=Paper=
{{Paper
|id=Vol-2086/AICS2017_paper_38
|storemode=property
|title=Adoption, Architecture and Technology of Enterprise IoT Systems – Towards a Framework of Concerns in IoT Environments
|pdfUrl=https://ceur-ws.org/Vol-2086/AICS2017_paper_38.pdf
|volume=Vol-2086
|authors=Popescu Mirona,Dumitriu Dan,Helfert Markus
|dblpUrl=https://dblp.org/rec/conf/aics/PopescuDH17
}}
==Adoption, Architecture and Technology of Enterprise IoT Systems – Towards a Framework of Concerns in IoT Environments==
https://ceur-ws.org/Vol-2086/AICS2017_paper_38.pdf
Adoption, Architecture and Technology of Enterprise IoT
Systems – Towards a framework of concerns in IoT
environments
Mirona Popescu1, Dan Dumitriu1, Markus Helfert2
1 Faculty of Entrepreneurship, Business
Engineering and Management, Politehnica University, Bucharest, Romania
{mirona.popescu15 ¦ dumitriud} @gmail.com
2 School of Computing, Dublin City University, Dublin, Ireland
Markus.Helfert@dcu.ie
Abstract. Internet of Things - IoT (Internet of Things) or Internet of Everything
defines a network of objects that incorporates electronic circuits that allow
communication via existing infrastructure (network INTERNET), wireless or
cable for many purposes, including monitoring or remote control. The existence
of a web interface through which a person may access the equipment of the house
is useful to increase efficiency and save resources. Both own homes as well as
companies, medical services, factories, state services and communities will
benefit from using connected IoT environments. These environments are
characterized by information sharing between equipment, devices and humans
help devices and the exchange of information between these. This research
outlines the benefits brought by IoT systems adopted in companies from different
viewpoints. We employ a case study in order to point to real experiences and
situations. A further objective of this paper is to propose a model of architecture
for the systems and examine their process of operating by analyzing the
technologies used to develop such systems and the security problems that can be
met.
Keywords: IoT systems, enterprises, IoT technology, machine learning, big data
1 Introduction
Invented in 1999 by British entrepreneur Kevin Aston, the concept of Internet of Things
(hereinafter IoT) represents a structure of physical objects, or "things", which have
electronic components, software, sensors, and Internet connections to collect and
distribute data. Objects are uniquely identified, "self-aware" and can communicate with
each other, locally or globally, without human intervention (based on IP connectivity).
IoT also represents the omnipresent connectivity concept for businesses, governments
and consumers, with their own management, monitoring, statistical computations and
data analysis systems.
The above definition can be extended beyond objects, including humans and animals
alike. For example, in IoT, a "thing" also refers to a person with a monitored heart
implant or a dog with a microchip under the skin of his head or any other natural or
artificial object to which an IP address can be attached and which is capable to transfer
data over an Internet network.
Internet of Things is already omnipresent, a daily presence in many people's lives.
For example, smart meters are installed in homes to coordinate and save electricity;
Internet-powered cars are used from OnStar (US) or eCall (EU) system that triggers an
automated response in case of an accident. Furthermore, the system allows to follow
cars if they are stolen or provides technical assistance when needed; on bodies as an
insertion in the smart shirt or shorts they run with, or as an auditory implant that
captures the noise of the fire truck and "moans" before they can hear it.
The beginnings of internet of things were: the barcode (1974); Radio Frequency
Identification (RFID) device; the incredible price drop of database storage devices
required to collect, store and process thousands of bits; the emergence of IPv6, the
Internet protocol that replaced the previous version, IPv4. With the new protocol, an
Internet address may be available in any object that has software stored in it: the
toothbrush, the coffee machine, the refrigerator, the dishwasher, etc. Technologically,
IoT has become a collection of bar codes, QR codes, RFID tags, NFC (Near Field
Communication) and SAAS (Machine-to-Machine) communication devices, active,
Wi-Fi and IPv6.
Pan [7] investigated a unique IoT experimental testbed for energy efficiency and
building intelligence, and identified the heterogeneous IoT devices as major challenges
to work together as a coherent system. Similar Al-Fuqaha [8] identified that the
heterogeneity of the IoT elements is challenging. To overcome the existing challenges,
solution have been suggested, e.g. on communication enabling the use of wireless
sensor network (WSNs) [9], enabling technologies and application services using a
centralized cloud vision [10], enabling technologies with emphasis on the RFID and its
potential applications [11]. However, the technologies currently used to implement IoT
bring some flaws especially when it comes to security, but it is working to strengthen
it to provide people and companies with the necessary safety and trust.
