Content uploaded by George Hatzivasilis
Author content
All content in this area was uploaded by George Hatzivasilis on Mar 25, 2021
Content may be subject to copyright.
The CE-IoT Framework for Green ICT Organizations
The interplay of CE-IoT as an enabler for green innovation and e-waste management in ICT
George Hatzivasilis, Nikos Christodoulakis, Christos
Tzagkarakis, Sotiris Ioannidis
Institute of Computer Science
Foundation for Research and Technology – Hellas
Heraklion, Crete, Greece
{hatzivas, christodoulakis, tzagarak, sotiris}@ics.forth.gr
Giorgos Demetriou
Ecole des Ponts Business School
Paris, France
g.demetriou@pontsbschool.com
Konstantinos Fysarakis
Sphynx Technology Solutions AG
Zug, Switzerland
fysarakis@shpynx.ch
Marios Panayiotou
Cablenet Communication Systems Ltd.
Nicosia, Cyprus
m.panayiotou@cablenetcy.net
Abstract—The growth of the global middle class provokes
significant increment in product consumption. As the available
resources are limited, Circular Economy (CE) raises as a
promising initiative towards the sustainable development. Except
from the traditional approaches of reusing or recycling products,
the current trend utilizes modern computer technologies and
involves a data-driven aspect. The Internet of Things (IoT) is the
main enabler for the integration of CE with technology. This
paper proposes a framework for implementing the cooperative
vision of CE and IoT. Via this solution, a pilot system is
developed in a medium size telecommunication company for
administrating the lifecycle of the deployed electronic equipment
and the management of the related supply chains. Mechanisms
and devices are maintained/repaired/fabricated in a regular
basis, green computing techniques are efficiently applied, and the
productive period is prolonged. When the business upgrades the
system, the retired counterparts can be sold in start-ups or gifted
in third-world countries. The overall approach extends the
working period of the well-maintained electronic assets not only
for the examined business but for the collaborating organizations
as well. Recycling companies can then trace this supply chain and
the assets’ status in order to define their investment strategy at
the end-consumer, contributing in the reduction of the electronic
waste problem in the third-world.
Keywords—Circular Economy; Internet of Things; Sustainable
development; Green computing; E-waste; Circularity by Design;
IoT; IIoT;
I.
I
NTRODUCTION
The continuous raise of the world population and the
expansion of the global economy are stressing the planet’s
physical resources. As the environmental sustainability is
threatened, several governmental initiatives try to tackle the
problem. Green innovation is gaining ground the last years due
to the environmental movement and climate change (e.g. [1],
[2]). In 1997, United Nations adopted the Kyoto protocol to
reduce the greenhouse gases emissions and counter global
warming. The main period of the protocol expired in 2012 and
its extension until 2020 was announced.
Circular Economy (CE) is such an act towards a
regenerative system, where the consumption of new resources
is minimized by extending the lifecycle of current products
([3], [4]). The economy is formed in closed loops and the
materials are reused over-and-over by the various actuators.
The main states of a CE model include long-lasting design,
maintenance, repair, reuse, remanufacturing, refurbishing, and
recycling. The adoption of the CE principles in a large scale
inevitably leads to economic transformation and the
establishment of new business models [5]. An ambitious
innovation package by the European Union (EU) is provoking
this evolution in the Old World [6] and the global economy.
Significant efforts are also devoted by the Chinese government
[7].
Except from the ordinary CE settings, the integration with
Internet of Things (IoT) technologies is now emerging [8], [9].
Pioneer services for smart city and intelligent transportation are
popular here, while other significant economic sectors, like the
Information and Communications Technology (ICT), have not
gain much attention.
This paper proposes the CE-IoT: a data-driven framework
that materialize the CE business model and the integration with
the IoT infrastructure. An organization observes the deployed
equipment status via system monitors and IoT devices that
gather related knowledge. The information is then collected
and processed by machine learning (ML) and data-analytics
algorithms. The results are parsed to decision support systems
(DSS) that outcome the organization’s administration strategy.
The reactive policy is implemented by artificially intelligent
agents that manage the system at runtime and facilitate the CE
actions. Green computing techniques can be also enforced
increasing the utility of the current products [10].
The presented solution is quite generic and applicable in
various CE-IoT settings. Nevertheless, we demonstrate its
deployment in ICT organizations. Thus, a preliminary version
is developed and the main functionality is tested under the
cloud infrastructure of the medium-scale ICT provider,
Cablenet
1
, with collaborating organizations being emulated. To
our knowledge, this is the first concrete implementation of the
interplay between the CE and IoT visions in the ICT domain.
