POLITEHNICA University of Bucharest [602375]

POLITEHNICA University of Bucharest
The Doctoral School on Automatic Control and Computers

RESEARCH REPORT

Scientific coordinator
Prof. Dr. Ing. Valentin Sg ârciu

PhD student: [anonimizat]
2016

2

POLITEHNICA University of Bucharest
The Doctoral School on Automatic Control and Computers

Scientific report no. 1

Internet of Things and Artificial Ingelligence

Scientific coordinator PhD student: [anonimizat]. Valentin Sg ârciu Aurel -Dorian Floarea

2016

3
Table of Contents
1 Introduction ………………………….. ………………………….. ………………………….. ………………………….. 4
2 Research ………………………….. ………………………….. ………………………….. ………………………….. …… 5
2.1 The state of the art ………………………….. ………………………….. ………………………….. …………….. 5
2.2 Hardware ………………………….. ………………………….. ………………………….. ………………………….. 9
2.3 Software ………………………….. ………………………….. ………………………….. …………………………. 12
2.4 Security ………………………….. ………………………….. ………………………….. ………………………….. 14
2.5 Communicatio n ………………………….. ………………………….. ………………………….. ……………….. 16
2.6 Studied solutions ………………………….. ………………………….. ………………………….. ……………… 18
2.7 Proposed solutions ………………………….. ………………………….. ………………………….. …………… 20
3 Conclusions ………………………….. ………………………….. ………………………….. ………………………… 23
4 Annexes ………………………….. ………………………….. ………………………….. ………………………….. …. 24
4.1 Annex 1: Smart Refrigerator poster ………………………….. ………………………….. ………………… 24
5 References ………………………….. ………………………….. ………………………….. ………………………….. 25

4
1 Introduction
The Internet of Things (IoT) is a unique phenomenon because it brings together many
strands of different tehnilogies from different industries. The challenge comes from the common
requirements for connectivity, security and data handling that cut across all businesses. However,
the IoT must also allow for sector specific applications and solutions. Therefore collaborati on is
highly necessary, even essential in some cases. In other cases convergence of telecoms, IT and
consumer electornics, along with softwware, applications and data handling is required.
This report presents the current trends in IoT research propelled b y applications and the
need for convergence in several interdisciplinary technologies. The following sections of this
research report present: a state of the art and common research directions throughout the IoT
industry in section 2.1, research on recent progresses in hardware related IoT challenges in
section 2.2, software related IoT challenges in 2.3 and both security and communication IoT
challenges in sections 2.4 and 2.5. A few functional IoT systems have been analyzed and are
presented in section 2. 6 and sections 2.7 outlines a proposed IoT system that was presented at the
ECAI conference that took place in Ploiesti, Romania at the end of June, 2016. This report
concludes with a discussion on future challenges and trends in section 3.

5
2 Research
The Io T refers to the set of devices and systems that interconnect real world sensors and
actuators to the Internet. This includes many different systems, including: Internet connected
cars; wearable devices including health and fitness monitoring devices , watc hes, and even
human implanted devices; smart meters and smart objects; home automation systems and
lighting controls; smartphones which are increasing being used to measure the world around
them; and wireless sensor networks that measuring weather, flood d efenses, tides and more.
There are of course two key aspects to the IoT: the devices themselves and the server -side
architecture that supports them. In fact there is often a third -category as well: in many cases there
may be a low power gateway that perfo rms aggregation, event processing, bridging etc. that
might sit between the device and the wider Internet.
In both cases the devices probably have intermittent connections based on factors such as
GPRS connectivity, battery discharging, radio interference, or simply being switched off.
There are effectively three classes of device. The smallest devices have embedded 8 -bit
System -On­Chip (SOC) controllers. A good example of this is the Open Source Hardware
platform Arduino: e.g the Arduino Uno platform and other 8 -bit Arduinos. These typically have
no operating system.
The next level up are the systems based on Atheros and ARM chips that have a very
limited 32 -bit architecture. These often include small home routers and derivati ves of those
devices. Commonly these run a cut -down or embedded Linux platform such as OpenWRT, or
dedicated embedded operating systems. In some cases they may not use an OS. An example of
this would be the recently announced Arduino Zero, or the Arduino Y un.
The most capable IoT platforms are full 32 -­‐bit or 64 -­‐bit computing platforms. These
systems, such as the Raspberry Pi or the BeagleBone, may run full a full Linux OS or another
suitable Operating System such as Android. In many cases these are eith er mobile phones or
based on mobile -phone technology. These devices may also act as gateways or bridges for
smaller devices: for example if a wearable connects via Bluetooth Low Energy to a mobile
phone or Raspberry Pi, which then bridges that onto the wid er Internet.
2.1 The state of the art
A reference architecture presente d in [1] consists of a set of layers. Each layer performs a
clear function. Layers can be instantiated by specific technologies. There are also some cross –
cutting/v ertical layers such as security/identity management.

6
The layers are: external communications (such as Web/Portal, Dashboard, APIs), event
processing and analytics (including data storage), aggregation or bus layer (ESB and Message
Broker), device communica tions and the devices. The cross -cutting layers are the device
management layer and the identity and access management.

Figure 1 reference architecture
The bottom layer of the architecture is the device layer. Devices can be of v arious types,
but in order to be considered IoT devices, they must have some communications that either
indirectly (such as ZigBee devices connected via a ZigBee gateway, Bluetooth or Bluetooth Low
Energy devices connecting via a Mobile Phone, Devices comm unicating via low power radios to
a Raspberry Pi ) or directly (such as Arduino with Arduino Ethernet connection, Raspberry Pi
connected via Ethernet or Wi -Fi, Intel Galileo connected via Ethernet of Wi -Fi) attaches to the
Internet.
Each device typically needs an identity. The identity may be one of the following: a unique
identifier (UUID) burnt into the device (typically part of the System -on-Chip, or provided by a
secondary chip, a UUID provided by the radio subsystem (such as Bluetooth identifier, Wi -Fi
MAC address), an OAuth2 Refresh/Bearer Token (this may be in addition to one of the above) or
an identifier stored in non -volatile memory such as EEPROM.
The communication layer supports the connectivity of the devices. There are multiple
potential prot ocols for communication between the devices and the cloud. The most well known

