting RESTful Web services. The broker is
also in charge of retaining the list of application-specific re-
sources and the corresponding Remora Web services APIs.
6. PRELIMINARY EVALUATION
As mentioned before, we adopt Contiki as our system
platform and our hardware platform is the popular TelosB
mote equipped with a 16-bit TI MSP430 MCU with 48KB
ROM and 10KB RAM. In the preliminary evaluation, we
assess the performance of the system based on the memory
overhead incurred by the REST Broker and the REST wrap-
ping component (implementer of the REST Wrapper API)
on ROM and RAM.
The memory footprints are categorized into a fixed over-
head and a dynamic overhead. The fixed overhead is the
minimum additional memory required for a distributed ap-
plication, regardless of the number of distributed services
running within the application. Table 1 shows the fixed
memory requirements, which turn out to be quite reason-
able with respect to both code and data memory. As seen
from the table, the main memory consuming module is the
REST framework. Therefore, the fixed memory cost can
be further reduced by switching to a more efficient REST
framework in the future.
Table 1: The fixed memory requirement of Remora
Web services framework.
Code Data
Memory Memory
Module (bytes) (bytes)
Remora Runtime 494 14
REST Wrapper 94 0
REST Broker 252 8
REST Framework 4668 76
Total 5508 98
The dynamic memory overhead is calculated based on the
number of distributed resources (services, components or op-
erations) in an application and the number of REST verb
mappings (GET, POST, PUT and DELETE) for each re-
source. The memory overhead of the former is variable ac-
cording to the length of a resource’s name, while the latter
consumes 2 bytes of ROM to retain the starting memory
address of a REST operation.
7. CONCLUSION AND FUTURE WORK
Distributing WSN services through standard and widely-
accepted communication protocols is of high importance.
In this paper, we presented a high-level programming ab-
straction in order to enable service distribution in WSN ap-
plications and relieve the programmer from the burden of
dealing with low-level APIs for developing distributed sen-
sor services. This component-based approach promises a
new abstraction for integrating sensor software modules to
the Internet through upgrading component-level services to
Web services over a lightweight RESTful architecture. This
flexible framework is also potentially able to exhibit sensor
services to other types of network protocols by implement-
ing platform-specific bindings. Our future work includes ad-
dressing the Remora remote binding, occurring when com-
ponent services within a sensor network need to remotely
communicate via a particular sensor network protocol.
8. ACKNOWLEDGMENTS
This work was partly funded by the Research Council
of Norway through the project SWISNET, grant number
176151.