The rest of the paper is organized as: Section 2 mentions
the related work in the field of CE and IoT. Section 3 sketches
the CE-IoT framework and the deployment on the ICT
organization. Sections 4 presents the machine learning and data
analytics procedures that enhance forecasting and the
establishment of green computing policies. Section 5 details
the decision making model and the administration of the CE
actions. Finally, Section 6 concludes and refers future work.
II. R
ELATED
W
ORK
A. CE and IoT
CE review studies in [3] and [4] present the business
principles of the new initiative. Further CE models and
indicative examples are detailed in [5]. Smart devices and the
IoT ecosystem enable new forms of interaction and business
models. The transformation of the traditional market into a
service-oriented setting is now a fact. The integration of CE
and IoT further promotes such data-driven service-oriented
architectures (SoA) [8], [9]. Related cases are presented below,
deriving the various examples from innovative industrial
sectors.
The tire company Michelin added Fleet Solutions [11], a
leasing application for EU trucking fleets, manner back in
2000. The business-to-business enterprise initiative leases "tire
services" per the kilometer, charging a flat fee that scales
primarily based on automobile type and distance pushed. The
organization states that the carrier reduces the risk of
fluctuating costs that include variability in tire performance,
purchase fees, and unpredictable damage costs from owning
the tires. Rather than requiring clients to pay prematurely for
the tire and fee of changing it, Michelin absorbs a number of
that risk.
Cars are from several perspectives one of the most beneath-
utilized sources in modern lifestyles. A research in Denmark
reveals that an average vehicle transports around 1.4 humans
on the street and spends about 23 hours per day taking up
parking space [12]. Empty automobile seats are the most
important excess ability within the shipping enterprise, and this
inefficiency has a poor impact at the environment and is
highly-priced for car proprietors. GoMore’s platform [13]
offers a provider that permits drivers to ask persons to join a
journey on an already planned course. This will increase the
quantity of people in vehicles and reduces the range of cars on
the streets, making automobile journeys greener and friendlier
for the environment. It results a travel solution that is
inexpensive for both the auto owner and the passenger. Given
that GoMore changed into mounted in 2005, more than 1
million users have visited and used the service. As an end
result, more than 300,000 seats have been bought only in
Denmark, making GoMore a European chief in combining
ridesharing and peer-to-peer vehicle condo, and Scandinavia’s
main commercial enterprise inside the sharing economy.
1
Cablenet: http://cablenet.com.cy/en/
In 2014, the French train manufacturer Alstom commenced
imparting HealthHub [14], a predictive protection device that
monitors the health of trains, train infrastructure, and signaling
structures. It uses advanced information analytics to extend and
maintain the beneficial life of trains. HealthHub uses records to
capture and diagnose the degree of performance of wheels,
brake pads, and put on-and-tear-inclined pantograph carbon
strips (the framework that conveys currents to trains from
overhead wires).
The German organization MAN Truck & Bus is a main
international provider of commercial and transport vehicles to
companies. It offers support and advice with car design for
fleets, operations, infrastructure, carrier, and maintenance. Last
year, MAN rolled out a data-driven service to assist
commercial truck and bus fleets with the transition to low-
carbon transportation [15]. Specifically, the corporation is that
specialize in helping companies with challenges which
includes variety making plans, charging infrastructure and
battery swapping issues, capacity planning and loading times,
and considerations round heating and air-conditioning
requirements.
B. Green ICT
On the other hand, ICT is one of the main contributors of
the financial system. The ICT infrastructure consumes the 3%
of the global energy and produces about 2% of the global CO
2
emissions [16]. The trend of green computing and networking
includes practices of selecting energy efficient computing
technologies and products, minimizing resource consumption
whenever possible [2]. The Climate Group and the Global
eSustainability Initiative estimated that such movements could
reduce the global emissions by 15%, achieving around an
eight-fold benefit comparing to carbon reduction expectations
[17].
In general, the green ICT paradigms try to reduce the
consumption of non-renewable resources (e.g. [7], [18], [19],
[20], [21]). The integration of the two domains [22], [23] tries
to provide efficiency along the supply chain for the involved
products [24], [25], offering better rental [26] and e-waste
services [27], [28]. New business models are then applicable
materializing the so called recycle-rewarding system (RRS),
like the e-waste management framework in [29].