The aim of this paper is to outline the benefits brought by implementing IoT systems
in companies regarding their field of business. Although many paper have examined
the benefits and challenges of IoT, this paper specifically focuses on a corporate
environment. We outline how IoT has appeared and the need of its existence in an
enterprise environment that aims to evolve and improve its actual status and numbers.
In our research, we employ a case study to examine IoT usage in Enterprises.
Furthermore, an overview of an architecture model is proposed based on the case study
findings. It illustrates the complex dependencies and workflows between its
components within an IoT environment.
2 Related Work
A great amount of resources and research have been directed towards the adoption of
IoT in various areas. Diverse application areas of such technologies are often
summarized with terms such as ‘Smart City’, ’Smart Home’, ‘Smart Buildings’ and
lately Smart Commerce (Pan, [7]). Smart environments include smart objects, such as
houses, buildings, sustainable urban infrastructure, cars, sensor technology, etc. Within
these environments, through the application of semantic web technologies and
intelligent applications, the systems may be personalized, responsive, and intuitive.
It is considered that the determinants of the proliferation and the success of IoT are
the emergence and development of methods of artificial intelligence, machine learning
and data mining. IoT is essentially a continuous flow between:
1. Body Area Network (BAN): Audible implant, smart jersey or shirt;
2. LAN (local area network): intelligent meter as the interface of our home;
3. WAN (wide area network): bicycle, car, train, bus, drone, all intelligent;
4. VWAN (very wide area network): the "smart" city, where e-services are
omnipresent, without being linked to physical locations.
The increasing importance of IoT environment are due to mainly 4 factors:
(1) The exponential increase of computer capacity and the emergence of IoT could
bury the world in huge volumes of unintelligible data. To solve this problem
before it becomes too complex to be approached, the use of Artificial
Intelligence (hereinafter AI) methods has become the favorite option of
companies that try to control Big Data. The purpose of AI is to take large
amounts of unstructured data - such as those produced by IoT devices - and
make actionable decisions about these data. In short, AI is data-giving
technology and produces significant instructions, such as commands given to
IoT devices to perform specific actions.
(2) Machine learning is the global term for algorithms that, automatically or with
some human support, identify patterns in the Big Data Rivers and determine
which IoT device behaviors tend to create the most desirable results. Based on
learning algorithms, computers can act without explicit programming. In the last
decade, machine learning technology has given people, among other things,
driverless cars, speech recognition, effective Internet search and a broad
understanding of the human genome.
(3) Analytics and Data mining techniques as the analytical process of exploring
large databases - Big Data - for finding and finding consistent patterns and / or
systematic relationships between variables, followed by applying patterns
detected to new data sets.
3 Concerns of IoT in Enterprises
3.1 IoT adopted by companies
The Internet of Things (IoT) is assumed to be a key part in the business world as
companies progressively search for original approaches to keep up their rival edge.
Research firm International Data Corp estimates that the IoT market will grow from
$655.8 billion in 2014 to $1.7 trillion in 2020. Another analyst, McKinsey and
Company noted there will be between 20 to 30 billion linked gadgets by 2020.
Buyers are starting to see the benefits of having their devices linked to each other
and controlled remotely, associating and matching up different gadgets going from their
wellness trackers to autos to home surveillance cameras. Companies all over the world
are investing large amounts of their capital in adopting IoT solutions to become
digitized and drag the advantages brought by the connected things and the data
generated by them.
Additionally, IoT will benefit all types of business by reinventing industries on three
levels:
Fig. 1. Levels of a business
This research outlines a part of the advantages brought by IoT systems that are
summarized in Fig2. and they are derived from the case studies discussed later in
section 4.
Fig. 2. Benefits of IoT (adapted after https://vmokshagroup.com/blog/6-ways-
businesses-can-take-advantage-of-iot/)
3.2 IoT Architecture
The architecture of an IoT is adopting a three-tier pattern which includes edge, platform
and enterprise tiers as follow:
The public networks and proximity are found in the edge tier. There is gathered all
the data received from devices and transmitted to devices through the gateway or
straight from the device into the Cloud provider, using the edge services and IoT
transformation and connectivity.