7
three potential protocols are: HTTP/HTTPS (and RESTful approaches on those), MQTT 3.1 /
3.1.1 and Constrained Application Protocol (CoAP).
HTTP is well known, and there are obvi ously many libraries that support it. Because it is a
simple text -based protocol, many small devices such as 8 -bit controllers can partially support the
protocol – for example enough code to POST or GET a resource. The larger 32 -bit based devices
can utili ze full HTTP client libraries that properly implement the whole protocol.
MQTT is a publish -subscribe messaging system based on a broker model. The protocol has
a very small overhead (as little as 2 bytes per message), and was designed to support lossy and
intermittently connected networks. MQTT was designed to flow over TCP. In addition there is
an associated specification designed for ZigBee -­‐style networks called MQTT -SN (Sensor
Networks).
The Constrained Application Protocol (CoAP) is a protocol from the IETF that is designed
to provide a RESTful application protocol modeled on HTTP semantics, but with a much smaller
footprint and a binary rather than text -­‐based approach. CoAP is a more traditional client -server
approach rather than a brokered approa ch. CoAP is designed to be used over UDP.
An important layer of the architecture is the layer which aggregates and brokers
communications. This is an important layer for three reasons: the ability to support an HTTP
server and/or an MQTT broker to talk to the devices; the ability to aggregate and combine
communications from different devices and to route communications to a specific device
(possibly via a gateway) and the ability to bridge and transform between different protocols – for
example to offer HTT P-based APIs that are mediated into an MQTT message going to the
device.
The aggregation/bus layer provides these capabilities as well as adapting into legacy
protocols. The bus layer may also provide some simple correlation and mapping from different
correlation models (e.g. mapping a device ID into an owner’s ID or vice -versa).
The Event Processing and Analytics Layer takes the events from the bus and provides the
ability to process and act upon these events. A core capability here is the requirement to s tore the
data into a database. The Device Management is handled by two components. A server side
system (the Device Manager) communicates with the devices via various protocols and provides
both individual and bulk control of devices. It also remotely mana ges the software and
applications deployed on the device. It can lock and/or wipe the device if necessary. The Device
Manager works in conjunction with the device management agent. There are multiple different
agents for different platforms and device type s.
The external communications layer offers a way for these devices to communicate outside
of the device -oriented system. This includes three main approaches. Firstly, the network needs a

8
web-based front -ends and portals that interact with devices and with the event -processing layer.
Secondly, it needs dashboards that offer views into the analytics and event processing. Finally, it
needs to interact with systems outside this network using machine -to-machine communications
(APIs). These APIs need to be manag ed and controlled and this happens in an API Management
system.
The Device Management is handled by two components. A server side system (the Device
Manager – DM) communicates with the devices via various protocols and provides both
individual and bulk con trol of devices. It also remotely manages the software and applications
deployed on the device. It can lock and/or wipe the device if necessary. The Device Manager
works in conjunction with the device management agent. There are multiple different agents f or
different platforms and device types. The Device Manager also needs to maintain the list of
device identities and map these into owners. It must also work with the Identity and Access
Management layer to manage access controls over devices.
The final la yer is the identity and access management layer. This layer needs to provide the
following services: OAuth2 token issuing and validation; other identity services including
SAML2 SSO and OpenID Connect support for identifying inbound requests from the Web l ayer;
XACML PDP; directory of users (e.g. LDAP); policy management for access control (Policy
Control Point).
The challenges and enablers for enabling things and objects to become more reliable, more
resilient, more autonomous and smarter is presented in [2]. This is realized through a set of
decentralized management mechanisms for targeting IoT -based systems in order to enable the
exploitation of millions of devices. Also, an architecture is presented that allows things to learn
based on others experiences. The proposed architectural approach also introduces situational
knowledge acquisition and analysis techniques in order to make things aware of conditions and
events affecting IoT -based systems behavior. The goal is to enable the provision of a scalable
and privacy -aware IoT infrastructure that considers social relationships among objects, while
being able to self -adapt itself according to environmental context changes in a decentralized and
real-time way. Core to the architecture are the ability of things to learn and evolve by exploiting
things social -behavior and use it as a basis to share and exchange experiences, as a means to
make things smarter and more autonomous. The proposed enabling technologies allow new
device platform s to be integrated at any time and become immediately available via the
decentralized management mechanisms that base coordination decisions on the role and
participation scheme of such devices in and across IoT networks.
The limitations of associated devi ces in the IoT in terms of storage, network and
computing are described in [3], which also presents a survey of integration components (like

9
Cloud platforms, Cloud infrastructures and IoT Middleware), technologies, applications and
research challenges [4]. The requirements of complex analysis, scalability, and data access,
require a technology like Cloud Computing to supplement this field. Moreover, the IoT can
generate large amounts of varied data and q uickly when there are millions of things feeding data
to Cloud Computing. The latter is a clear example of Big Data, that Cloud Computing needs to
take into account. In addition, some integration proposals and data analytics techniques are
surveyed as wel l as different challenges and open research issues are pointed out.
Necessary background and fundamentals are presented in [5] to understand current efforts
in IoT, Web of Things (WoT) and Social Web of Things (SWoT) by reviewi ng key enabling
technologies. These efforts were investigated in detail from several different perspectives such as
architecture design, middleware, platform, systems implementation, and application in hand.
Moreover, a large number of platforms and applic ations were analyzed and evaluated from
various alternatives that have become popular during the past decade. Finally, the authors
addressed associated challenges and highlight potential research to be perused in future.
An Analytic Hierarchy Process (AHP) model was proposed in [6] for assessing the
viability of IoT applications, which consists of 11 technology (technical practicalit, technical
reliability, cost efficiency and standardization), market (market demands, user accep tance,
business model and ecosystem building), and regulation factors (industry regulation, consumer
protection regulation and governmental support). The model was applied to assess and compare
the prospect of three IoT applications (IoT healthcare, IoT lo gistics, and IoT energy
management), inviting the perception of 31 ICT experts. The results showed that IoT logistics is
the most promising IoT application, due to its strong market potential. IoT healthcare was
perceived to have strength in wide consumer market demand, but reliability issues and regulation
issues were posed. Meanwhile, IoT energy management showed strengths only in governmental
support. The results also imply that it would be wiser for the government to focus on eliminating
regulatory bott lenecks.
A thorough and high -level analysis about how M2M and the IoT are being implemented
and deployed in various industries is presented in [7]. The book outlines the global context of
M2M and the move towards IoT , including technology and business drivers, presents the
technology building blocks of M2M and IoT solutions and covers real -world implementation
examples of M2M and IoT solutions.
2.2 Hardware
One of the most challenging problems IoT faces is its dependabili ty (reliability and
availability), since a device failure might put people in danger or result in financial loss. The lack