In the current ICT setting, a typical business would use the
bought equipment and serve the customers’ operations. When a
damage occurs, the counterpart is either repaired or replaced by
a new one. The retired equipment is threw away or recycled. If
the company participates in a collaborative movement, the
machinery could be re-sold in a low price or gifted to an
organization in a third-world country. Such actions have led to
an important increase of the Internet users and the expansion of
the ICT sector in these countries [27]. On the other hand, the
problem of environmental pollution and electronic waste (E-
waste) has emerged. As these end-users do not have the
adequate capacity to repair occurring breakdowns of the
outdated equipment, it becomes useless and ends up in
stockpiles, landfills, or swamps [28].
The CE-IoT framework integrates the two sectors of CE
and IoT. The efficiency of this combinatory approach is tested
on the ICT industry. The case study that is presented in this
paper, details the appliance of the proposed setting on
Cablenet. Cablenet is a privately-owned company in Cyprus
that offers telecommunication and entertainment services to the
retail and corporate markets. The network capacity is
terminated in key European locations and operates a fully
redundant, self-healing ring in mainland Europe over multi-
10G capacity. Among the various Cablenet’s products and
services, CE-IoT is deployed and administrates a cloud
services platform that is offered to the end-customers. With
CE-IoT in place 6 main goals are accomplished:
1. The current host organization monitors the assets status
and maintains them regularly and proactively, avoiding
breakdowns and economic loss
2. Green computing is applied by promoting policies that
distribute the computational/communicational effort
among inactive intervals, limiting the equipment
overworking during peak periods
3. The damaged products are repaired or fabricated and
they are reused, extending their beneficial lifecycle
4. When the company performs a system upgrade, the
retired equipment is sold to startups or gifted in third-
world countries, contributing to the collaborative acts
5. If the equipment cannot be fixed, it is recycled, reusing
the available materials in order to manufacture new
products
6. The recycling organizations can gather information
from the involved parties in order to facilitate their
investment strategies closer to the end-consumer,
improving their operation
III. CE-I
O
T
F
RAMEWORK
In this subsection, we present a preliminary implementation
of the CE-IoT framework on real devices and the deployment
on the cloud platform. The system is also applied in an
emulated supply chain of the telecommunication sector where
data are collected in real-time by IoT technologies and
decisions are taken automatically regarding the CE aspects of
the system’s maintenance and equipment supply.
A. The Cloud Services Platform
The evaluated cloud services platform is hosted on
Cablenet’s infrastructure and the provided services include: i)
hosted exchange email, ii) web hosting with personalized cloud
control panel, iii) cloud server and mini Virtual Private Server
(VPS), iv) cloud backup with support on workstations, servers,
and Virtual Machines (VMs), and v) Domain Name System
(DNS) hosting. The high-level architecture of Cablenet Cloud
Platform includes different parts as depicted in Fig. 1. More
precisely, the most significant parts compiling the platform
infrastructure and functionalities are listed below:
• Cloud platform is deployed utilizing Cablenet’s
Datacenter Virtual Infrastructure (DVI). DVI is
physically protected by a number of protection
measures focusing on controlling physical access,
mitigating power‐related risks, providing air
conditioning and fire suppression
• One of the major roles on the provision of DVI is the
heavy-duty support by the Internet Protocol (IP) /
Network infrastructure, which is the backbone towards
providing high quality connectivity and resiliency for
the DVI cloud services
• Backbone/Core network operates based on multi-10G
self-healing and fully meshed setup and provide
connectivity over IP Multiprotocol Label Switching
(MPLS) / Virtual Private Network (VPN) throughout all
points of Cablenet presence and beyond
• Data access is provided via multi-10G redundant
connections towards the cloud blade center depending
on the data exchange traffic and needs, as well as in
similar manner storage is directly connected on the
cloud blade chassis for the storage/inventory
• The DVI includes installations of several products and
components to provide its services and ensure their
security from an IT perspective, including Microsoft’s
Exchange server and tools for clustering and replication
such as the VMWare High Availability Cluster,
Microsoft Cluster Service on VMware vSphere and
VMware vSphere Replication
• DVI’s set up includes also tools for security,
anti‐malware management, data encryption and secure
communications, as well as tools for managing the
quality and Security and privacy requirements operation
of the data center infrastructure (e.g., tools for executing
backup processes, identifying improvements to server
application or maintenance levels, adjusting parameters
and physical resources such as memory, CPU, disk).