Cloud provider represents the platform tier, which is in charge to process and analyze
the information received from the edge tier and it provides API management and
visualization. It has the capacity to launch commands of control from the enterprise
network to the public one.
Enterprise network defines the enterprise tier and it has the data from the enterprise,
a enterprise user directory and the applications. The data goes flow to and from the
network using the transformation and connectivity component. The data stored in the
enterprise is gathered from structured and unstructured sources of data and real-time
data that comes from stream computing.
Systems that are based on IoT are relying on application logic and control logic in a
hierarchy of locations that need timescales and datasets to inform the decisions. They
are parts of the code that may be executed directly on devices at the edge of the
networks or in the IoT gateways that are in the proximity of the devices. Other parts of
the code are carried out in the enterprise network or the provider cloud services.
Security systems and IoT governance have the role to span the architecture’s
elements to provide policies and control for the data and applications which are defined
and enabled in the whole system.
The figure below will illustrate the process described above and the connections
between all the components of an IoT system:
Fig. 3. IoT Overview (adapted after IBM Model of Architecture, source:
https://www.ibm.com/developerworks/cloud/library/cl-grush-smart-toothbrush-
bluemix-trs/index.html)
3.3 Technologies used for developing IoT systems
The basic principle behind the Radio frequency identification (hereinafter “RFID”)
chips is radio wave technology. When waves pass through copper wire coils an
electrical signal is generated which, in turn, powers the chip. The power obtained is
used to send a reply. The following sequence of events details the aforementioned
process
1. A RFID transceiver sends a pulse of radio waves. This pulse contains an
identifying number
2. If an RFID tag is nearby, the pulse hits the tiny built-in antenna, creating
electricity. This electricity makes the circuit "come alive" momentarily.
3. The chip checks the ID number transmitted from the receiver. If it matches, then
the chip transmits its stored information in the form of a reply radio wave.
4. The transceiver picks up the reply signal from the RFID chip and the transaction
is done.
The current problem with using RFID chips is that the current radio signal lacks in
power. Thus, it cannot reply to a receiver in a location further than a few centimeters.
Be that as it may, the amount of energy it gets and the strength of the reply signal is
a percentage of the power of the original pulse. So on the off chance that a person utilize
a substantially more effective pulse, the chip can send a considerably more powerful
reply. A sufficiently solid pulse can support the range to a few meters.
In this case "hacking" comes into play. With a sufficiently solid transceiver, a human
could set it up to check all the RFID contributes everybody's card as they passed by.
Obviously, the person should send the right code to each RFID chip to motivate it to
transmit an answer. The answer may likewise be scrambled, and that can be hard to
break. Be that as it may, nothing is "uncrackable", and the original RFID chips did not
use any form of encryption. Some still do not.
The following table details the contents found inside the receiver data packet:
Fig. 4. Receiver data packet content
ID is the unique identifier of the card. It can be used this for a basic check, but beware
there are cards with unlocked sector 0 that can have this number reprogrammed. The
number is also not really unique, only unique to a manufacturer's production run of 232
cards (for cards with a 4-byte UID). This might be fine depending on what you are
doing with it.
ATQA indicates the type of the card and SAK indicates the manufacturer. More
details can be found inside the ISO/IEC 14443 Type A document.
The sectors are the data storage to a card. The sector 0 is typically write protected as
that stores the information presented above. If you look at your dump you can pick out
some of these values in there.
The remaining sectors are available for storage on most cards. However, 16 bytes at
the end are reserved for access control to the data. This includes two keys of 6 bytes
each and 4 bytes that control what authenticating with these keys allows you to do with
the data (read/write/increment, etc.). A new card has keys consisting of 6 0xff bytes.
If security is concerned, one or more sectors of the card need to be populated with
additional data to match against beyond the UID. To authenticate a card, the ATQA,
SAK, and UID need to be used to look up the read key and expected data in the card,
and to authenticate the sectors with the data that should be matched, in the end a
compare is done between what is found inside the card to what you is already stored.
