An Agentified Use of the Internet of Things

Abstract

The massive use of connected objects and the development of the Internet of Things leads us to ask questions about the way of use and the intelligence of these objects. First, we will present the main lines of Internet of Things research related to the intelligence. Then, we will highlight some situations showing the boundaries of the different axes of research and present the advantage of using the software agent paradigm. Finally, we will our vision of a smart use of connected objects by merging with the concept of software agents to expose directly the functionalities and capacities of the objects through software agents in purpose to be more accountable to the sustainable development.

Full text

An agentified use of the Internet of Things
Joël Kwan
LIM – Université de la
Réunion
Sainte Clotilde, La Réunion
joel.kwan@univ-reunion.fr
Yassine Gangat
LIM – Université de la
Réunion
Sainte Clotilde, La Réunion
yassine.gangat@univ-
reunion.fr
Denis Payet
LIM – Université de la
Réunion
Sainte Clotilde, La Réunion
denis.payet@univ-reunion.fr
Rémy Courdier
LIM – Université de la Réunion
Sainte Clotilde, La Réunion
remy.courdier@univ-reunion.fr
Abstract—The massive use of connected objects and the
development of the Internet of Things leads us to ask questions
about the way of use and the intelligence of these objects. First,
we will present the main lines of Internet of Things research
related to the intelligence. Then, we will highlight some situations
showing the boundaries of the different axes of research and
present the advantage of using the software agent paradigm.
Finally, we will our vision of a smart use of connected objects by
merging with the concept of software agents to expose directly
the functionalities and capacities of the objects through software
agents in purpose to be more accountable to the sustainable
development.
Keywords— Internet of Things; connected objects; collective
and adaptative system; virtual objects; software agents; sustainable
development; green IT
I. INTRODUCTION
With the development of ubiquitous computing,
particularly the domain of Internet of Things (IoT), there are
many opportunities in both researches and applications. In this
paper, we will focus on the intelligence of these connected
objects.
The main concept of the IoT is to connect all objects in the
physical environment to the Internet in order to expand it in the
physical world. The IoT is an evolution of the Internet for
interacting in the digital world from the physical world and
vice versa through exchanges between the devices on Internet
network [1][2]. It combines several aspects and technologies
such as pervasive computing, Internet protocol, sensors and
actuators technologies and the principle of embedded systems
that are all merged to form a system where the physical and
digital world meet and allow various interactions [3]. Thus, this
construction allows somehow to bring Internet into the physical
world.
At present time, there is a trend toward computerization and
integration of embedded devices in everyday objects (home
automation, wearables, smart multimedia devices and
appliances) [4]. Pervasive computing refers to the use of
increasingly widespread spontaneously and implicitly devices
communicating with each other through the physical or digital
interfaces that dues to their very small dimensions will be
integrated into the everyday objects, becoming almost invisible
to users [5].
More important, nearly all other devices are intended to
provide a smart behavior and virtually expose it through an
Internet connection. So, we have several connected objects that
provide data online as we can see in the Fig. 1. As we can see
in this figure the IoT is a cross-domain that may provide
support in many areas.
Fig. 1 Schema of R. Buyya representing Internet of Things Schematic
showing the end users and application areas based on data
The number of objects that compose the Internet of Things
are increasing over time [6][7]. This explosion of use results in
the mass production of devices and will have an impact on
energy consumption but also on the environment side. Besides
massive use of the batteries for powering these devices, the
objects will sooner or later be doomed to be replaced for reason
related to the product lifetime thus generating a huge amount of
electronic garbage. Each of us will have some responsibility in
terms of sustainable development and Green IT. It would be
interesting to reflect on how we can use the connected objects
to limit the redundancy of equipment and features while
keeping a certain level of comfort.
For understanding on how to improve the interaction
between connected objects and people, we will initially
concern ourselves with existing research in connected objects,
especially the intelligence. Then, we will see that many of
these researches tend to explore in the domain of software
agents. Thus, based on the software agent paradigm, we will