Fig. 1. Cloud services platform architecture.
B. IoT Technologies
A popular choice for materializing an intelligent application
is the agent technologies (e.g. [30], [31], and [32]). Software
agents collect ambient information from the underlying
subsystems and manage them based on artificially intelligent
procedures. The most commonly utilized and supported
enabler is the Java Agent DEvelopment platform (JADE) [33].
For the semantic representation of the knowledge and
reasoning concepts, we select the Discrete Event Calculus
Knowledge Theory (DECKT) – an implementation of Event
Calculus (EC) in the rule engine Jess [34]. Reasoning is
formed as an agent’s behavior and is embodied in JADE
agents.
The various agents can also exchange knowledge. The FIPA
standardized Agent Communication Language (ACL) [35] is
selected for this purpose. The agents communicate well-
formed messages based on the ACL format containing events,
alarms, or other reports and notifications for the evaluated
system.
C. CE Decision Making
In the telecom CE-IoT scenario, we deploy JADE agents for
administrating distinct subsystems of the examined
organization (the abovementioned cloud services platform
components). The agents receive messages from their
underlying IoT devices and process them based on the
DECKT-enabled reasoning process. These local agents are
close to the edge system and perform the required embedded
intelligence.
For this demonstration setting, the smart agents model the
Location, Condition, and Availability (LCA) perspectives of
the system. The agents are aware for each component’s
location, its working condition, and the availability. More
specifically, the three properties are determined as:
• Location: the physical location of the component (i.e.
the room where a server is installed)
• Condition: good or requiring maintenance, repair,
refurbishment, and recycling
• Availability: working, ready for reuse, or broken
Once entered in the organization, the component’s location
is recorded by the system operator. Initially, it is of good
operational condition and is either placed and working
immediately or waits to be used. Nevertheless, when a
component is integrated in the system and performs the
desirable functionality, the operator determines the
maintenance strategy. Three maintenance types are defined as:
• Periodic: implemented as a triggered event in the
accountable agent’s reasoning behavior after a specified
period (e.g. every three months)
• Decay: the agent collects usage data for the component
and raises a maintenance event when it exceeds a
threshold (e.g. after 1000 hours of operation)
• Or a combination of the two.
Meanwhile, if the component is broken, the operator specifies
the new condition as requiring repair/refurbishment/recycling
and the maintenance events are postponed. If the component is
repaired, the good operational condition is restored and the
component is available for reuse.
The local agents also exchange this information to similar
agents that are placed in the upper system layers and the
backend infrastructure that implement the high level business
intelligence. This includes, among others, the presentation of
integrated reports to the high level management or even the
automated communication with the relevant agents from
cooperated businesses and the planning of adequate and timely
equipment supplies.
Except from the LCA properties, the agents also gather the
data for the ML algorithms that enable the green computing
features. The ML procedures and detailed in the next section.
D. Scalability and Performance
As concerns the scalability and performance of the agent
artificial intelligence and reasoning, Jess implements an
efficient pattern matching method. The computational
complexity is linear to the working memory size, and more
specifically, it is strongly affected by the number of facts that
are required in order to model the components of the
underlying subsystem. In the examined scenario, each
component requires around 10-20 facts to be modelled,
resulting only in a low processing overhead in the nanosecond
range. Fig. 2 depicts the agent GUI, 4 emulated components
and the triggered event where the component C1 needs
maintenance.
Fig. 2. The CE-Iot agent GUI.
We demonstrate the proposed CE-IoT framework in an
emulated environment, where BeagleBone embedded devices
imitate the operation and the LCA properties of the
organization’s components. Totally, we deploy 4 devices that
communicate information to the relevant agent that runs on a
local computer. On average, it takes 57 ms to perform a
reasoning operation, requiring 40 MB RAM and 1.8 MB code
size.
IV. G
REEN
C
OMPUTING
&
E
DGE
I
NTELLIGENCE
In this subsection, we further extend the main monitoring
framework with machine learning (ML) techniques (e.g. [36],
[37]). Lightweight ML algorithms at the edge process the
collected data from the local agents. Then, predictions are
made concerning the maintenance of the monitored assets.
Data analytics are also performed at the backend with
mainstream ML technologies enhancing the overall business
logic and the applied CE strategies.
A. Smart sensing and prediction
Smart monitoring/prediction is heavily based on the AI
embedded sensing platform used in the field (layer) to sense
the physical world, obtain the raw signals and thus, collecting
the data in a parallel fashion. These units also include self-
learning, self-adaptive, self-configuration, and self-
management capabilities. The platform utilizes a sophisticated
design methodology imposing a hierarchical, naturally
distributed, thus scalable, processing chain.
