444 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO.

5, OCTOBER 2019

On the Security Aspects of Internet of Things: A

Systematic Literature Review
Evandro L. C. Macedo, Egberto A. R. de Oliveira, Fabio H. Silva, Rui R. Mello Jr, Felipe M. G. França,
Flavia C. Delicato, José F. de Rezende, and Luís F. M. de Moraes

Abstract: Internet of Things (IoT) has gained increasing visibility abling the integration of heterogeneous technologies. As a con-
among emerging technologies and undoubtedly changing our daily sequence, IoT leads to the development of novel applications,
life. Its adoption is strengthened by the growth of connected de- as well as the improvement of existing ones, enabling unprece-
vices (things) as shown in recent statistics. However, as the number dented data collection.
of connected things grows, responsibility related to security aspects With 29 billion of connected things estimated by 2022 [2], the
also needs to increase. For instance, cyberattacks might happen
spread of IoT paves the way to a myriad of applications that can
if simple authentication mechanisms are not implemented on IoT
applications, or if access control mechanisms are weakly defined.
improve life quality of our society. Considerable benefits can
Considering the relevance of the subject, we performed a system- be seized, for instance, in healthcare, smart cities, smart home
atic literature review (SLR) to identify and synthesize security is- applications and intelligent transport systems (ITS). All these
sues in IoT discussed in scientific papers published within a period applications avail improvements due to the increased amount of
of 8 years. Our literature review focused on four main security as- data that can be obtained from things, which helps decision mak-
pects, namely authentication, access control, data protection, and ing processes. However, all these benefits come together with
trust. We believe that a study considering these topics has the po- the responsibility of providing them in a secure way. If any of
tential to reveal important opportunities and trends related to IoT these systems are compromised, not only financial issues could
security. In particular, we aim to identify open issues and tech- happen. For example, in a healthcare IoT application, a tam-
nological trends that might guide future studies in this field, thus pered sensor may expose private data or deliver wrong measures
providing useful material both to researchers and to managers and of heart rate leading to prescription errors; or in an ITS if traf-
developers of IoT systems. In this paper, we describe the protocol
fic lights control are hacked, this may cause the occurrence of a
adopted to perform the SLR and present the state-of-the-art on the
car crash. Thus, in this context, besides financial loss, a security
field by describing the main techniques reported in the retrieved
studies. To the best of our knowledge, ours is the first study to flaw can lead to violation of data privacy and, in the worst cases,
compile information on a comprehensive set of security aspects in it may even incur physical damage to human beings. On the one
IoT. Moreover, we discuss the placement, in terms of architectural hand, having connectivity at anytime for anything and anywhere
tiers, for deploying security techniques, in an attempt to provide is tempting and brings many advantages, but on the other hand,
guidelines to help design decisions of security solution developers. new requirements and challenges arise that need to be consid-
We summarize our results showing security trends and research ered in the design of IoT systems and applications. According
gaps that can be explored in future studies. to the authors in [3]–[6], security issues, such as privacy, autho-
rization, verification, access control, information storage, and
Index Terms: Access control, architecture, authentication, data management, are major challenges in an IoT environment. As
protection, internet of things, IoT, security, techniques, trust. we will discuss in Section I.A, many of the characteristics inher-
ent to the IoT environment, such as the large scale, dynamism,
and heterogeneity of devices, contribute to making such security
I. INTRODUCTION aspects challenging. Addressing these challenges and providing