propose a new way of conceptualizing the connected objects
and the way of use.
II. ARTIFICIAL INTELLIGENCE IN THE FIELD OF CONNECTED
OBJECTS
With the objective of improving the living comfort and
facilitating the use of such a technology environment where IT
is ubiquitous and pervasive, the user is immersed in a world
where there will be more boundaries between the digital world
and the physical world. Of course, to get to this point of
immersion, there are many requirements to fulfill.
Besides of the material and technical limits (connectivity,
battery life, enslavement, precision, miniaturization, etc.), we
have to improve the software side of such infrastructure in
order to have a consistent whole application giving the
impression of a globally unified software that inhibits the
hardware diversity and topology on which their operating is
based. The goal is to burst their physical wraps into a coherent
and diffuse entity that is totally transparent to the user.
With this aim in mind, we have to set up a form of complex
organization system between these connected objects
preferably without or with a minimal of central nodes. A code
unit represented as an entity would be embedded in different
physical elements that would allow to be interconnected and
take knowledge of the environment where they are deployed
without human intervention. These connected objects would be
dynamically configured regarding the profile or behavior that
fits better to the environment [8][9]. This would require that
these entities would be endow with proactivity when they are
alone or collectively. From a simplistic point of view, an entity
will have to find itself its place in the system where it evolves
and will have to contribute to the operating of the overall
system of the surrounding environment. We also have to take
in consideration that any entity will possibly be mobile or
moveable (e.g.: smartphone, presence sensors, wearables,
personal computer, drones, etc.).
Obviously deploying such kind of systems requires some
organization. To simplify this, we could make each entities
work as teams so that each team member can interact with
others members of a same team in a privileged way through
well compartmentalized interaction spaces. As another
viewpoint, each device could be part of one or several of these
privileged spaces of data exchanges. The possible mobility of
equipment implies that the system is so flexible that it can deal
with departures and arrivals of entities as node in the
infrastructure. These spaces must be adaptable for dynamically
accept comings and goings of new entities without human
interventions.
Thanks to the evolution of technology and the ease of
access to the materials, there are researches which are focused
on the current and future change in the domain of ubiquitous
computing and its association with others domains such as
multi-agent systems [10], learning systems [11], the
convergence between the digital and the physical world [5], or
wireless networks [12]. There are proposals with multi-
approaches in the area of IoT [13][14], introducing a first level
of reflection to link Collective and Adaptive Systems and the
Internet of Things and can be referred to expand these
proposals to the more generic domain of the connected objects.
In a similar vein, others teams are also trying to offer a solution
based on artificial intelligence and software agents to improve
the vision of ubiquitous computing [15].
Multi-agent models are exploited for engineering and
conception of complex and pervasive system [16] while
dealing with unforeseen events [17] and offers formalization
that take into account both collective management and
modeling centered on individuals to solve issues related to
complex systems [18]. Researchers and particularly researchers
in software agent community have attempted and proved that
complex issues can be resolved by using group of agents
working together. The power of self-organization capacity of
the agent based models permit to have highly dynamic and
steady architectures [19]. Moreover, by endorsing the self-
adaptation capacity to the agents, a phenomena of self-
adaptability will emerge on the system itself.
One of the big challenges of this design is to bring multi-
agent approaches to a physical and real-time dynamic and open
operating context that requires constant adaptability with a bare
minimum of actions from the users who are themselves
included in the system. Applying multi-agent systems to real
environment introduces as well new problems. In simulation,
the environment is more or less controlled by configuration and
initial states. In real situation, there are possibilities of
unknown events at any time.
There are papers questioning about the architecture of IoT
which try to expand the concept with a multi-agent approaches
in more or less precise context.
 Mzham and al. [14] propose a solution to augment the
IoT with intelligent software agents by introducing the
concept of Agents of Things. They have review the
architectures of IoT concept and identify limitations
and deficiencies. They end to propose an agent based
proposition resulting from a review concept software
agent architectures and benefits.
 Leppänen and al. [13] also propose an agent based
solution to resolve the continuous transition,
requirement of software adaptation and system
evolution. They have questioned about a method of
integration of autonomous smart objects with mobile
agents, with open standards for communication and
cooperation support without a specific middleware
solution.
 Ranganathan and al. [15] discuss about the design of
systems and application in ubiquitous environments to
take account of heterogeneous devices, mobile users
and rapidly changing contexts. So they focused on the
context-awareness of the ubiquitous agents which
allow them to be adaptive to different situations. They
describe their middleware for developing context-
aware applications based on a predicate model of
context using rules or machine learning approaches to
decide their behavior.
 In the same logic, Sadri and al. [20] also propose a
solution based on artificial intelligence and agent