10
of a design tool for assessing the dependability of IoT applications at the early planning and
design phases prevents system designer s from optimizing their decisions so as to minimize the
effects of such faults on the network devices. To overcome this issue [8] proposes a
dependability evaluation tool for IoT applications, when hardware faults and permanent link
faults are considered. The authors surveys some of the most relevant research on reliability and
availability evaluation for the IoT and give a brief introduction to Fault Tree Analysis (FTA) and
the basic concepts used in their proposal. They then d escribe their proposed tool for the
dependability evaluation of IoT applications which is organized into four modules: an IoT
network editor, a fault tree generator, a fault tree editor, and a visual reporting tool. The
dependability evaluation starts by p roviding information about the network topology in the first
module. After that, the device's failure and repair process and network failure condition are
defined. The fault tree generator module uses this information to create a fault tree corresponding
to the IoT network. This fault tree can then be edited and visualized in the fault tree editor.
Finally, the metrics of interest are evaluated in the visual reporting tool.
Automating actions based on collected context from IOT -controlled systems is one of the
most important requirements of IOT systems; [9] seeks to satisfy this requirement by offering a
context -aware service framework on top of IOT controlled systems. The fault management
process in electric power distribution networks is taken here as a case study and used for
implementing their proposed f ramework. The authors discuss the different aspects that need to
be covered in such a framework and its components. They used the concept of context to refer to
the signals collected by an IOT controlled system along with specialists' GPS locations and
workloads. Despite the importance of context adaptation, the research was focused more on
understanding the different components needed to solve the studied research problem.
Additionally, the authors proposed the adoption of services architecture to provide useful
services to both specialists and their managers. The obtained results from simulating and testing
the framework showed a significant improvement in the task management process compared to
the traditional approach.
Connecting trillions of smart dynam ic objects to the Internet which interact and collaborate
together independently from any physical location necessitates the proposal of an adequate
architecture and routing process. [10] proposed a novel design and overlay arc hitecture that
fulfills the Internet of Things requirements. Their approach is mainly based on Distributed Hash
Table protocols to afford the required flexiblity and to handle efficiently mobility and churn
cases and it includes a complete architecture tha t integrates things identifiers assignment,
routing, churn and mobility management.

11
Since smart objects are placed massively in every place in the world, it is beneficial to
partition the network into two main parts: the first part is made up of access poi nts and the
second part is made up of smart objects. Access points have an important memory size and
important processing capabilities. They are static and do not have any energy constraints since
they have access to power. This enables them to have an ext ended coverage.
Each access point is responsible of smart objects in its coverage. It stores important
information related to these smart objects. Each node should have a location -independent
identifier in order to be reachable wherever it is in the world and also a Virtual Cord Protocol
(VCP) routing table that maintains information about its physical/virtual neighbors.
The churn is the continuous and massive nodes joining and departure. This phenomenon is
frequent in Internet of Things. That's why an adeq uate protocol should be employed to handle
churns efficiently. When a node joins the network, it starts by broadcasting HELLO messages
containing its global identifier. Each node that receives this message, responds it by a Hello
response message containin g its local and global identifiers as well as the identifier of the access
point at which it belongs. If the joining node receives only messages from nodes belonging to the
same network (having the same access point identifier), it joins this network using VCP joining
procedure and computes its local identifier. Then it sends a notification message containing its
local and global identifiers, to the access point responsible of this network using VCP. When the
access point receives the notification message, it adds the pair (global address, destination
address) of the joining node to the addresses table.
Their proposed approach uses a novel mobility management protocol called MOBYDHT.
When the mobile node still belongs to the same local network, MOBYDHT uses the same
procedure defined in VCP to update the local identifier of the mobile node. When the mobile
node does not receive HELLO messages from its local network nodes, this means that it should
change its corresponding local network and its corresponding a ccess point.
Several problems of IoT and WoT which have not been fully solved are outlined in [11].
The first problem arises when smart devices operate in a dynamic environment. There are
multiple scenarios in which links betwe en smart devices cannot be established in advance, as it
would have been possible in a mostly static environment such as a smart home, thus a
mechanism for discovering available devices is required.
The second problem is specific for Web of Things and it r egards asynchronous
communication patterns. Many internet -connected devices use a very simple data transfer
strategy: when a new piece of information is available, it is sent to a web -service. This pattern
perfectly fits original web standards. However, wh en data needs to be pushed asynchronously to
a remote device, some other schemes are required.

12
2.3 Software
IoT platforms are the central piece in the Internet of Things architecture that connect the
real and the virtual worlds and enable communication between objects. The authors of [12] offer
a more sophisticated overview over any IoT platform. According to them, a platform should
platform consists of a variety of important building blocks: connectivity and normalization,
device m anagement, database, processing and action management, analytics, visualization,
additional tools, and external interfaces.
Every IoT platform starts with a connectivity layer. It has the function of bringing different
protocols and different data formats into one “software” interface. This is necessary in order to
ensure all devices can be interacted with and data is read correctly.Advanced devices usually
provide an API that allows for a standardized communication interface to the platform. If this is
not the case then software agents have to be developed and installed on the hardware in order to
enable the IoT platform to establish a stable connection.
The device management module of an IoT platform ensures the connected objects are
working properly and i ts software and applications are updated and running.
Data storage is a central piece in an IoT platform and must handle all the requirements
generated by the devices data: volume, variety, velocity and veracity. An IoT platform therefore
usually comes wit h a cloudbased database solution that is distributed across different sensor
nodes. It should be scalable for big data and should be able to store both structured (SQL) and
unstructured data (NoSQL).
The data that is captured in the connectivity and normal ization module and that is stored in
the database gets processed in the processing and action management module which is a rule –
based event -action -trigger that allows performance of "smart" actions based on specific sensor
data.
The analytics engine encomp asses all dynamic calculations of sensor data, from basic data
clustering to deep machine learning.
Visualization comes in the form of line -, stacked -, or pie charts, 2D – or even 3D -models.
The visualization dashboard that is available to the manager of th e IoT platform is often also
included in the prototyping tools that an advanced IoT platform provides.
Advanced IoT platforms often offer an additional set of tools for the developer and the
manager of the IoT solution. Development tools allow the IoT deve loper to prototype and test
the IoT case. Sometimes even in the form of what -you-see-is-what -you-get-editors (WYSIWYG)
that lets the developer create simple smartphone apps for visualizing and controlling connected
devices.

