Service Composition Approaches for IoT Research Paper I would like to paraphrase section 2,3,4,5 and 6 from the attached research paper. The word count mus

Service Composition Approaches for IoT Research Paper I would like to paraphrase section 2,3,4,5 and 6 from the attached research paper. The word count must be between 5000 to 6000 without excluding any details and with all references which are mentioned in these sections. See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/330053078
Service composition approaches for Internet of Things: a review
Article in International Journal of Communication Networks and Distributed Systems · January 2019
DOI: 10.1504/IJCNDS.2019.10017271
CITATION
READS
1
451
4 authors:
Idir Aoudia
Saber Benharzallah
Université de Biskra
University of Batna 2
2 PUBLICATIONS 1 CITATION
49 PUBLICATIONS 64 CITATIONS
SEE PROFILE
SEE PROFILE
Laid Kahloul
Okba Kazar
Université de Biskra
Université de Biskra
100 PUBLICATIONS 235 CITATIONS
319 PUBLICATIONS 346 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Reconfigurability View project
web services composition based MAS View project
All content following this page was uploaded by Idir Aoudia on 03 November 2019.
The user has requested enhancement of the downloaded file.
SEE PROFILE
194
Int. J. Communication Networks and Distributed Systems, Vol. 23, No. 2, 2019
Service composition approaches for internet of
things: a review
Idir Aoudia*
LINFI Laboratory,
Biskra University, Algeria
Email: idir.aoudia@yahoo.fr
*Corresponding author
Saber Benharzallah
LINFI Laboratory,
Biskra University, Algeria
and
Computer Sciences Department,
Batna 2 University, Algeria
Email: sbharz@yahoo.fr
Laid Kahloul and Okba Kazar
LINFI Laboratory,
Biskra University, Algeria
Email: kahloul2006@yahoo.fr
Email: kazarokba@yahoo.fr
Abstract: The internet of things (IoT) is the integration of information space
and physical space, becoming more and more popular in several places. In
order to answer the complicated user request, most IoT devices do not work
alone, a multiple service composition must be made, and it is defined as the
services composition. A number of approaches to IoT’s service composition
have been proposed in the literature. We have noticed a lack of studies that
review the different works of the service composition in IoT. Thereby, in this
paper, we present a review of existing approaches for service composition in
IoT. It describes and compares them among each other with respect to some
criteria, which we have defined based on the different challenges of service
composition in IoT. Finally, we conclude by discussing open research issues
and avenues for further work.
Keywords: internet of things; IoT; IoT service composition; adaptability;
context; smart object; resource constraints; IoT service composition challenges.
Reference to this paper should be made as follows: Aoudia, I.,
Benharzallah, S., Kahloul, L. and Kazar, O. (2019) ‘Service composition
approaches for internet of things: a review’, Int. J. Communication Networks
and Distributed Systems, Vol. 23, No. 2, pp.194–230.
Copyright © 2019 Inderscience Enterprises Ltd.
Service composition approaches for internet of things
195
Biographical notes: Idir Aoudia is a PhD student at LINFI Laboratory, Biskra
University 07000, obtained his Master 2 SIM degree in 2015 from Batna
University (Algeria). His current research interests include web services,
service composition and internet of things.
Saber Benharzallah is an Associate Professor and a Researcher in the Computer
Science Department of Batna 2 University (Algeria). He received his PhD
degree in 2010 from Biskra University. He is currently the team leader of
semantic interoperability and their applications in information systems and web
services, in LINFI Laboratory (Biskra University). His current research
interests include semantic interoperability and their applications, semantic web,
web services, context management and internet of things.
Laid Kahloul received his PhD in Computer Software and Theory from the
Computer Science Department, Biskra University, Biskra, Algeria, in 2012.
Currently, he is an Associate Professor with the Computer Science Department,
Biskra University, Algeria. His current research interests include Petri nets,
high level Petri nets and software engineering.