As RFID has been around for a while, the technology has matured, and has spun off
a newer technology: near-field communication (NFC). The emergence of NFC seems
to have also sparked confusion. What’s the difference between RFID and NFC? RFID
is a one-way process. Information is transmitted from an encoded memory chip (known
as a “smart tag”) via an antenna to an RFID reader. There are two types of RFID tags:
active and passive. Active RFID tags contain a power source, so they can broadcast a
signal, up to 100 meters away. This capability makes RFID a strong choice for asset
tracking. Passive RFID tags have no power. They’re activated by an electromagnetic
signal sent from the RFID reader. The signal doesn’t travel as far as active RFID, so
they’re used for short read ranges. Passive RFID falls into one of three frequency
ranges: Low frequency: 125-134.2 kHz High frequency: 13.56 MHz Ultra-high
frequency: 856-960 MHz NFC is based on RFID protocols. The devices run at passive
RFID’s high frequency. NFC reads smart tags because, like RFID, it features a
read/write operation mode. But NFC goes farther than RFID. The technology has two-
way communication—unlike RFID’s one-directional limitation—using one of two
modes: card emulation and peer-to-peer (P2P). For example, a smartphone enabled with
NFC (and many of them are nowadays) can pass information back and forth to another
NFC device. Contactless payment is an example of card emulation mode. Any time you
redeem rewards points via your phone, you’re also using NFC’s card emulation feature.
P2P comes into play when you “bump” your mobile device with another one to share
information. Maybe you’re passing music back and forth, swapping special deals, or
playing a game with the friend sitting next to you. You can even tap your device with
a router, to get on that network without having to use a password. NFC will soon likely
replace QR codes in some advertisements and promotional materials. Consumers will
no longer have to scan a QR code to get to the intended location, but can simply use the
NFC mode to instantly get the information that the advertiser wants them to have. There
is still plenty of space across today’s industries—from retail to manufacturing,
transportation to healthcare—for RFID’s one-way communication, but NFC is paving
another path along the ever-winding information highway.
[http://www.aaarfid.com/UploadFiles/RFID%20VS%20NFC.pdf]
4 Case Studies
Beyond connecting devices to IT systems, the true value of IoT is the ability to track,
almost in real time, the inherent qualities or relationships existing in data that are always
growing in volume, speed and variety. This section highlights some sectors or industries
whose activities were influenced by IoT systems.
Asset tracking had been widely welcomed by the O&G industry. "There is
dependably an inclination that insufficient resources are being conveyed rapidly
enough," says Henry Rosen, senior vice president of sales and marketing for Geoforce,
Inc., an O&G equipment asset tracking and utilization solutions provider. "With the
move from RFID implementations to Global Positioning System (hereinafter GPS)
technology and IoT arrangements, investment funds have hit endeavor adopters'
primary concerns in two ways: working and capital uses. For instance, working
expenses are lessened with fuel investment funds and diminishment in man hours; and
capital expenses are decreased with expanded volumes of turns and speed of
utilization."
Consider the instance of an O&G rental hardware supplier battling with the expenses
of dealing with its remotely conveyed resources. Given the multifaceted nature of the
gear, experts needed to research client complaints on location, regularly spending time
in between sites. Sensors were put on basic parts of the hardware that would
automatically send alarms if gear related issues occur and give enough information to
specialists to permit some remote investigating and determination. Site visits
diminished by 50 percent, reducing personnel, vehicle, and fuel costs. Furthermore, the
measure of time required for issue solving dropped from hours to minutes, bringing
about less machine downtime. With bring down workloads, specialists could be
conveyed to different parts of the business.
IoT can possibly empower O&G organizations to investigate and find new resource
locations in much deeper and complex deposits. Sensor gadgets used on-field at
exploration sites can assemble and transmit seismic information to a focal observing
area, making information streams that may give new understanding into potential new
sources. IoT arrangements may give more noteworthy accuracy to reveal potential
O&G deposits at bigger depth underneath the earth's surface, possibly enhancing the
likelihood of new disclosures and taking into account better resource/hardware and
workforce arranging.
Another possibility is to look at how as an oil field administrations organization may
focus on its client connections by utilizing the ongoing information spilling out of oil
generation clients, joined with inner gear execution information and freely accessible
products information. The oil field administrations organization could utilize the data
to screen clients production rates relative to fluctuating energy prices to anticipate
equipment needs by geography and potentially offer new, valuable services.