13
Built -in application programming interfaces (API), software development kits (SDK), and
gateways are the key to the integration of 3rd -party systems and applications.
Even though all IoT platforms contain a connectivity and normalization discussed above, if
the number of devices is very large then further optimizations must be made because widespread
deployment of devices generate enormous amounts of data which is capable of supporting an
almost unlimited set of high value proposition applications for users.
An effective context -aware me thod to cluster sensors in the form of Sensor Semantic
Overlay Networks (SSONs), inspired by ant clustering algorithm, is proposed in [13], in which
sensors with similar context information are gathered into one cluster. Firstl y, sensors are
grouped based on their types to create SSONs. Then, their meta -heuristic algorithm called
AntClust has been performed to cluster sensors using their context information. Furthermore,
useful adjustments have been applied to reduce the cost of sensor search process and an adaptive
strategy is proposed to maintain the performance against dynamicity in the IoT environment.
Experiments show the scalability and adaptability of AntClust in clustering sensors. It is
significantly faster on sensor sea rch when compared with other approaches.
Another similar sensor selection problem solved through a filter -based approach is
presented in [14]. Given a predefined list of activities, the main problem is which sensors are
most im portant to accurately recognise these activities. Their solution to the sensor selection
problem is based on an information -theoretic approach, which uses information gain to measure
the effectiveness of a sensor in classifying an activity. To accomplish t his, the authors first
determine the amount of information they can get from each sensor. Then they evaluate the
effectiveness of their proposed method on two publicly available smart home datasets .
The implementation of an algorithm to detect outliers usi ng Big Data processing is
presented in [15]. The aim of their research was to propose an outlier detection procedure using
the K -means algorithm and Big Data processing using the Hadoop platform and Mahout
implementation integrated with their chosen IoT architecture. Withing their research the authors
describe the main IoT concepts, the K -means algorithm and Big Data technology. Next, they
proposed an architecture and described its implementation details and experimental validation.
Within [16], the authors describe a formal semantic model to abstract the interaction
among the components for analyzing the deadlock connections, and then the dynamic probe
detection (DPD) algorithm is used to detect the deadlock loops. They offer a formal semantic
model using probe -based connectors, based on deadlocks in CBS and they present a case study
to better describe the strategy. Their results show that the proposed strategy can achieve both
lower proces sing cost and higher reliability.

14
2.4 Security
The best solution found to handle a massive amount of data and to offer services at the
same time is cloud computing. Yet, the wide adoption of this tehnology, especially for
application areas such as pervasive he alth care, assisted living, and smart cities, is hindered by
severe privacy concerns of the individual users. Hence, user acceptance is a critical factor to turn
this vision into reality. To address this critical factor [17] presents a comprehensive approach to
privacy for IoT cloud computing. In their research, the authors present privacy concerns of end –
users and privacy considerations of service providers that arise in a scenario similar to the ones
listed above. Also, they p resent important related work and formalize the challenges and
requirements for realizing privacy -preserving cloud -based services for the IoT. In the end they
describe he design and implementation of UPECSI, their solution for User -driven Privacy
Enforceme nt for Cloud -based Services in the IoT.
Enforcing security in IoT environments has been identified as one of the top, but also last,
barriers for realizing the vision of smart, energy -efficient homes and buildings. the risks related
to the use and potentia l misuse of information about homes and consumers but also companies
and their employees requires substantial investigation. A risk analysis applied on a smart home
automation system developed in a research project involving leading industrial actors has b een
conducted in [18]. Out of 32 examined risks, 9 were classified as low, 4 as high and most of the
identified risks were considered moderate. The severe risks are related to the software
components, as well as human behavior.
The pervasive, complex and heterogeneous properties of IoT make its security issues very
challenging. In addition, the large number of resourcesconstraint nodes makes a rigid lightweight
requirement for IoT security mechanisms. The attribute -based encrypt ion (ABE) is a popular
solution to achieve secure data transmission, storage and sharing in the distributed environment
such as IoT. However, the existing ABE schemes are based on expensive bilinear pairing, which
make them not suitable for the resources -constraint IoT applications. In [19], a lightweight no –
pairing ABE scheme based on elliptic curve cryptography (ECC) is proposed to address the
security and privacy issues in IoT. The security of the proposed scheme is based on the ECDDH
assumption instead of bilinear Diffie –Hellman assumption. By uniformly determining the criteria
and defining the metrics for measuring the communication overhead and computational
overhead, the comparison analyses with the existing ABE schemes ar e made in detail. The
results show that the proposed scheme has improved execution efficiency and low
communication costs .