Towards this twofold direction (of monitoring and
prediction), the set of AI sensing units can be placed in the
specific locations of the ICT infrastructure measuring a wide
range of qualitative characteristics such as traffic load, CPU
usage, memory consumption, room temperature or humidity
etc. The robustness of the ICT network can be reflected in the
aforementioned parameters. Dependability could be
compromised by equipment aging, overworking, sudden
environmental conditions, etc. Those threats can lead to
failures resulting in significant economic losses.
The critical infrastructure is monitored through the AI
embedded sensing platform. The collected data are processed
locally and then transmitted to the upper IoT architecture
level, which includes a data center capable of further
analyzing the acquired information.
During monitoring, raw/physical data are collected and
processed (by each sensing unit) in a real-time fashion. It
should be highlighted that monitoring is strongly related with
event detection and thus, online machine learning algorithms
can be applied in order to identify specific abnormal patterns
within the streaming data. Moreover, feature extraction and
feature selection can be applied in each sensing unit (in a
distributed manner) to reveal the hidden inter-sensor data
information structure. The estimated features could then be
transmitted to the backend for further processing, i.e.,
performing event detection. Several off-the-shelf approaches
could be adopted for event or anomaly detection such as
Hidden Markov Models (HMMs), autoencoders, Long Short
Term Memory (LSTM) autoencoders, k-NN (nearest
neighbor), and Local Outlier Factor (LOF) ([38], [39]).
Moreover, data fusion-based event detection algorithms could
be devised to achieve higher detection rates by exploiting
multisensory data fusion.
As aforementioned, the analyzed data are sent to the
backend for prediction purposes. The aggregated data can be
exploited for long-term prediction towards identifying specific
time of failures, estimating the parts of the infrastructure that
are about to malfunction and in general ensuring that potential
problems are discovered well before they actually occur.
Prediction of critical events can be performed using decision
trees, support vector machines (SVMs), deep learning and
other machine learning time-series forecasting/prediction
techniques ([38], [39]).
It is obvious from the description above that monitoring is
taking place at the edge of the IoT system related with online,
real-time detection, while prediction is related to short or long-
term estimation of critical events. The extracted special data
transferred from the lower edge level to the upper cloud/data
center level can be used for prediction. However, continuous
data aggregation can be exploited to retrain the predictor and
an updated version of the monitoring algorithm can be
obtained by sending down data that are computed during the
prediction procedure. As a result, this information from the
upper level can be used to obtain an updated robust
monitoring version at the edge level, since we can determine
what type of sensed data is needed to be acted on quickly and
what can be analyzed at a later time. In some cases, it may be
determined that certain data points are not worth capturing and
analyzing at all, particularly in instances where enormous
amounts of data are being generated.
The outcome of the local computations is a decision on how
to act in case of a detected event. In some cases, the decision
making is going to be affected by the feedback from the
backend as a result of a long-term monitoring and prediction
procedure of the whole system. The expected long-term
outcome is to minimize the risk of infrastructure failure and
accomplish an unobstructed operation.
Finally, the stakeholders can update the policy accordingly
if needed and disseminate it in a top-bottom fashion though
the IoT system. More specific, stakeholders such as IoT
vendors or service providers can exploit the processed data in
order e.g. to replace failed or malfunctioned IoT equipment
which is critical for the smart infrastructure
monitoring/prediction capabilities.
V. B
USINESS
D
ECISION
M
AKING
P
OLICY
The CE-IoT framework implements a utility-driven
operation by exploiting the aforementioned smart sensing and
data analysis procedures. A decision support system (DSS) is
utilized for this purpose. The Analytic-Synthetic method,
called Stochastic UTA (UTilites Additives – Additive utility)
[40], evaluates and rates the alternative choices (Sell, Dispose,
Repair). The utilization method is described below.
A. The CE decision making model
The presented stochastic method is used in the
interconnected environment with sensors/monitors on the
products sending data about their lifecycle to the backend. The
CE-IoT collects information about the “health” of each product
and diagnoses if it needs ‘repair’, ‘buy new’ or ‘sold as an asset
to another consumer’. The products are considered as assets
and can reentered to the CE with a price depending from the
market request.