Telematics and car pool maintenance
For companies that keep up car pools, telematics can make strides vehicle operations
to lessen hazard and costs and also enhance client satisfaction. Constant upkeep and
hazard administration can offer assistance organizations guarantee that items and
administrations are in great working request. GPS recipients can speak with electronic
GSM gadgets introduced in every vehicle to permit endeavors "to screen the area,
developments, status, and conduct of a vehicle or pool of vehicles," as indicated by
Stein Soelberg, executive of advertising at KORE Telematics. Such arrangements can
identify issues in vehicles before they are utilized. For instance, one maker of cranes
and heavy machinery put sensors in tires to decrease the danger of machine failure. The
sensors permitted consistent checking of tire weight with the goal that the maker could
foresee potential imperfections and proactively settle them.
Telematics arrangements can possibly help vehicle makers get a constant stream of
rich insight about their vehicles. As IoT networks collect copious amount of
information produced from the pool, it can empower organizations to disentangle and
present new ideas or highlights in their current product line-up. For instance, the sensors
in heavy machinery hardware can give rich data back to the maker about how and where
the item is utilized and how that influences execution over its lifecycle. This knowledge
may prompt item configuration changes. It additionally yields knowledge into clients'
organizations, which the producer may offer back to the client as a form of potential
feedback regarding the current contract.
mHealth and remote wellbeing observing
Pete Celano, director of consumer health initiatives in the Innovation Group at
MedStar Health, has seen firsthand the advantages of IoT arrangements in medical
services and how mHealth and remote wellbeing arrangements will upset medical
services through guidance ahead of time and preventive care. "In a machine-to-machine
interoperability case, the potential outcomes of remote medical gadgets are changing
remote patient observing, particularly as there are structural movements occurring in
the medical services industry." Hospital frameworks, for instance, take advantage from
additional clever medical hardware that upgrade preventive care through new sorts of
medical sensors that empower remote, constant observing of imperative wellbeing
related measurements, independent of where the patient is found.
Celano depicts how a patient at home "can transmit wellbeing related information to
the cloud by means of a non-stop remote system service, giving early-cautioning signs
to facility's nurses that the patient is in linked to, or going into pressure." Medical
experts can react with fitting actions. "During a time where doctor's facilities are getting
to be payers as well, there is a need to oversee healing center confirmations proactively.
Progressively, if suppliers do not provide results, there is no replace for the old test-
heavy, experience based model and implies the smart medical services frameworks are
hustling to make programs that shield patients from winding up in the crisis room."
This shrewd observing capacity prompts other human services administrations and
medications, for example, cutting edge wellbeing dashboards and health centers.
Market revenue for IoT developments in social insurance was evaluated at $10.6 billion
out of 2012 and is relied upon to twofold by 2016.
Remote wellbeing checking starts to upset the conventional treatment condition and
the occurrence based model. The blend of wearables and other intelligent hardware that
enables a person to catch, track, break down and share information about themselves,
will open up considerably more problematic issues: the individual, as opposed to the
social insurance supplier, will claim the information about his/her body. Wellbeing and
health suppliers will utilize the information to offer more customized treatment plans
in light of the person's specific circumstance and reactions to past measures. People
will associate with a differing biological community of health suppliers who can enable
people to obtain insight out of their information to enhance wellbeing and better utilize
specialized medical services suppliers when they are not well.
IoT can possibly empower oil and gas, transportation and healthcare organizations
to expand and find new resource locations, to disentangle and present new ideas or
highlights in their current product through telematics arrangements for companies that
keep up car fleets, as well as having a significant impact in the development of
preventive care.
5 Concluding Remarks and Future Research
The Internet of Things is a paradigm shift and an ontological change. The fundamental
notions of what it means to be human and what it means to be "in humanity" are based
on subject-object dichotomies. IoT introduces a third dimension in the current
existential relationships: Big Data, specific algorithms, and a realistic scenario always
present in any object-subject interaction. If we are to refer to the "next epochal
invention" of mankind, IoT is just as important as the fire or the book - a paradigm shift,
a disruptive event, the creation of a new vision of the present times.
There is a myriad of areas where IoT technologies are currently applied or will soon
be applied: intelligent homes, smart cities, automotive, health, etc.