Okba Kazar is an Associate Professor at the Computer Science Department of
Biskra which he helped create and director of LINFI Laboratory at the
University of Biskra. He is the author of numerous publications, member of
international conference program committees and editorial board in various
journals. His research interests are artificial intelligence, multi-agent systems,
web applications and information systems.
1
Introduction
The internet of things (IoT) is an emerging technology, which can alter the industry,
environment, social and medical field. IoT is a fusion between information space and
physical space, known as the interconnection of embedded computing, sensors, mobile
devices, or other uniquely identifiable objects, which are able to interact with each other
and cooperate with the surrounding environment to reach common goals, by exploiting
the existing internet infrastructure. With the rapid development of microchips, sensor
devices, networks and software in recent years, IoT is becoming more and more popular
in several applications. An example is smart city, which are an extension of smart
buildings that are themselves an extension of smart home, which is one of the hottest
applications in people’s daily life.
IoT consists to connect a large number of daily objects to the internet, giving them
their own identity, allowing them, among others, to offer functionalities and to collect
information in the form of a service.
In order to answer the complicated user request, most IoT devices do not work alone,
a multiple service composition must be made, and it is defined as the services
composition. The services composition consists in combining the functionality of
multiple services within a single process in order to answer complex demands that a
single service could not satisfy (Berardi et al., 2003). The service composition provides
the ability to perform complex activities from existing services, and the composed
services form a new service, which can be reused in another composition.
196
I. Aoudia et al.
Actually, the research community active on the service composition in IoT is still
highly fragmented, and, most are focused on single application areas or unique
technologies. We believe that this fragmentation is potentially harmful for the
development of IoT’s service composition technologies. Also, there is a lack of studies
that review the different works of the service composition in IoT.
In this survey article, we aim at presenting a review of existing approaches for IoT’s
service composition, including a description and comparison between them with respect
to some key requirements. This paper also presents a comparison between traditional web
service composition and IoT service composition.
The main objective is to give the reader the opportunity of understanding what has
been done (protocols, algorithms, proposed solutions) and what still remains to be
addressed, as well the paper represents a support for researchers to focus on their efforts
and to deliver lasting solutions in this field.
We hope that this review paper will be able to help to fill the existing communities, to
encourage cross-collaborations and ensure that the challenges related to service
composition in the IoT will be better addressed, so that research can be exploited
successfully.
In summary, the main contributions of this paper are:
1
Identifies research problems by introducing some specifics criteria, which are based
on the fundamentals challenges of service composition.
2
An exhaustive comparative study between the IoT service composition approaches
based on a set of well-focused and specific criteria.
3
It aims to examine research trends and suggest future directions. In doing so, we
intend to emphasise service composition in IoT.
4
It summarises the current state–of–the–art of service composition in IoTs.
The remainder of this paper is organised as follows: in Section 2, we introduce some
challenges of IoT and then in Section 3, we describe some web service composition
standard and their application in IoT. In Section 4, we introduce specific criteria, which
will be the basis of our study. In addition, we present different approaches, already,
proposed for service composition in the IoT. Finally, in this section we propose a
comparative study of these approaches based on specific criteria cited before. Section 5
presents a discussion about our study, while in Section 6, we review the related work.
Finally, we conclude the paper by discussing open research issues and avenues for further
work in Section 7.
2
Challenges of service composition in IoT
Service composition in IoT has many challenges, IoT’s services are mainly deployed in
resource-constrained devices with limited energy storage and processing capability and
that is comparing to internet services which are deployed in computers with rich resource
and powerful function (Yang and Li, 2015). Besides, with the rapid increasing number of
devices, the composition must be designed from centralised to distributed, and the cost of
service composition in IoT needs to be considered.