technologies for ubiquitous computing. They also
highlight that such applications will play an important
role in the future.
While lot of works have been done to introduce agent
approaches in the area of ubiquitous computing and Internet of
Things, there are more or less hazy outlines, propositions and
concepts or partial solutions due to the youthful of the domains.
III. USE CASES
The driver of our use cases will be over the composition of
the physical connected objects found in a room (Fig. 2) where a
normal user would live. In this room, there will be:
 Open/close sensors on different doors for detecting
when doors are opened when the user is away
 A doorbell for notifying guest at the entrance.
 A fire alarm system in case of fire in the kitchen.
 An alarm for notifying unwanted intrusion.
 A presence sensor for detecting movement inside the
room.
 A personal computer.
Usually, there are the exact representation of the connected
objects in the digital world.
Fig. 2 Schema of a room equipped with connected objects
A. Finding an alternate for a faulty alarm
By staying in the category of the connected objects
available to the general public, there are often multi-sensor
devices such as the fire alarm system. In this kind of device,
there are at least a piezo buzzer working as the alarm, a LED
light for notifying the state of the battery, a gas/smoke sensor
and in some elaborate version a temperature sensor for
avoiding false positive. So we can notice that there are
potentially two other alarms (the doorbell, the fire alarm
system) in this room. If we can directly expose the features of
the fire alarm system, we could use the alarm inside of the
device as an alternative for the current alarm when there are
burglars or fire in the house in case of failure. The fact is that
the alarm feature of this deployed system is redundant, so the
presence of the alarm dedicated to intrusion notification is
optional.
Fig. 3 Alternative to a faulty alarm
B. Enhancing the comfort
Without adding new connected objects in this room, we
could also classically enhance the comfort of the user, the
presence sensor can also be used to activate lights or powering
on the TV as there are users in the room.
More important, the main objective of this paper is the fact
that the intelligence brought in the domain of Internet of
Things and generally the connected objects is the fact that we
could have the same level or greater level of comfort and
interaction without the obligation of multiplying the number of
devices around us. With the presence of a computer in this
room, this new inserted device in the environment could be
mixed with the available features. For example, the TV could
use the speakers of the computer for enhancing the quality of
the sound by passing from a 2.1 speaker to an equivalent of a
5.1 home cinema kit.

Fig. 4 New virtual connected object from existing physical connected objects
IV. OUR PROPOSITION
The motivation to propose a multi-agent approach in a
collective and adaptive perspective as a solution of the
requirements arises from technical and conceptual limitations
issued from current connected objects systems. There are
already trails proposed in several papers that attempt to explore
and outline the future of the IoT and ubiquitous computing.
There are many possibilities of enhancing the current state of
connected objects and ubiquitous computing. A new agent
based approach is to look at a large deployment of such system
in various fields mixed with a smart used of the existing
components inside each devices. This allows to identify the
common requirements to have the more adaptive and
robustness environment that could be used in several ways
regardless the hardware limitations and regarding the context-
awareness.
The main concept of IoT is to bring the physical object to
the digital world in order to manipulate it in a software
environment. Today’s trend is to perceive an object of IoT as a
service for easing the handling by the common user. We find
ourselves very quickly with a multitude of devices scattered in
the everyday environment. However, many of these devices
have unwanted redundancy capabilities or features in on
deployed system. Instead of exposing the physical presence of
a connected object composed of several components (actuators
and sensors) inside a software environment, it would be more
interesting and advantageous to go one level further and
directly expose the capabilities and features of the involved
physical object.
Usually, the users interact with the equivalent of the
connected objects on the software side. What prevents us to
create virtual objects and manipulate them as well as current
objects of Internet of Things? By breaking down the physical
object to a finer level of granularity, we will benefit from a
greater malleability and functionality of an existing
environment while allowing to keep an ease of use. Each object
or virtual objects will be represented as a software agent. On
top of the fact that the objects will be provide by a certain level
of intelligence resulting of the proactive nature and self-
organization of a software agent, there are also three significant
advantages:
 The physical object will be indirectly interacted
through the agent features. The uniqueness of the
technology and communication methods remains
inside the agent mechanics, so there will be only a
homogeneous and uniformed communication between
agents. By doing this, the concept transcends the
boundaries that limit the combination of different
technologies (switching between different
communication protocols and types of communication,
the possibility to enslave a non-connected device
through a do-it-yourself assembly, …).
 By permitting to see the physical objects as an assembly
of minimalistic virtual objects, we can create new
virtual objects by gathering virtual objects to find an
alternative solution of a material issue. In one hand, it
would be not necessary to add a new physical object in
a system to increase the available features. And in an
Fig. 5 Comparative schema of a conventional view and agentified view of the deployed system in a room