The biggest advantage of using IoT technology is savings in city budgets. Mixing
IoT systems with Big Data could save US 1,200 billion dollars. Installing intelligent
transport systems in major cities (e.g. New York, Chicago, Los Angeles, San Francisco)
would contribute with 800 billion a year starting in 2030. Other areas where smart cities
will get spending cuts are water and energy consumption, domestic and industrial waste
management, public safety (firefighters, police officers, rescuers, and traffic
controllers) or intelligent buildings (e.g. Seattle is a world leader in its efforts to build
buildings with low energy consumption).
In the public smart cities sector, it can be also met IoT devices that fulfill a multitude
of roles - from monitoring the level of rivers to triggering flood alerts, identifying
different types of offenders, etc.
Intelligent cities technologies will cost US for example 27.5 billion by 2023, and the
number of these cities will increase to 88 in 2025, according to a report made by the
New Jersey Institute of Technology (NJIT). Intelligent city data will be processed and
analyzed to determine how to improve the quality of life in each city, from managing
financial resources to fighting crime, and so on.
Implementation of IoT on a global scale, affecting billions of people and devices,
has not only benefits. Two aspects - cyber security and protection of personal privacy
- seem to be the essential risks that IOT is currently facing. Secondarily, there can be
listed issues related to the integration, reliability and scalability of the data used.
Often, Internet connections are vulnerable regardless of their type (Wi-Fi, Bluetooth,
cellular, satellite, or microwave) or how they are connected to the PAAS (Platform-As-
A-Service) or other servers and services set to search, process, or receive data from an
IoT application. Vulnerability classes discovered, for example, in the hydrocarbon
industry include undocumented protocols, unsafe protocols, weak passwords and
"backdoors" (various ways to circumvent legitimate users authentication).
The purpose of this paper is to present the relevance of IoT systems by evaluating
companies that have adopted it in various domains to automate some of the processes
and increase sales, leading to firm stabilization and recognition of the company through
the profit obtained and the fulfillment of the consumers' requirements. Adopting them
is a win-win situation for both the company and the consumer. The paper also presents
an architecture model proposed for IoT systems, but also the technologies that are
currently used with their advantages and disadvantages. In conclusion, the research
succeeds in highlighting the importance of technology-based evolution, artificial
intelligence and IoT, which is currently expanding due to the many functions that help
people once they benefit from it.
Acknowledgement: This work was supported, in part, by Science Foundation Ireland
grant 13/RC/2094 and co-funded under the European Regional Development Fund
through the Southern & Eastern Regional Operational Programme to Lero - the Irish
Software Research Centre.
References
1. http://www.contributors.ro/media-tech/the-internet-of-things-iot-lumea-obiectelor-
inteligente-beneficii-%C8%99i-riscuri/
2. https://www.reddit.com/r/Futurology/comments/4bgy6f/an_excellent_overview_of_the_in
ternet_of_things/
3. http://www.businessnewsdaily.com/5450-internet-of-things-business-opportunities.html
4. http://wso2.com/library/articles/2017/02/six-business-benefits-of-the-internet-of-things/
5. https://vmokshagroup.com/blog/6-ways-businesses-can-take-advantage-of-iot/
6. https://www2.deloitte.com/us/en/pages/technology-media-and-
telecommunications/articles/internet-of-things-iot-enterprise-value-report.html
7. Pan G, Qi G, Zhang W, et al. (2013). Trace analysis and mining for smart cities: issues,
methods, and applications. Communications Magazine, IEEE, 51(6), 120-126
8. https://www.ibm.com/developerworks/cloud/library/cl-grush-smart-toothbrush-bluemix-
trs/index.html
9. Atzori L, Iera A and Morabito G, (2010), The internet of things: A survey, Computer networks,
54(15), 2787-2805
10. Al-Fuqaha A, Guizani M, Mohammadi M, Aledhari M and Ayyash M (2015), Internet of things:
A survey on enabling technologies, protocols, and applications, IEEE Communications Surveys
& Tutorials, 17(4), 2347-2376.
11. Gubbi J, Buyya R, Marusic S and Palaniswami M, (2013), Internet of Things (IoT): A vision,
architectural elements, and future directions, Future generation computer systems, 29(7), 1645-
1660
12. Yang D L, Liu F and Liang Y D, (2010), A survey of the internet of things, In Proceedings of the
1st International Conference on E-Business Intelligence (ICEBI2010), Atlantis Press.