I NTERNET of Things (IoT) [1] is a disruptive paradigm that suitable security solutions is essential for the development and
brings the next wave of evolution of the Internet by extending widespread dissemination of the IoT paradigm.
communication among any kind of smart object (thing), and en-
A. Security Challenges in IoT
Manuscript received March 4, 2019. IoT, as an ecosystem composed of the merging of heteroge-
This study was partially funded by Conselho Nacional de Desenvolvimento
Científico e Tecnológico (CNPq), under grant 307378/2014-4 and by Fundação neous network technologies, not only inherits the same security
de Amparo à Pesquisa do Estado de São Paulo (FAPESP) under the grant problems from traditional wired networks, wireless networks,
2015/24144-7 for Flavia C. Delicato, by CNPq under grant 312357/2017-6 for mobile networks, and sensor networks, but also incorporates
Jose F. de Rezende and under 309781/2016-7 for Felipe M. G. França, by Fun-
dação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) and Rede- new issues derived from its very specific characteristics.
Rio (the state academic backbone network) under the grant 150.134/2010 for Let’s begin our discussion by the sensor devices in charge of
Luís F. M de Moraes, and in part by the Coordenação de Aperfeiçoamento de
Pessoal de Nível Superior – Brasil (CAPES) – Finance Code 001.
the acquisition and monitoring of environmental variables, and
The authors are with the Computer Engineering and Systems Program, Fed- major responsible for the integration between the physical and
eral University of Rio de Janeiro, email: {evandro, moraes}@ravel.ufrj.br, virtual worlds. Sensors nodes have limited computing and stor-
{egberto, fabhenr, ruirodrigues, felipe}@cos.ufrj.br, [email protected],
[email protected].
age capacities, which prevent the implementation of robust and
Evandro L. C. Macedo is the corresponding author. complex security mechanisms, for example for authentication
Digital Object Identifier: 10.1109/JCN.2019.000048 purposes. It is necessary to implement lightweight technologies
1229-2370/19/$10.00 
c 2019 KICS

Creative Commons Attribution-NonCommercial (CC BY-NC).

This is an Open Access article distributed under the terms of Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0)
which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided that the original work is properly cited.
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 445