another hand, the environment allows a greater
robustness by providing possible alternative solutions to
potential hardware failures. Moreover, the unwanted
redundant feature can be avoided by having the choice
to use virtual connected objects instead of adding a new
physical connected object in the system.
 Regarding the user involvement in the conception and
deployment of a such system, the environment can
provide a greater level of malleability by permitting to
create virtual objects from physical objects or virtual
objects. There will be no assembly limit and it will
permit possibly to create new features without adding
new physical device. The range of connected objects
could be more freely increased if we can have access to
the capacities and features of the components that
constitutes the physical device
V. CURRENT AND FUTURE WORKS
Our team are currently working on a software architecture
that would take control of connected objects. The final form of
the platform aims to handle standard or do-it-yourself
connected objects with wires or wireless protocols, or more
basically direct control on sensors and actuators if needed.
With the aim of keeping the ease of use of the current
Internet of Things, we have split the architecture un several
platforms with different domain of tasks.
Currently, we are at the first versions of the software
platform named FullJS [21] which is a deployment platform
written in JavaScript. The main objective of this platform is to
deploy and support an application system in an environment
composed of one or more computers (of any nature and size)
and possibly be connected to electronic equipment such as
actuators and sensors.
Alongside, we also need a way to properly communicate
between all of the objects regardless the type and brand of
device. So we are working on a uniformed device architecture
which permit to enslave existing devices or created devices by
providing utilities and the ability to design dynamic
configuration such as the number and the specifications of
sensors/actuators at any time of a device. This architecture
allows us to increase the compatibility of the devices and also
serves as a transparent way of universal connection between
components of different devices. Presently, we have set a
minimum device specification for being at the same hardware
level as the Arduino.
In parallel, we also want to improve the smart side of the
agent in our collective and adaptive system by adding some
awareness execution contexts.
VI. CONCLUSION
This article highlights a possible form of the use of the
Internet of Things and more generally the connected objects
associated with collective and adaptive system brought by the
software agent approach. Several requirements have been
identified in order to achieve this vision of design. Multi-agent
approaches may be exploited to have for organization,
intelligence and a unified platform. By agentifying, connected
objects or part of devices, these agents could be used to create
virtual connected objects to be used at the same way as the
physical objects in the system of IoT. In this concept, the IoT
or more generally the network of connected objects becomes
the network of teamed agent regardless the limit of the physical
wraps which allow to deploy less devices but allow to keep a
minimum level of comfort. For reaching this stage of
evolution, there are still many challenges whether scientific and
technical, but we can easily perceive the benefits and steps of
progress of such progression in the near future in both
technology and sustainable development domains.
REFERENCES
[1] R. Khan, S. U. Khan, R. Zaheer, and S. Khan, “Future Internet: The
Internet of Things Architecture, Possible Applications and Key
Challenges,” 2012 10th Int. Conf. Front. Inf. Technol., pp. 257–260,
2012.
[2] N. Bari, G. Mani, and S. Berkovich, “Internet of things as a
methodological concept,” Proc. - 2013 4th Int. Conf. Comput.
Geospatial Res. Appl. COM.Geo 2013, pp. 48–55, 2013.
[3] E. Borgia, “The Internet of Things vision: Key features, applications
and open issues,” Comput. Commun., vol. 54, pp. 1–31, 2014.
[4] J. Sliwa and E. Benoist, “Pervasive Computing - The Next Technical
Revolution,” 2011 Dev. E-systems Eng., pp. 621–626, 2011.
[5] M. Conti, S. K. Das, C. Bisdikian, M. Kumar, L. M. Ni, A. Passarella,
G. Roussos, G. Tröster, G. Tsudik, and F. Zambonelli, “Looking ahead
in pervasive computing: Challenges and opportunities in the era of
cyber–physical convergence,” Pervasive Mob. Comput., vol. 8, no. 1,
pp. 2–21, 2012.
[6] M. Nemri, “Demain, l’Internet des objets,” Note d’analyse - Fr. Strat.,
vol. 22, 2015.
[7] John Greenough, “The ‘Internet of Things’ Will Be The World’s Most
Massive Device Market And Save Companies Billions Of Dollars,”
Business Insider UK, 2014. [Online]. Available:
http://uk.businessinsider.com/how-the-internet-of-things-market-will-
grow-2014-10?r=US&IR=T. [Accessed: 30-Sep-2016].
[8] G. Clair, E. Kaddoum, M.-P. Gleizes, and G. Picard, “Self-Regulation
in Self-Organising Multi-agent Systems for Adaptive and Intelligent
Manufacturing Control,” 2008 Second IEEE Int. Conf. Self-Adaptive
Self-Organizing Syst., pp. 107–116, 2008.
[9] M.-P. Gleizes, “Self-adaptive Complex Systems,” 2012, pp. 114–128.
[10] G. Fortino, A. Guerrieri, W. Russo, and C. Savaglio, “Internet of
Things Based on Smart Objects,” pp. 1–27, 2014.
[11] E. Gilman, I. S. Milara, E. Gilman, I. S. Milara, M. Cortés, and J.
Riekki, “Learning Systems Towards User Support in Ubiquitous
Learning Systems,” no. October, 2015.
[12] C. Jimmy, S. Stanley, S. Lily Leng-Hiang, K. Sze Ling, T. Yin-Leng,
and B. L. D. Henry, “Ubiquitous Computing History, Development,
and Scenarios,” in Ubiquitous Computing: Design, Implementation and
Usability, 2008, pp. 1–8.
[13] T. Leppänen, J. Riekki, M. Liu, E. Harjula, and T. Ojala, “Mobile
Agents-Based Smart Objects for the Internet of Things,” 2014.
[14] A. M. Mzahm, M. S. Ahmad, and A. Y. C. Tang, “Enhancing the
Internet of Things ( IoT ) via the Concept of Agent of Things ( AoT ),”
vol. 2, pp. 101–110, 2014.
[15] A. Ranganathan and R. H. Campbell, “A Middleware for Context-
Aware Agents in Ubiquitous Computing Environments,”
ACM/IFIP/USENIX Int. Conf. Middlew., pp. 143–161, 2003.