9. REFERENCES
[1] I. F. Akyildiz and I. H. Kasimoglu. Wireless sensor
and actor networks: research challenges. Ad Hoc
Networks, 2(4):351–367, 2004.
[2] P. Buonadonna, J. Hill, and D. Culler. Active Message
Communication for Tiny Networked Sensors. In
INFOCOM ’01: Proceedings of the 20th Annual Joint
Conference of the IEEE Computer and
Communications Societies, Alaska, USA, 2001. IEEE.
[3] M. Ceriotti et al. Monitoring heritage buildings with
wireless sensor networks: The Torre Aquila
deployment. In IPSN’09: Proceedings of the 2009
International Conference on Information Processing in
Sensor Networks, pages 277–288, San Francisco, CA,
USA, 2009. IEEE.
[4] G. Coulson et al. A generic component model for
building systems software. ACM Trans. Comput.
Syst., 26(1):1–42, 2008.
[5] L. de Souza, P. Spiess, D. Guinard, M. Köhler,
S. Karnouskos, and D. Savio. SOCRADES: A Web
Service Based Shop Floor Integration Infrastructure.
In The Internet of Things, volume 4952 of LNCS,
pages 50–67. Springer, 2008.
[6] A. Dunkels, B. Gronvall, and T. Voigt. Contiki - a
lightweight and flexible operating system for tiny
networked sensors. In LCN ’04: Proceedings of the
29th Annual IEEE International Conference on Local
Computer Networks, pages 455–462, Tampa, FL, USA,
2004. IEEE Computer Society.
[7] A. Dunkels, F. Österlind, and Z. He. An Adaptive
Communication Architecture for Wireless Sensor
Networks. In SenSys’07: Proceedings of the 5th
international conference on Embedded networked
sensor systems, pages 335–349, Sydney, Australia,
2007. ACM.
[8] D. Estrin, D. Culler, K. Pister, and G. Sukhatme.
Connecting the Physical World with Pervasive
Networks. IEEE Pervasive Computing, 1(1):59–69,
2002.
[9] R. Fielding. Architectural Styles and the Design of
Network-based Software Architectures. University of
California, Irvine, USA, 2000. PhD thesis.
[10] N. Gershenfeld, Raffi, R. Krikorian, and D. Cohen.
The Internet of Things. Scientific American, pages
76–81, 2004.
[11] D. Guinard, V. Trifa, T. Pham, and O. Liechti.
Towards physical mashups in the web of things. In
INSS’09: Proceedings of the 6th International
Conference on Networked Sensing Systems, pages
196–199, Pittsburgh, PA, USA, 2009. IEEE.
[12] J. Hill, R. Szewczyk, A. Woo, S. Hollar, D. Culler, and
K. Pister. System Architecture Directions for
Networked Sensors. SIGPLAN Not., 35(11):93–104,
2000.
[13] D. Hughes et al. LooCI: a Loosely-Coupled
Component Infrastructure for Networked Embedded
Systems. In MoMM’09: Proceedings of the 7th
International Conference on Advances in Mobile
Computing and Multimedia, pages 195–203, Kuala
Lumpur, Malaysia, 2009. ACM.
[14] H. Liu, M. Bolic, A. Nayak, and I. Stojmenovic.
Taxonomy and Challenges of the Integration of RFID
and Wireless Sensor Networks. IEEE Network,
22(6):26–35, 2008.
[15] T. Luckenbach, P. Gober, K. Kotsopoulos,
Andreas Kim, and S. Arbanowski. TinyREST: a
Protocol for Integrating Sensor Networks into the
Internet. In REALWSN’05: Proceedings of the
Workshop on Real-World WSNs, Stockholm, Sweden,
2005.
[16] M. Marin-Perianu et al. Decentralized Enterprise
Systems: a Multi-Platform Wireless Sensor Network
Approach. Wireless Communications, IEEE,
14(6):57–66, 2007.
[17] A. Milenković, C. Otto, and E. Jovanov. Wireless
sensor networks for personal health monitoring: Issues
and an implementation. Computer Communications
(Special issue: Wireless Sensor Networks:
Performance, Reliability, Security, and Beyond),
29:2521–2533, 2006.
[18] OSOA. The Service Component Architecture.
http://www.oasis-opencsa.org/sca.
[19] N. B. Priyantha, A. Kansal, M. Goraczko, and
F. Zhao. Tiny Web Services: Design and
Implementation of Interoperable and Evolvable Sensor
Networks. In SenSys’08: Proceedings of the 6th ACM
Conference on Embedded Network Sensor Systems,
pages 253–266, Raleigh, NC, USA, 2008. ACM.
[20] E. Souto, G. Guimar aes, G. Vasconcelos, M. Vieira,
N. Rosa, and C. Ferraz. A message-oriented
middleware for sensor networks. In MPAC’04:
Proceedings of the 2nd Workshop on Middleware for
Pervasive and Ad-hoc Computing, pages 127–134,
Toronto, Canada, 2004. ACM.
[21] C. Szyperski. Component Software: Beyond
Object-Oriented Programming. Addison-Wesley
Longman Publishing Co., Inc., Boston, MA, USA,
2002.
[22] R. T. Berners-Lee and M. Fielding, L. Uniform
Resource Identifier (URI): Generic Syntax, 2005.
http://www.ietf.org/rfc/rfc3986.txt.
[23] A. Taherkordi, F. Loiret, A. Abdolrazaghi, R. Rouvoy,
Q. L. Trung, and F. Eliassen. Programming Sensor
Networks Using Remora Component Model. In
DCOSS’10: Proceedings of the 6th International
Conference on Distributed Computing in Sensor
Systems, pages 45–62, Santa Barbara, CA, USA, 2010.
Springer.
[24] UPnP. Universal Plug and Play.
http://www.upnp.org.
[25] G. Werner-Allen, J. Johnson, M. Ruiz, J. Lees, and
M. Welsh. Monitoring volcanic eruptions with a
wireless sensor network. In EWSN’05: Proceedings of
the Second European Workshop on Wireless Sensor
Networks, pages 108–120, Istanbul, Turkey, 2005.
[26] D. Yazar and A. Dunkels. Efficient Application
Integration in IP-based Sensor Networks. In
BuildSys’09: Proceedings of the First ACM Workshop
on Embedded Sensing Systems for Energy-Efficiency in
Buildings, pages 43–48, Berkeley, CA, USA, 2009.
ACM.

A Component-Based Approach For Service Distribution In Sensor Networks

More Related Content

Similar to A Component-Based Approach For Service Distribution In Sensor Networks (20)

More from Kim Daniels (20)

Recently uploaded (20)

A Component-Based Approach For Service Distribution In Sensor Networks