The CE-IoT agents send data to a server which also collects
data from other CE users and from the market. These data are
used by the DSS using the Analytic-Synthetic approach in
which a Decision Maker has to take the Decision. As is shown
in Fig. 3, a product is considered as an asset. The
consumers/businesses buy this product due to their needs.
Fig. 3. The CE-IoT business model for a single user.
The Stochastic Method is applied when we want to create a
Hierarchy Model for our decisions. The Stochastic UTA
method implements this Analytic-Synthetic approach [40]. The
philosophy of this approach in the context of the multi-criteria
analysis lies in the estimation of a preference model, which
emerges as a conclusion from a given expression of the overall
preference on alternative activities. The aim is to provide
support for decision-making actions using operational models
[41].
In the traditional form of the synthetic approach followed
by the majority of the multi-criteria analysis problems, the
composition model of the individual criteria is a priori known,
while the overall preference is unknown [40].
The Stochastic UTA method exhibits the total utility of
each of the three alternatives: sell, buy or repair. The decision
maker takes the final decision about what to do with the
product. Also, indexes are included that reveal the stability of
the proposed model, like the Average Stability Index [40].
Fig. 4 illustrates an emulated iteration of the CE-IoT loop.
As aforementioned our organization (Cablenet) uses its
equipment to serve end-customers. When the system is
upgraded, the related machinery is sold to start-ups in UK.
Similarly, these companies can re-sell the CE assets when their
utility goes beyond a threshold. Afterwards, the assets are
gifted to third-world countries that use older technologies in
their networks. When the products are no longer working, they
are recycled in businesses that are close to these areas. Finally,
a subset of useful materials are exported back to the
manufacturers and the loop starts again.
Fig. 4. The CE-IoT emulated application.
VI. C
ONCLUSION
The ongoing interplay of the CE and IoT sectors enables
new business opportunities while promoting green
technologies. This paper presents the CE-IoT framework,
which forms an enhanced recycle-rewarding system for ICT
organizations. Except from recycling, the overall approach
extends the productive period of the electronic assets, enhances
the e-waste management, and improves the economic
operations of the involved stakeholders. A preliminary
implementation on a real equipment is detailed, which also
involves AI, ML, and DSSs. The CE-IoT is applied on a
medium-size ICT provider and the overall trace of a real
supply-chain scenario is described. The ML techniques
promote green computing/networking policies on the
organization that currently operates the valuable assets, while
the AI and DSS procedures administrate the CE-IoT
functionality and provide valuable insights for the decision
making staff.
However, as the cooperating organizations exchange the
CE assets, we need a transparent way to administrated and
authorize the performed actions. Thus, the usage of
blockchains as a distributed ledger that will maintain this
information is considered as future work.
A
CKNOWLEDGMENT
This work has received funding from the European Union
Horizon’s 2020 research and innovation programme H2020-
DS-SC7-2017, under grant agreement No. 786890 (THREAT-
ARREST), as well as the Marie Skodowska-Curie Actions
(MSCA) Research and Innovation Staff Exchange (RISE),
H2020-MSCA-RISE-2017, under grant agreements No.
777855 (CE-IoT) and No. 778229 (Ideal Cities).
R
EFERENCES
[1] C. M. Angelopoulos, S. Nikoletseas, and G. C. Theofanopoulos, “A
smart system for garden watering using wireless sensor networks,” 9
th
MobiWac, ACM, Florida, USA, pp. 167-170, Nov. 2011.
[2] C. M. Angelopoulos, G. Filios, S. Nikoletseas, and D. Patroumpa, “A
holistic IPv6 test-bed for smart, green buildings,” IEEE International
Conference on Communications (ICC), IEEE, Budapest, Hungary, pp. 1-
12, Nov. 2013.
[3] Y. Kalmykova, M. Sadagopan, and L. Rosado, “Circular economy –
From review of theories and practices to development of implementation
tools,” Resources, Conservation and Recycling – Sustainable Resource
Management and the Circular Economy, Elsevier, vol. 135, issue 1, pp.
190-201, Aug. 2018.
[4] A. Heshmati, “A review of the circular economy and its
implementation,” International Journal of Green Economics,
Inderscience, vol. 11, issue 3/4, pp. 251-288, Feb. 2018.
[5] A. Urbinati, D. Chiaroni, and V. Chiesa, “Towards a new taxonomy of
circular economy business models,” Journal of Cleaner Production,
Elsevier, vol. 168, issue 1, pp. 487-498, Dec. 2017.