15
As more end -users start using smart devices, such as smart phones or smart home
appliances, the need to have more scalable, manageabl e, understandable and easy to use access
control mechanisms increases. A description of the Capability Based Access Control (CapBAC)
system is offered in [20]. CapBAC is a system that enterprises, or even individuals, can use t o
manage their own access control processes to services and information. The proposed
mechanism supports rights delegation and a more sophisticated access control customization.
The authors are currently developing in the EU FP7 IoT@Work project for managi ng access
control to some of the project’s services. They provide a more extensive analysis and coverage
of the access control issues and of the work we are currently doing, as well as of the aspects
related to access control usability for which a capabili ty based access control system can be more
effective in end -user’ and consumer’ scenarios (e.g. domotics, everyday life, e -Health) as
compared to traditional access control approaches.
The authors of [21] consider security aspects for M2M communications and they propose a
security gateway application (SGA) including the lightweight symmetric key cryptographic
negotiation function, secure E2E M2M key exchange generation function and secure E2E M2M
messages d elivery function. The proposal of the SGA is newly suggested to improve the gateway
application (GA) of the M2M service layer in the IoTs referencemodel. In their research the
authors prove that it could prevent various attacks via theoretical security ana lyses. They
describes the system's overall architecture and presence and introduces the designed SGA
approach for E2E M2M service. They then evaluates the proposed schemes in terms of security
and provides theoretical analyses and describe why the proposed SGA is suitable to be a new
standard interface.
A centralized approach to efficiently distribute and manage a group key in generic ad hoc
networks and Internet of Things is presented in [22]. According to the authors, this is needed to
reduce the computational overhead and network traffic due to group membership changes caused
by users' joins and leaves . Their proposed protocol takes advantage of two possible leave
strategies. First, at a pre -determined time selected when the u ser joins the group or, second, at an
unpredictable time, as in the case of membership revocation. The proposed protocol is applied to
two following relevant scenarios: secure data aggregation in IoT and Vehicle -to-Vehicle (V2V)
communications in Vehicular Ad hoc Networks (VANETs).
Due to the close connection with the physical world, it is an important requirement for IoT
technology to be self -secure in terms of a standard information security model components.
Autonomic security should be considered as a c ritical priority and careful provisions must be
taken in the design of dynamic techniques, architectures and self -sufficient frameworks for
future IoT. Over the years, many researchers have proposed threat mitigation approaches for IoT

16
and WSNs. [23] considers specific approaches requiring minimal human intervention and
discusses them in relation to self -security. The authors brings together a broad range of ideas
linked together by IoT, autonomy and security. More particularly, they look at threat mitigation
approaches in IoT using an autonomic taxonomy and finally sets down future directions.
2.5 Communication
CoAP was described earlyer as a simpler alternative to the HTTP for connecting
constrainedsmart objects to theWeb. The adop tion of the protocol depends on its relative
advantage, and the cost –benefit associated with the use of the protocol is a significant factor
affecting a protocol adoption decision. [24] aims at deepening the understanding of th e cost –
benefits of CoAP and identifies the application scenarios where its use is likely to be
economically justifiable. The authors analyze the costs of using CoAP and HTTP in (WoT)
applications, by identifying the components of the total cost of ownershi p (TCO) model for these
applications and by studying the factors affecting individual costs. The use of the model is then
demonstrated by means of comparing the TCO of CoAP and HTTP in an environment
monitoring application scenario. The results of the anal ysis suggest that the simpler hardware
requirements of CoAP smart objects, as well as the lower communication overhead of the
protocol and the resulting reduced power consumption lead to cost advantages in the application
scenarios where: the smart objects are large in volume and are deployed in the field, engage in
frequent communications with the Web that are charged for on the basis of the volume of the
data being transferred, and are sleeping between the communication sessions.
Process and industry auto mation have strong demands for reliable communication and
security guarantees, which an IoT architecture has to incorporate from the start. Today,
deployment and commissioning of complex production processes or Internet -enabled
applications interacting wit h production systems still require a time consuming and errorprone
manual network configuration process. This is due to the need to maintain a high level of
determinism, safety, and security of the production process itself and avoiding both safetycritical
failures and costly production interruptions. The IoT@Work European funded research project
coordinated by Siemens is focusing on the industry automation and its networking and
communication needs. This project will deliver tools and runtime mechanisms ba sed on IoT
technologies to significantly simplify commissioning, operating, and maintaining complex
production processes. The contribution of this project will be focused on using self -configuration
mechanisms, enabling what is called secure Plug&Work IoT. [25] introduces the problem
domain and its challenges and present the approach followed in the project.

17
A novel cross -layer module for the IoT is proposed in [26] to accurately capture both the
high heterogeneity of the IoT and the impact of the Internet as part of the network architecture.
The use of cross -layer communication schemes to provide adaptive solutions for the IoT is
motivated by the high heterogeneity i n the hardware capabilities and the communication
requirements among things. The fundamental part of the module is a mathematical framework,
which is developed to obtain the optimal routing paths and the communication parameters
among things, by exploiting the interrelations among different layer functionalities in the IoT.
Moreover, a cross -layer communication protocol is presented to implement this optimization
framework in practical scenarios. The results show that the proposed solution can achieve a
global communication optimum and outperforms existing layered solutions. The novel cross –
layer module is a primary step towards providing efficient and reliable end -to-end
communication in the IoT.
Current M2M communication standardization efforts try to form alize the interfaces
between M2M nodes based on the perspective of exchanging uniform small data size with low
sampling rate only. However, many devices will require support for more heterogeneous traffic
patterns, with different network capacity. [27] introduces a communication concept for
supporting gracefully a heterogeneous set of devices. This paper analyses the effect of traffic size
in M2M transactions and propose a concept to adapt gracefully to support heterogeneous tra ffic
patterns in M2M systems. To prove its feasibility, the concept is exemplified on top of oneM2M
(a set of specifications that enable users to build platforms, regardless of existing sector or
industry solutions, according to [28]) architecture and implemented as part of the Fraunhofer
FOKUS OpenMTC toolkit. Additionally, the concept was applied to a deployment in an E –
Health pilot and practical measurements during functional evaluation are reported.
DNS architecture for the IoT is described in [29]. Similarly to the existing DNS
infrastructure on the Internet, the DNS for the IoT translates unique identifiers (URIs) of
physical objects to concrete network addresses, from which information about su ch objects, such
as status or location, can be extracted. The authors propose an experimental DNS infrastructure
for IoT, in the domain of transport logistics. They have simulated the behaviour of the DNS
infrastructure using a 3 tier hierarchy of domain n ame servers and a three level caching strategy.
Preliminary results indicate that a DNS approach could be a feasible proposition to realise object
tracking in IoT.
The research and the implementation of home automation are getting more popular
because the IoT holds promise for making homes smarter through wireless technologies. There is
a main requirement that make a wireless protocol ideal for use in the IoT, that is the energy
efficiency. Bluetooth Low Energy (BLE) has a high potential in becoming an impo rtant