in the devices, which generally do not provide the highest de- issues in the IoT field. The authors in [3] provide an overview
grees of protection. Another characteristic related to IoT de- of IoT focusing on enabling technologies, protocols, and appli-
vices is their heterogeneity, pervasiveness, and potential mo- cation issues, and highlight security as one of the main chal-
bility. Since literally any physical object can be instrumented lenges that must be addressed in IoT. Alaba et al. [9] focus
and become part of the IoT, there is a large increase in the at- on the state-of-the-art of IoT security threats and vulnerabili-
tack surface. Appliances, home utilities, day-to-day objects that ties, offering a taxonomy of threats and discussing possible cy-
were not designed with security issues in mind, by being net- berattacks. Authors in [11] review eight IoT frameworks com-
worked are now posing vulnerabilities to be exploited by mali- mercially available, pointing out the security features of the re-
cious users. Besides, for identification and access control pur- spective architectures. Khan et al. [12] explore open challenges
poses, it is quite difficult to manage identities considering such on IoT security, presenting a taxonomy of security issues, fo-
a myriad of (potentially mobile) devices. Regarding the trust as- cusing on blockchain-based solutions. Sfar et al. [13] present a
pect, it is challenging to rely on devices that can be easily tam- roadmap of security in the IoT through a systemic and cognitive
pered since they might be mobile. approach. By cognitive, the authors mean that their approach
At the network layer, IoT ecosystems suffer from the same provides complex and dynamical interactions between process,
vulnerabilities of wireless networks in general, aggravated by its people, technology and organization, in order to give the flex-
high dynamism, the need to integrate different technologies and ibility for the system to be able to analyze different situations
the lack of standards currently in use. There is as yet no domi- and perform the most suitable measures to guarantee reliabil-
nant technology nor a stack of standardized protocols for use in ity and security. They also show related surveys that cover dif-
IoT. This has often led to the adoption of proprietary protocols ferent security aspects (authentication, access control, privacy,
and the creation of ad hoc network architectures, not always fo- confidentiality, trust, data protection, availability etc), each one
cused on providing security solutions at all layers, which can with a different focus. Irshad [14] presents a study with focus
result in vulnerabilities to be exploited in cyberattacks. on information security management frameworks for IoT, and
The heterogeneity of IoT applications and their nature im- in [5] the authors discuss about security problems in IoT consid-
pose additional security challenges. Many applications require ering their characteristics concerning the application layer (e.g.,
the acquisition of sensitive data, such as personal data (includ- data protection, software vulnerabilities), network layer (e.g.,
ing body variables, activities and location) or industrial process DoS attacks, network congestion, authentication), and percep-
monitoring data. The need to protect the data trafficked in IoT is tion layer (e.g., replay attack, fake node). They also point out
crucial. Besides that, different IoT application fields have differ- security measures to deal with these problems. Finally, in [4] the
ent industry standards and related security specifications. This authors provide a survey in which secure communication proto-
makes it difficult to adopt unique and integrated security frame- cols based on IP are explored in the IoT context. In particular,
works, leading to the need for customized solutions. the authors discuss the applicability and limitations of existing
Finally, considering the intrinsic features of this emerging IP-based Internet security protocols, as well as other types of se-
type of ecosystem constituted by IoT, it is common the occur- curity protocols used in wireless sensor networks. Such analysis
rence of opportunistic, ad hoc interactions among devices and is based on a taxonomy focusing on key establishment mecha-
users, leveraged by some specific contexts. For instance, a mo- nisms in the context of the IoT.
bile device can make its resources available only for users that However, these studies focus on specific approaches to cover
are in its neighborhood for a given period of time. Therefore, security issues in IoT and do not provide a holistic view of es-
there is a trend of adopting more relaxed business models, in- sential aspects to provide security at different levels for these
stead of formal contracts between parties. In this context, trust systems. Our study discusses the following security aspects: (i)
in the participating parties is a key issue and the ad hoc na- Authentication; (ii) access control; (iii) data protection; and (iv)
ture of interactions makes the establishment of trust a challeng- trust. We consider them as very comprehensive and encompass-
ing aspect. In addition, the need arises to deal with the context ing further aspects. For instance, data protection includes pri-
(temporal, geo-spatial, among others) of the interactions that oc- vacy, trust includes availability, etc. Moreover, existing studies
cur in the system. Security solutions should therefore ideally do not discuss architectural aspects in terms of the places of im-
be context-aware and incorporate context information into deci- plementation of the security techniques. With the analysis and
sions, for example, regarding authentication and access control. synthesis of the studies retrieved in our literature search, we aim
to shed light on techniques that are used as security solutions
B. Context and Goal of this Study and discuss where the reported techniques were implemented
Given the relevance of the subject, many references in the cur- in terms of the architectural stack of IoT systems. We believe
rent literature [3]–[5], [7]–[13] present studies that highlight the that the type of overview and discussions provided in our pa-
importance of considering security aspects during the develop- per may help researchers to envision opportunities and reflect
ment of IoT solutions, from different perspectives. In [7], the on the state-of-the-art to identify important challenges and re-
authors analyze the most relevant available solutions for secu- search gaps regarding security in IoT. In addition, the discussion
rity aspects such as access control, privacy, trust, confidentiality, about architectural aspects may allow to scale security solutions
authentication, among others. For each considered aspect, they according to restricted computational resources of IoT devices,
focus on proposals that address security solutions and security thus helping developers of IoT systems. Among prior studies,
middleware applied to mobile devices. They also present ongo- different research methodologies were used to conduct the re-
ing projects from the European Commission that address such view of the literature. One of the methodologies commonly used
446 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO. 5, OCTOBER 2019