[16] F. Zambonelli, A. Omicini, B. Anzengruber, G. Castelli, F. L. De
Angelis, G. D. M. Serugendo, S. Dobson, J. L. Fernandez-Marquez, A.
Ferscha, M. Mamei, S. Mariani, A. Molesini, S. Montagna, J.
Nieminen, D. Pianini, M. Risoldi, A. Rosi, G. Stevenson, M. Viroli,
and J. Ye, “Developing pervasive multi-agent systems with nature-
inspired coordination,” Pervasive Mob. Comput., vol. 17, no. PB, pp.
236–252, 2015.
[17] F. D. Macías-Escrivá, R. Haber, R. Del Toro, and V. Hernandez, “Self-
adaptive systems: A survey of current approaches, research challenges
and applications,” Expert Systems with Applications, vol. 40, no. 18.
pp. 7267–7279, 2013.
[18] J. Ferber, “La kénétique: des systèmes multi-agents à une science de
l’interaction,” Rev. Int. systémique, pp. 13–27, 1994.
[19] S. Kernbach, Structural Self-Organization in Multi-Agents and Multi-
Robotic Systems. Logos Verlag Berlin, 2008.
[20] F. Sadri, “Ambient intelligence,” ACM Comput. Surv., vol. 43, no. 4,
pp. 1–66, Oct. 2011.
[21] D. Payet, “FullJS,” 2016. [Online]. Available: http://fulljs.onover.top.