18
technology for the IoT in low power, low cost, small devices. However, specific techniques can
be used in such a way as to further reduce the energy consumption of BLE. To this end, [30]
proposes a fuzzy logic based mecha nism that determine the sleeping time of field devices in a
home automation environment based on BLE. The proposed FLC determines the sleeping time
of field devices according to the battery level and to the ratio of Throughput to Workload
(Th/Wl). Simulati on results reveal that using the proposed approach the device lifetime is
increased by 30% with respect to the use of fixed sleeping time.
2.6 Studied solutions
In [31] an IPv6 and Zigbee home network frame is described briefly and the real products
are introduced capable of home networking like home gateway and information appliance.
ZigBee is an emerging network technology as a wireless communication standard that is
capable to satisfy such requirements with advantage of low cost, low power and wider coverage,
and therefore it is very suitable for the development and utilization of home. IPv6 offers new
function such as address that is extended from 32 bits to 128bits, auto -configuration and more
powerful security.
The structure of the smart home network requires three different devices: coordinator,
router and end device. Home environment can then be monitored, and the required data can be
delivered to the coordinator or home gateway. After program execution, proper control
command s will be sent from the gateway through coordinator to home appliances. A supervisory
control Intranet system which is low cost and high performance can be implemented by the
ZigBee technology. A terminal node first sends the data to coordinator node and t hen this
coordinator node retransmits the data to the aiming terminal node in return. Since all appliances
have their own IP addresses based on the IPv6, they can be directly connected to external
network. All appliances can be controlled locally and centr ally through the hand remote control
device at home and also remotely by the long -distance PC, which entered the home gateway
machine through the Internet .
An IOT infrastructure for collaborative warehouse order fulfillment based on RFID,
ambient intellige nce and multi -agent system is described in [32]. It consists of a physical devices
layer, a middleware ambient platform, a multi -agent system and an enterprise resource planning.
It integrates a bottom -up approach with decision support mechanisms such as self -organization
and negotiation protocols between agents based on the "com -peration" concept, which comes
from competition and cooperation. This approach was selected to improve reaction capabilities
of decentralized managemen t of warehouses in a dynamic environment. A collaborative

19
warehouse example was conducted to demonstrate the implementation of the proposed
infrastructure.
Since the IoT concept was introduced, the ideea of using IOT to help safety supervising
authorities of coal mine in order to strengthen the supervision on enterprises implementing
principal responsibility for safety was put forward, and specific application of IOT in safety
supervision of coal mine wer e analyzed thoroughly. A program of networked remote inspecting
technology is presented in [33] which provides a new way for innovating supervising.
The primary goal of [34] is to identify those par ameters that significantly affect students'
focus during the lectures. The authors describe and analyze the impact of several parameters of
the physical environment in a classroom on students’ focus, where the term "focus" refers to the
students' subjectiv e feeling of their ability to concentrate on a lecture at a given moment. The
measured several parameters in a real classroom environment using different low -cost smart
devices. The research is based on the dataset collected from 14 recorded lectures atten ded by 197
students. Five parameters were measured of the physical environment and extracted 22 features
from the lecturer’s voice. After analyzing collected measurements, they identified eight
parameters that have shown to have statistically different val ues for "focused" and "not focused"
segments. After performing additional series of trials we identified three parameters that could
be removed from the original dataset without changing classifier's accuracy, which left us five
uncorrelated parameters tha t have shown to have significant impact on students' focus.
The results of a quasi -experimental design with 500 households in a target group (with
smart meters) and 500 – in a control group (without smart meters) in Latvia is analyzed in [35].
Lorenz curves were used to quantify the reduction in electricity consumption and to study the re –
distribution of energy allocation due to roll -out of smart meters. Results on energy distribution
for the target group suggest that the reduct ion of electricity consumption play a minor role on
energy equity in a short -term, but effect is growing over last months of the study. These results
could be explained by the hypothesis that investments in energy efficient lighting and home
appliances tak e time.
The recent deployment of wireless sensor networks in Smart City infrastructures has led to
very large amounts of data being generated each day across a variety of domains, with
applications including environmental monitoring, healthcare monitoring and transport
monitoring. To take advantage of the increasing amounts of data there is a need for new methods
and techniques for effective data management and analysis to generate information that can
assist in managing the utilization of resources intelli gently and dynamically. Through [36], a
Multi -Level Smart City architecture is proposed based on semantic web technologies and

20
Dempster -Shafer uncertainty theory. The authors' proposed architecture is described and
explained in terms of its functionality and some real -time context -aware scenarios.
Several systems have been proposed in the last few years for guiding cognitively -impaired
people in performing daily activities, mostly based on binary sensors, cameras and other senso rs
such as RFID tags. Cameras are very intrusive, binary sensors (such as movement detectors) give
only basic information, and other types of sensors (such as RFID) need complex deployment.
[37] presents a new assistive expert system based on electric device identification to address the
problem of guidance and supervision in the performance of activities for people with cognitive
disorders living in a smart home. The system was solely based on a single power analyzer placed
in the electric panel. The authors proposed an algorithmic approach used to recognize erratic
behaviors related to cognitive deficits and provides cues to guide the person in the completion of
an ongoing task. This was achieved through load signatures study o f appliances represented by
three features, active power, reactive power and line -to-neutral, which allowed the determination
of the errors committed by the resident. This system was implemented within a genuine smart –
home prototype equipped with household appliances used by the patient during his morning
routines. Different multimedia prompting devices were used. The system was tested with real –
case scenarios modeled from former clinical trials, allowing demonstration of accuracy and
effectiveness in assis ting a cognitively -impaired resident in the completion of daily activities.
2.7 Proposed solutions
The paper presented at the ECAI 2016 conference in Ploiesti, Romania describes the
concept, architecture, building process and functionality of a device belonging to the smart
objects, smart monitoring devices and home automation systems familly: a smart r efrigerator that
is exposed as an IoT object and interacts with the items stored within, gathers information about
them, process this information into relevant data that is later conveyed through an IoT platform
to its owners.
The proposed system revolves around the core concept of product identification based on
RFID technology. The concept itself is built on a set of well -defined functionalities. Below we
give an overview of the most relevant ones:
 identifying new products: Similar to the intelligent refr igerator system for storing
pharmaceutical product containers described above, the proposed system will scan
all products at a fixed time interval and identify whether or not a certain product
was already marked as found in a previous scan. If not, then th e product is
considered new and a web request is made to a global database from which