due to its capability of providing a set of well-defined, repro-

ducible, and auditable steps, is the systematic literature review
(SLR) [15], which comprises a research protocol that can be
used in other studies. The motivation to use SLRs as an inves-
tigation method should start from real problems detected in the
industrial practice, as it is the case of IoT security. The results
of SLRs should, in turn, affect and improve this practice.
In our study we adopt the SLR methodology. An SLR is de-
fined as a secondary study that applies a well-defined approach
to determine, evaluate and interpret scientific evidence related
to a specific research question, in a way that is unbiased and
repeatable. In short, in this paper we present a comprehensive
study with the following differentials and main contributions:
• We adopt a well-defined research protocol based on the SLR
methodology that can be replicated and verifiable;
• We present a comprehensive review on the state of the art re-
garding four major security aspects;
• We discuss the location, in terms of architectural tiers, of
security techniques implementations, thus providing hints that
can help developers to visualize which tier(s) would be the best
place to implement a specific security technique.
We believe that a study considering these topics is relevant
and has the potential to reveal opportunities and trends related
to IoT security. In particular, we aim to identify open issues and
technological trends that might guide future studies in this area,
thus providing useful material both to researchers as well as to Fig. 1. Phases and stages of an SLR.
managers and developers of IoT systems.
The rest of this article is organized as follows. Section II de-
scribes the steps and protocol adopted for the SLR execution. In In Fig. 1 we present all phases and respective stages that com-
Section III we present some statistics for selected studies. The pose the SLR process. The Planning phase corresponds to the
analysis of each security aspects is presented in Section IV, fol- set of tasks related to designing and preparing the protocol. The
lowed by the discussion of the results, in the Section V. Finally, Conducting phase refers to the tasks that will apply the protocol
we conclude the paper in Section VI. until the primary studies are selected. Finally, at the Document-
ing phase the SLR will be produced in a paper or report format.
II. RESEARCH METHODOLOGY
In this section we present the methodology used to conduct A. Planning
the literature search and the selection of the studies to be in-
In this section we will discourse the tasks performed at the
cluded in our analysis. An SLR is based on the choice and sub-
Planning phase of our SLR. This way, all the activities executed
sequent analysis of a series of scientific articles, according to a
to prepare our protocol will be described below.
well-defined protocol, so that the results can be reproduced, ver-
ified, and audited. An SLR is basically composed of three phases
[16], namely: (i) Planning; (ii) Conducting the review; and (iii) A.1 Requirements Identification and Research Questions
reporting (Documenting) the review. Activities of defining the
research questions, developing a review protocol and validating The first step to perform an SLR is the recognition of the
the review protocol are carried out during the Planning phase. need of performing the study, which corresponds to Stage 0 of
The Conducting phase comprises identifying relevant study, se- an SLR, according to Fig. 1. In our case, this task was accom-
lecting primary studies, assessing study quality, extracting re- plished by identifying the needs to uncover gaps and trends re-
quired data, and synthesizing data. In the last phase, Document- lated to the four IoT security aspects addressed in this study.
ing, researchers write a review report and validate it. According Therefore, it is necessary to identify some research questions
to [17] such approach has the benefits of providing an overview (RQ) to be answered from the inputs provided by the analysis of
of a specific domain, identifying the state-of-the-art on the re- relevant studies, which will constitute the primary studies. Spec-
searched field, which is interesting for researchers who need a ifying this RQs corresponds to the Stage 1 of the SLR protocol.
first approach on a particular subject. In addition, this kind of For the specific case of our SLR, based on recent researches
study enables the identification of research trends, raising the [18] in the field of authentication, access control, data protec-
most discussed aspects and open issues, indicating possibilities tion, and trust applied to the IoT, the following RQs, and their
of research in less discussed aspects. respective goals have been defined, as listed in Table 1.
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 447

Table 1. Research question.

Research questions Goals
What are the main security techniques that consider data To obtain an overview of access control, authentication,
RQ1
protection, access control, authentication, and trust in IoT? data protection and trust techniques already proposed.

What are the trends and gaps concerning authentication, To identify trends and gaps among considered
RQ2
access control, data protection, and trust? security aspects.