[6] G. Schulze, “Growth Within: A Circular Economy Vision for a
Competitive Europe,” Ellen MacArthur Foundation and the McKinsey
Center for Business and Environment, pp. 1-22, Nov. 2016.
[7] J. Sarkis and H. Zhu, “Information technology and systems in China’s
circular economy: implications for sustainability,” Journal of Systems
and Information Technology, Emerald, vol. 10, issue 3, pp. 202-217,
2008.
[8] A. Khalamayzer, ‘7 companies advancing the circular economy by
selling products as a service,’ GreenBiz, Feb. 2018,
https://www.greenbiz.com/article/7-companies-offering-circular-
economy-service .
[9] I. Askoxylakis, “A framework for pairing circular economy and the
Internet of Things,” IEEE International Conference on Communications
(ICC), IEEE, Kansas City, MO, USA, May 2018.
[10] J. Soldatos, N. Kefalakis, M. Serrano, and M. Hauswirth, “Design
principles for utility-driven services and cloud-based computing
modelling for the Internet of Things,” International Journal of Web and
Grid Services, Inderscience, vol. 10, issue 2-3, pp. 139-167, Mar. 2014.
[11] K. B. Slimane and D. Chaney, ‘Towards marketing of functionality:
insights from Michelin and Hilti,’ Journal of Strategic Marketing, Taylor
& Francis, vol. 23, issue 3, pp. 224-237, 2015.
[12] State of Green, “Circular consumption through innovative business
models,” Nov. 2017, https://stateofgreen.com/en/partners/state-of-
green/news/circular-consumption-through-innovative-business-models/ .
[13] K. Lyndegaard, ‘Combining spatial and network analysis: a case study
of the GoMore network,’ Esri European User Conference (EUC), Esri,
Split, Croatia, pp. 1-11, Oct. 2014.
[14] A. T. S. Worth and B. Escartin-Claveria, ‘Alstom’s approach to railway
condition monitoring,’ 6
th
IET Conference on Railway Condition
Monitoring (RCM), IET, Birmingham, UK, pp. 1-6, Sept., 2014.
[15] M. Orehek and C. Robl, ‘Model-based design of an ECU with data- and
event-driven parts using auto code generetion,’ IEEE ICRA, IEEE,
Seoul, South Korea, pp. 1346-1351, May 2001.
[16] Information & Communication Technology (ICT), 1
st
International
Workshop on Green Wireless 2008 (W-GREEN), Lapland, Finland,
2008, http://www.cwc.oulu.fi/workshops/W-Green2008.pdf.
[17] sustainableIT, “Green computing makes good business sense,” 2011,
http://www.sustainableit.com/green-computers-make-good-business-
sense/.
[18] G. Cecere, N. Corrocher, C. Gossart, and M. Ozman, “Technological
pervasiveness and variety of innovators in Green ICT: a patent-based
analysis,” Research Policy, Elsevier, vol. 43, issue 10, pp. 1827-1839,
Dec. 2014.
[19] Z. S. Andreopoulou, “Green Informatics: ICT for green and
sustainability,” Agricultural Informatics, vol. 3, issue 2, pp. 1-8, Dec.
2012.
[20] A. Ozturk, K. Umit, I. T. Medeni, B. Ucuncu, M. Caylan, F. Akba, and
T. D. Medeni, “Green ICT (Informaiton and Communication
Technologies): a review of academic and practitioner perspectives,”
International Journal of eBusiness and eGoverment Studies, The Social
Sciences Research Society, vol. 3, issue 1, pp. 1-16, 2011.
[21] L.-D. Radu, “Determinants of green ICT adoption in organizations: a
theoretical perspective,” Sustainability, MDPI, vol. 8, issue 8, article
731, pp. 1-16, Jul. 2016.
[22] X. Qiu, H. Luo, G. Xu, R. Zhong, and G. Q. Huang, “Physical assets and
service sharing for IoT-enabled Supply Hub in Industrial Park (SHIP),”
International Journal of Production Economics, Elsevier, vol. 159, issue
1, pp. 4-15, Jan. 2015.
[23] Z. Pang, Q. Chen, W. Han, and L. Zheng, “Value-centric design of the
Internet-of-things solutions for food supply chain: value creation, sensor
portfolio and information fusion,” Information Systems Frontiers,
Springer, vol. 17, issue 2, pp. 29-319, Apr. 2015.
[24] A. Aminoff and O. Kettunen, “Sustainable supply chain management in
a circular economy – towards supply circles,” International Conference
on Sustainable Design and Manufacturing (SDM), Springer, SIST, vol.