21
information like product name, manufacturer and expiration date is fetched and
stored locally.
 identifying removed products: Using the same fixed interval scanning mechan ism
described above, the system identifies products that are missing from a scan,
although they were present inside the refrigerator in previous scans. If this scenario
is repeated consecutively three times then the system assumes that the product has
been permanently removed from the refrigerator and, based on the functionality
stated below, it will add it to a shopping list described in the following below.
 inventory and shopping list: Not everyone buys the same products over and over
again but there are those few products that are essential in every household’s
refrigerator (e.g. milk, eggs, cheese, etc.). For this purpose, an Essential Inventory
was created in which the refrigerator owners can add new products that they wish
to have available at all time s or save an already available product in the Essential
Inventory. A product that is marked for the Essential Inventory and has been
permanently removed from the refrigerator, as described above, will be added to
the shopping list. The Shopping List is a w eb application that the refrigerator owner
can access on any internet connected device (smartphone or tablet). The list of
products is a live feed because it does not interact directly with the refrigerator and
the data is stored off site, in a remote data base.
The monitoring system consists of 5 main components, 4 of which are hardware
components and the fifth is a software component: the RFID kit which can be broken down into
the RFID reader and the RFID antenna, the Arduino Breakout Kit, the Intel Edison and finally
the remote IoT platform used to aggregate the data received from the local system.
In order to have the widest coverage inside the refrigerator the antenna was to be placed
and the top or bottom of the refrigerator. Using a vertical implementa tion will result in not
requiring multiple antennas, one for each level compartment, thus reducing the costs and the
entire system complexity.
The ideas presented in this paper can be further developed in several directions. First of all,
further research is needed to effectively overcome one of the RFID's biggest weakness which
represents the radio waves propagation through metal and liquids.
Secondly, the proposed solution did not touch any of the security and privacy concerns that
currently affect the Io T paradigm development.
Lastly, the proposed solution does not cover the products stored in a refrigerator that are
not container in a RFID -equipped packaging like fruits of vegetables.

22
Although the solution presented in this paper works well in the contex t of a smart
refrigerator, the concept can be adapted and implemented in all contexts that require managing
and tracking storage items or inventories in enclosed small and medium sized areas. The added
value of this paper is represented by more than the te chnical solution and context that were
chosen, the concept itself being new and bringing new opportunities and ideas to the IoT
industry. As a short example, a similar solution can be implemented inside a wardrobe in order
to assess what clothes are availa ble and what are the optimal matches that can be wore in a
specific day, based on what was removed from the wardrobe the day before, outside weather
forecast, desired color scheme and other situational parameters. Many more implementations can
be found eve n outside of a home. This is why the concept adopted in this paper is part of the
added value.

23
3 Conclusions
Technology development will obviously continue to grow at an increasing pace. Embedded
technologies and sensing will be further miniaturized and redu ced in cost, and we will see an
ever-increasing instrumentation of these embedded technologies in the environment around us in
the objects and goods we own and use daily. Where technologies have many times been used for
simpler use cases, a shift is alread y underway with more complex use cases involving heavy
machinery, autonomous operations, and advanced remote control of vehicles and robots. In
addition, M2M and IoT will create a shift in business models across a number of industries in
conjunction with t he move from product sales towards a service economy, where a consumer,
rather than buying a product, buys a service, where the product is connected for its entire
lifetime.
The M2M and IoT markets are expected to see significant growth in coming years but in
order to do so both M2M and IoT ecosystems will need to adopt further standars to act as a glue
between all the fragmented and expansive systems implemented to this day. A standardisation of
this type is vital to enable interoperability in a cost efect ive way.
The dynamic rapidly changing and technology -rich digital environment enables the
provision of added -value applications that exploit a multitude of devices contributing services
and information. As IoT techniques mature and become ubiquitous, empha sis is put upon
approaches that allow things to become smarter, more reliable and more autonomous.
Realizing the vision of sustainable IoT applications requires the enhancement of IoT
technologies with new ways that will enable things and objects to become more reliable, more
resilient, more autonomous and smarter.