A.2 Source Bases and Search Strings studies to be included in the literature review only if they present
a scientific contribution to the body of knowledge on security
According to Fig. 1, the Stage 2 of an SLR corresponds to
aspect in IoT context. For this present paper, the following cri-
the definition of: Source bases, search strings, inclusion and ex-
teria were identified for inclusion and exclusion of studies, as
clusion criteria, and quality criteria. In this section, we will see
presented in Tables 2 and 3, respectively.
details about source bases and search strings definition.
Regarding the exclusion criteria, a filter was created, called
As a way to identify the primary studies necessary to our
“Relevance Criterion”, which aims to verify the relevance of a
study, four scientific databases were used, namely IEEExplore,
selected study, from the point of view of the number of existing
Scopus, Science Direct, and Web of Science. These electronic
citations. The motivation behind the definition of this criterion
databases consist of some of the most relevant digital libraries
was that we consider that studies without a minimum number of
for scientific studies in the area of Computer Science.
citations have little impact and have no relevant scientific poten-
It is important to explain that the ACM Digital Library
tial. This filter consists of papers having at least one citation per
(https://dl.acm.org), another well-known source of relevant sci-
year (considering the period from 2010 to 2016). For example,
entific information in the field, was also initially considered.
an article published in 2015 should have at least 4 citations. For
Nevertheless, the absence of the Abstract field on the search
articles published in 2017 or 2018 this filter was not applied,
results export template became an issue to the adopted review
since they are considered quite recent.
protocol as it would preclude applying the inclusion and exclu-
As it can be seen in Table 3, some criteria aim to discard stud-
sion criteria detailed ahead. Regardless, since search results are
ies not related to the subject of our study, for instance, EC1,
frequently repeated between different platforms, we consider re-
EC2 and EC3. Our goal is to find primary studies that can
moving ACM Digital Library from the group of source bases
be useful to help answering the formulated research questions.
does not cause a material loss in the final outcome of our study.
Considering this, we propose the EC11 to remove surveys from
As previously mentioned, one of the characteristics of an SLR
the results of queries, as this type of publication is not consid-
is to allow the proposed process to be reproduced. For that,
ered a primary study. The last exclusion criteria, EC12, aims
search strings were used to select primary studies. This search
to discard studies that do not suggest any solution or new tech-
strings are specific to each library, as listed below:
nique that can be applied on any of the four aspects focused on
• IEEE Xplore, Science Direct, and Scopus: this paper. Finally, it is important to point out that many of these
(internet of things OR internet-of-things OR iot OR web of exclusion criteria are commonly used in SLR. In this sense our
things OR wot) AND (access control OR identity OR authen- criteria was inspired in [17] and [20].
tication OR data protection OR data security) AND (trust OR
trust-based OR architecture OR framework) A.4 Quality Criteria
• Web of Science: After applying the criteria for selection of studies described in
TI=(internet of things OR internet-of-things OR iot OR web Section II.A.3, an optional step is to filter the selected studies by
of things OR wot) AND TS=(access control OR identity OR some quality criteria. According to Wohlin et al. [21], there is
authentication OR data protection OR data security) AND no universally agreed-on and applicable definition of study qual-
TS=(trust OR trust-based OR architecture OR framework) ity, although the most practical means for quality assessments
It is important to say that due to the specificity of each digi- are checklists. For this purpose, the authors used Kitchenham et
tal library, the strings have some differences. However, the ade- al. [19] guidelines as inspiration to define our quality criteria:
quacy of each string to its respective library can be considered • QC1 - Is there a clear statement of the aims of the research?
an important step in the process of performing an SLR, due to • QC2 - Is the proposed architecture/algorithm/protocol feasi-
its ability to obtain suitable and relevant primary studies. ble (it has been or can be applied to a real scenario)?
• QC3 - Were the experiments fully/properly analyzed and ex-
A.3 Inclusion and Exclusion Criteria plained? Does the results support the ideas shown in the paper?
Once the initial choice of studies from the search in the According to [19], the importance of these criteria is to pro-
databases is done, the next steps are to screen the retrieved stud- vide more detailed inclusion/exclusion criteria than the general
ies and select the relevant ones that will be the focus of a more in ones, to weight the relevance of individual studies after syn-
depth analysis. According to the SLR methodology, described thesizing, to emphasize the differences between studies, and to
in [19], it is necessary to establish rules for selecting the relevant guide for further research. This way, it is possible to minimize
studies, specifically, we must define a set of inclusion/exclusion bias and maximize validity of the studies.
criteria. The rationale behind the adopted criteria is to select Still according to [19], the quality criteria of an SLR may
448 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO. 5, OCTOBER 2019

Table 2. Inclusion criteria.