Service composition approaches for internet of things
197
IoT environment is known as dynamic, the information of IoT’s service is instable (its
state switches between availability/unavailability) comparing to internet services (Yang
and Li, 2015). In addition, due of the independence of each entity, problems of
interaction and communication can appear, besides the probability of errors such as
failures and service changes in IoT is high. A monitoring mechanism should be used to
ensure fault tolerance and to monitor the IoT system; however, this can lead to a high
cost (Yang et al., 2014a).
Due to the dynamic evolution of user needs and context, the composition must be
adaptive and make a real-time update, to ensure continuity of services to the user and
improve the user interaction with the environment.
It is suitable for IoT’s service composition system to introduce a non-functional
parameters strategy as quality of service (QoS) to choose and compose sensory
information service, and that is because, IoT information service composition aims to
provide user with sensory information (Yang and Li, 2015). Contrary to internet service
composition, who aims to solve programmatic issue, focuses on the demonstration of
feasibility and logicality.
An IoT system connects the physical world in cyberspace (Chen et al., 2014), thus
one of the challenges of the IoT’s service composition system is the management of
protection and security system to resist malicious attacks in order to survive in hostile
environments.
Among the above challenges, the following three aspects are considered the most
difficult:
•
finding the most optimal composition, by using for example non-functional
parameters to improve the behaviour of the composition and its components
•
protecting and monitoring the devices with restricted monitoring resources in
dynamic environment
•
managing dynamic change of the environment and resources.
3
The application of standard web services composition on IoT’s services
composition
There are two modes to compose web services, the mediation mode which follows the
orchestration model, where all global dependencies are known for at least one service
(called the mediator) before execution. On the other side we talk about the P2P mode
which follows the model of choreography.
First of all, in service-oriented architecture (SOA), the services composition requires
that services must be described in a standard way in a directory, and offer rich, accurate
and subject exchange formats (or encapsulation of exchange) submitted to strict rules but
easily scalable, to allow the exploitation of available resources. As far as, this approach is
meaningful in the construction of a service, which will be offered to very numerous users,
as in IoT relationships evolve, and the organisation is brought to be rethought. The IoT’s
applications involve personal elements, which have to answer individual problems. The
intervention of an architect is then required for the construction of a solid aggregation of
services aimed the widest population of users as possible. Each IoT’s application may be
198
I. Aoudia et al.
used only by its designer and his family, because it is adapted more specifically to their
use, and/or some of their objects. In IoT, the circuit designer/consumer is reduced, and
sometimes these roles are played by the same person (Cherrier, 2013). In IoT and highly
constrained environments, it seems appropriate to disqualify simple object access
protocol (SOAP), including its heaviness to the service it is supposed to make. Most of
the current operations can be performed through HTTP, more simply, lighter and
concisely. Brevity and lightness of exchanges are exactly what IoT requires, and that is
what RESTful (Fielding and Taylor, 2002) approach offers.
The choreography and the orchestration of services are two different approaches that
we wonder about their recoveries; if they express the same rules and satisfy the same
needs. A study (zur Muehlen et al., 2005) was interested in the use of choreography under
SOAP and REST, and compares their strengths and weaknesses facing the workings of
the choreography. REST provides a weak coupling and clarity of the definition of the
desired operations that appear in the message, while SOAP is used to describe the
complexity of certain transactions and provides monitoring of exchanges, particularly
interesting to debug the cinematic dialogue. The approach choreography is adapted to
IoT; it allows reducing the use of network resources involved in IoT. The choreography
respects the energy constraints, because the direct interactions between nodes require less
network then data feedback to a central point. In addition, the processing capabilities
added to objects (admittedly limited, but present) allow the implementation of the
algorithms describing the task to be performed on the object itself (Cherrier, 2013).
Practice shows that the direct application of standard web services on IoT’s services
do not fit and will bring new problems; because IoT’s services are different from
traditional services and their environment is dynamic. The biggest difference between
traditional and IoT’s services is that traditional service is a virtual entity but IoT’s
services are directly related to the physical world (Li and Jiang, 2012), providing services
that satisfies the user and that constitute one of colossal problematic of IoT.