24
4 Annexes
4.1 Annex 1: Smart R efrigerator poster

25
5 References
[1] P. Fremantle, "A Reference Architecture for the Internet of Things," WSO2, 2015.
[2] D. Kyriazis and T. Varvarigou, "Smart, Autonomous and Reliable Internet of Things,"
Procedia Computer Science, vol. 21, p. 442 – 448, 2013.
[3] M. Díaz, C. Martín and B. Rubio, "State -of-the-art, challenges, and open issues in the
integration of Interne t of things and cloud computing," Journal of Network and Computer
Applications, vol. 67, p. 99 –117, 2016.
[4] D. Miorandi, S. Sicari, F. D. Pellegrini and I. Chlamtac, "Internet of things: Vision,
applications and research challenges," Ad Hoc Networks, vol. 10, no. 7, p. 1497 –1516,
2012.
[5] I. Mashal, O. Alsaryrah, T. -Y. Chung, C. -Z. Yanga, W. -H. Kuo and D. P. Agrawal,
"Choices for interaction with things on Internet and underlying issues," Ad Hoc Networks,
vol. 28, p. 68 –90, 2015.
[6] S. Kim and S. Kim, "A multi -criteria approach toward discovering killer IoT application in
Korea," Technological Forecasting and Social Change, vol. 102, p. 143 –155, 2016.
[7] J. Holler, V. Tsiatsis, C. Mulligan, S. Avesand, S. Karnouskos and D. Boyle, From
Machin e-to-Machine to the Internet of Things, Academic Press, 2014.
[8] I. Silva, R. Leandro, D. Macedo and L. A. Guedes, "A dependability evaluation tool for the
Internet of Things," Computers & Electrical Engineering, vol. 39, no. 7, p. 2005 –2018,
2013.
[9] S. A. Amin, A. Ali -Eldin and H. A. Ali, "A context -aware dispatcher for the Internet of
Things: The case of electric power distribution systems," Computers & Electrical
Engineering, vol. 52, p. 183 –198, 2016.
[10] G. Fersi, "A Distributed and Flexible Architecture for Internet of Things," Procedia
Computer Science, vol. 73, pp. 130 -137, 2015.
[11] S. Efremov, N. Pilipenko and L. Voskov, "An Integrated Approach to Common Problems in
the Internet of Things," Proced ia Engineering, vol. 100, pp. 1215 -1223, 2015.
[12] K. L. Lueth and J. Kotzorek, "IoT Platform, The central backbone for the Internet of
Things," IoT Analytics, 2015.
[13] M. Ebrahimi, E. ShafieiBavani, R. K. Wong, S. Fong and J. Fiaidhi, "An adaptive meta –
heuristic search for the internet of things," Future Generation Computer Systems, 2015.
[14] S.-L. Chua and L. K. Foo, "Sensor Selection in Smart Homes," Procedia Computer Science,
vol. 69, pp. 116 -124, 2015.
[15] A. M. Souza and J. R. Amazonas, "An Outlier Detect Algorithm using Big Data Processing
and Internet of Things Architecture," Procedia Computer Science, vol. 52, pp. 1010 -1015,
2015.
[16] C. Li, L. Huang, L. Chen and W. Luo, "Deadlock detection and recov ery for component –
based systems," Mathematical and Computer Modelling, vol. 58, no. 5 -6, p. 1362 –1378,
2013.
[17] M. Henze, L. Hermerschmidt, D. Kerpen, R. Häußling, B. Rumpe and K. Wehrle, "A
comprehensive approach to privacy in the cloud -based Interne t of Things," Future
Generation Computer Systems, vol. 56, p. 701 –718, 2016.
[18] A. Jacobsson, M. Boldt and B. Carlsson, "A risk analysis of a smart home automation
system," Future Generation Computer Systems, vol. 56, p. 719 –733, 2016.
[19] X. Yao, Z. Chen and Y. Tian, "A lightweight attribute -based encryption scheme for the

26
Internet of Things," Future Generation Computer Systems, vol. 49, p. 104 –112, 2015.
[20] S. Gusmeroli, S. Piccione and D. Rotondi, "A capability -based security approach to ma nage
access control in the Internet of Things," Mathematical and Computer Modelling, vol. 58,
no. 5 -6, p. 1189 –1205, 2013.
[21] H.-C. Chen, I. You, C. -E. Weng, C. -H. Cheng and Y. -F. Huang, "A security gateway
application for End -to-End M2M communication s," Computer Standards & Interfaces, vol.
44, p. 85 –93, 2016.
[22] L. Veltri, S. Cirani, S. Busanelli and G. Ferrari, "A novel batch -based group key
management protocol applied to the Internet of Things," Ad Hoc Networks, vol. 11, no. 8, p.
2724 –2737, 2013.
[23] Q. M. Ashraf and M. H. Habaebi, "Autonomic schemes for threat mitigation in Internet of
Things," Journal of Network and Computer Applications, vol. 49, p. 112 –127, 2015.
[24] T. Levä, O. Mazhelis and H. Suomi, "Comparing the cost -efficiency of CoAP and HTTP in
Web of Things applications," Decision Support Systems, vol. 63, p. 23 –38, 2014.
[25] A. M. Houyou and H. -P. Huth, "Internet of things at work European Project: Enablin g Plug
and Work in Automation Networks," Embedded World Conference 2011, 2011.
[26] C. Han, J. M. Jornet, E. Fadel and I. F. Akyildiz, "A cross -layer communication module for
the Internet of Things," Computer Networks, vol. 57, no. 3, p. 622 –633, 2013.
[27] A. Elmangoush, A. A. Corici, R. Steinke, M. Corici and T. Magedanz, "A Framework for
Handling Heterogeneous M2M Traffic," Procedia Computer Science, vol. 63, pp. 112 -119,
2015.
[28] "oneM2M: The Interoperabiliry Enabler For The Entire M2M Ecosyst em," oneM2M
Parthers, 2015.
[29] B. Karakostas, "A DNS Architecture for the Internet of Things: A Case Study in Transport
Logistics," Procedia Computer Science, vol. 19, pp. 594 -601, 2013.
[30] M. Collotta and G. Pau, "Bluetooth for Internet of Things: A fuzzy approach to improve
power management in smart homes," Computers & Electrical Engineering, vol. 44, p. 137 –
152, 2015.
[31] Z. Zou, K. -J. Li, R. Li and S. Wu, "Smart Home System Based on I PV6 and ZIGBEE
Technology," Procedia Engineering, vol. 15, pp. 1529 -1533, 2011.
[32] P. J. Reaidy, A. Gunasekaran and A. Spalanzani, "Bottom -up approach based on Internet of
Things for order fulfillment in a collaborative warehousing environment," Inter national
Journal of Production Economics, vol. 159, p. 29 –40, 2015.
[33] Z. Yinghua, F. Guanghua, Z. Zhigang, H. Zhian, L. Hongchen and Y. Jixing, "Discussion
on Application of IOT Technology in Coal Mine Safety Supervision," Procedia
Engineering, vol. 43, pp. 233 -237, 2012.
[34] A. Uzelac, N. Gligoric and S. Krco, "A comprehensive study of parameters in physical
environment that impact students’ focus during lecture using Internet of Things,"
Computers in Human Behavior, vol. 53, p. 427 –434, 2015.
[35] U. Bariss, L. Timma and D. Blumberga, "Smart Metering Pilot Project Results," Energy
Procedia, vol. 61, pp. 2176 -2179, 2014.
[36] A. Gaur, B. Scotney, G. Parr and S. McClean, "Smart City Architecture and its Applications
Based on IoT," Procedia Computer Science, vol. 52, pp. 1089 -1094, 2015.
[37] C. Belley, S. Gaboury, B. Bouchard and A. Bouzouane, "Nonintrusive system for assistance
and guidance in smart homes based on electrical devices identification," Expert Systems
with Applications, vol. 42, no. 19, p. 6552 –6577, 2015.