52, pp. 61-72, Apr. 2016.
[25] G. Alexandris, S. Alexaki, V. Katos, and G. Hatzivasilis, “Blockchains
as enablers for auditing cooperative circular economy networks,” 23
rd
International Workshop on Computer Aided Modeling and Design of
Communicaiton Links and Networks (CAMAD), IEEE, Barcelona,
Spain, pp. 1-7, Sept. 2018.
[26] J. Kallio, M. Antikainen, and O. Kettunen, “A tool rental service
scenario,” 11
th
UBICOMM, Barcelona, Spain, pp. 60-65, Nov. 2017.
[27] A. M. Adeola and M. Othman, “An overview of ICT waste
management,” 7
th
International Conference on Networked Computing,
IEEE, Gyeongsangbuk-do, South Korea, pp. 109-115, Sep. 2011.
[28] M. H. Masud, N. B.T. A. Malik, and Md. K. Shukran, “Evaluating the
effectiveness of energy efficient and eco-leveling ICT infrastructures for
environmental sustainability,” 3
rd
International Conference on
Engineering and ICT (ICEI), Melaka, Malaysia, pp. 1-5, Apr. 2012.
[29] J. Sanekata, M. Tani, T. Nishi, Y. Hamatsuka, and H. Sugai,
“Environmental impact assessment logic for ICT systems named SI-
LCA and its case example,” IEEE International Symposium on
Electronics and the Environment, IEEE, Orlando, FL, USA, pp. 6-11,
May 2007.
[30] K. Fysarakis, G. Hatzivasilis, I. Askoxylakis, and C. Manifavas, “RT-
SPDM: Real-Time Security, Privacy and Dependability Management of
Heterogeneous Systems,” In: Tryfonas T., Askoxylakis I. (eds) Human
Aspects of Information Security, Privacy, and Trust. HAS 2015. Lecture
Notes in Computer Science, vol 9190. Springer, Cham, pp. 619–630,
2015.
[31] K. Fysarakis, G. Hatzivasilis, C. Manifavas, and I. Papaefstathiou,
“RtVMF: A Secure Real-Time Vehicle Management Framework,” IEEE
Pervasive Comput., vol. 15, no. 1, pp. 22–30, Jan. 2016.
[32] G. Hatzivasilis, I. Papaefstathiou, D. Plexousakis, C. Manifavas, and N.
Papadakis, “AmbISPDM: managing embedded systems in ambient
environment and disaster mitigation planning,” Applied Intelligence,
Springer, pp. 1–21, 2017.
[33] JADE framework: http://jade.tilab.com/
[34] T. Patkos and D. Pleousakis, “DECKT: epistemic reasoning for ambient
intelligence,” ERCIM News magazine, Special Theme: Intelligent and
Cognitive Systems, vol. 2011, issue 84, pp. 1-30, Jan. 2011.
[35] FIPA ACL: www.fipa.org/repository/aclspecs.html .
[36] P. V. Klaine, M. A. Imran, O. Onireti, and R, D. Souza, “A survey of
machine learning techniques applied to self-organizing cellular
networks,” IEEE Communications Surveys & Tutorials, IEEE, vol. 19,
issue 4, pp. 2392-2431, Oct. 2017.
[37] P. Stone and M. Veloso, “Multiagent systems: a survey from a machine
learning perspective,” Autonomous Robots, Springer, vol. 8, issue 3, pp.
345-383, June 2000.
[38] M. Langkvist, L. Karlsson, and A. Loutfi, “A review of unsurpervised
feature learning and deep learning for time-series modeling,” Pattern
Recognition Letters, Elsevier, vol. 42, issue 1, pp. 11-24, June 2014.
[39] S. B. Kotsiantis, I. D. Zaharakis, and P. E. Pintelas, “Machine learning: a
review of classification and combining techniques,” Artificial
Intelligence Review, vol. 26, issue 3, pp. 159-190, Nov. 2006.
[40] E. Grigoroudis and Y. Siskos, “Customer satisfaction evaluation,”
International Series in Operations Research & Management Science,
Springer, vol 139, pp. 1-308, 2010.
[41] E. Jacquet-Lagreze and Y. Siskos, “Preference disaggregation: 20 years
of MCDA experiene,” European Journal of Operational Research,
Elsevier, vol. 130, issue 2, pp. 233-245, Apr. 2001.