4
Existing approaches for service composition in IoT
This section discusses some popular proposed approaches that can be considered to
compose IoT’s services. Then in the second part, we propose a comparative study of
these proposals approaches. We started our investigation by introducing some specifics
criteria, which are based on the fundamentals challenges of service composition cited
before in Section 2.
•
Dynamic composition: the composition can take place during the execution of the
application to integrate new features and functionalities, following the dynamic
change in the environment and resources.
•
The adaptation: following the dynamics changes in the environment, resources and
the evolution of users’ needs and the context, the composition can take place during
the execution of the application to integrate new functionalities, and the system must
make a real-time update and therefore a decision on the composition choices can be
made during the execution in agreement with this. The adaptation criterion is based
on the following sub-criteria:
Service composition approaches for internet of things
1
2
3
4
5
199
type (moment): there is an offline adaptation technique (during the
decomposition and precision of the user’s goal, or during selection abd
discovery) and online adaptation technique (made on execution)
the technique used in the adaptation system
the deployment places of the adaptation system
the description: the behaviour of the adaptation system after detecting
unaccepted changes
the optimisation used to improve the efficiency of the adaptation system.
•
Independence and extensibility: the composition’s entities must be described
independently without assuming the existence or absence of the others. The
composition mechanism must also allow the addition and removal of entities at
runtime, so the composition must be recalculated automatically.
•
Automatic identification and resolution of failure and interaction problems: because
of the independence of each entity, problems of interaction and communication can
appear (conductivity, breakdowns, battery charge and others …); this will disrupt the
result of the composition. The probability of errors such as failures and service
changes in IoT is high. If an exception occurs, the system will be able to recompose
the service composition so that the new composition may likely work with different
services without failure.
•
Distributed and decentralised composition: the execution of the composition must be
done in a decentralised and distributed way, due to the increasing number of devices,
as well as the resource constraints (energy and processing capacity) of each device.
•
Trust, protection and security protocol: an IoT system evolves with new nodes and
existing nodes, the entities of the IoT system are mainly carried by humans or
human-operated devices, the system consists essentially of uncensored IoT devices
offering a wide variety of services, many of them (owners) will be malicious for their
own gain, so a trust management protocol is needed to resist malicious attacks and
survive in hostile environments.
•
The optimisation: the optimisation uses non-functional parameters to improve the
behaviour of the composition and its components. Although satisfying functional
requirements is important for building the system, optimising non-functional
preferences can be equally crucial. For example, a composition, which minimises
response times and costs, and ensures availability, is required. These parameters can
be dynamic or static. Static parameters are generally known at the deployment time
and are usually not updated during the execution time. However, the dynamic
parameters represent the variable characteristics of a given service such as the
response time, energy level, availability and their values are provided at the service
invocation time by a monitoring process.
•
The performance: the process of composition requires calculations. These
calculations are required to compose a service which satisfies the purpose of the
user, this could lead to a large consumption of time, due to the number of services
available in the IoT which increases enormously.
•
Results obtained: evaluation and experimental results for each proposed approach.
200
I. Aoudia et al.
•
The model used: which model is used for modelling the relationships between IoT
services.
•
The standards and protocol used in the service composition system.
•
The representation of services: which architectural style is used for service
representation in IoT system.
4.1 Overview of proposals
In this part, we are going to present several approaches already proposed for service
composition in the IoT.
In Chen et al. (2015), a distributed social network-based approach for IoT device
management and service composition has been proposed. The authors encapsulated IoT
devices into web services using RESTful style; they modelled the relationships between
IoT services by social networks, and classified it into three dimensions which are
location, type and correlation. Schemes of service registration, selection and management
are carefully designed for each of these three dimensions, and thus the social network is
divided into thr…
Purchase answer to see full
attachment