Inclusion criteria
IC1 The study presents or discusses opportunities, challenges, or schemes related to authentication, access control, data
protection or trust in IoT
IC2 The study presents or discusses security architectures related to authentication, access control, data protection or
trust in IoT

Table 3. Exclusion criteria.

Exclusion criteria
EC1 The study is not related to IoT
EC2 The study is not related to security in IoT
EC3 The study does not address authentication, access control, data protection or trust in IoT
EC4 The study is a previous version of a more complete study about the same research subject
EC5 The study does not have an abstract or the full text is not available
EC6 The study is not written in English, which is the most common language in scientific papers
EC7 The study was published before 2010
EC8 The study consists of a book chapter
EC9 The study was not approved by the "Relevance Criterion"
EC10 The study does not present references
EC11 The study is a survey
EC12 The study does not present a proposal for solution

need to consider a measurement scale for each item, instead of This corresponds to the Stage 3 of the SLR, according to Fig. 1.
a simple Yes/No answer, since it sometimes can be misleading. We extracted meta-data of these studies and inserted them into
Hence, for each quality criterion above, it is only possible to ad- a results table, to be used during the application of inclusion,
mit one of the following three answers: “yes”, “no”, and “par- exclusion, and quality criteria. The fields in the results table are
tially”. Each “yes” answer equals 1.0 point; 0.5 point for each described below:
“partially” answer and 0.0 for “no” answers. Finally, articles that • Database, Author, Title, Journal, Pages, Year, Keywords, Ci-
do not sum 2.0 or more points are excluded by the quality crite- tations, URL, Booktitle, Abstract, Publisher, and DOI.
ria shown in (1). Once the results table is created, the Stage 4 of the SLR can
be performed. In Fig. 2, the Stages 4 (paper selection) and 5
P QC1 + P QC2 + P QC3 >= 2.0 (1) (quality evaluation) of an SLR are shown, where we can see the
The PQCn corresponds to the score of a quality criteria of number of papers excluded in each step. Note that this figure
number n. The values stipulated for each quality criterion aim is divided into three different steps. The first step (from 1186
to assign equal weights for each one, so that the paper has to to 501 papers) was more “mechanical” than the others and was
reach at least 66% of this criteria, similar to [20]. It‘s important developed by all of the four researchers at the same time, in
to notice that papers excluded by these quality criteria are not a sequential way, without division of the work. In the second
considered poor quality papers, but rather it does not meet the step (from 501 to 223 papers), the papers were distributed to the
requirements defined for our specific research goals. researchers, who performed their work in parallel. In this step
So, with the definition of the source base and search the inclusion and exclusion criteria were applied on the title,
strings (Section II.A.2), inclusion and exclusion criteria (Sec- abstract and keywords (Screening). Lastly, in the step 3 of the
tion II.A.3) and quality criteria used (Section II.A.4), we fin- paper selection process (from 223 to 131), the 223 papers were
ished the Stage 2 and the Planning phase. distributed again to the researchers. However, in this step we
applied the inclusion, exclusion and quality criteria looking at
B. Conducting the full paper. More details about these steps presented at the
Fig. 2 will be described below.
In this section we present more details about the application • Step 1: The first task concerning the exclusion/filtering of
of the previously described protocol. The following stages will papers consists in removing from the results table the records
be detailed: Identification of research sources, paper selection, that presented “anomalies” returned by the search. For exam-
quality evaluation, data extraction and synthesis. ple, records that returned inconsistency on some fields, such as
publications missing author names, abstract or title. Thus, we
B.1 Steps of the Selection Process and Quality Evaluation
removed 25 search anomalies. Another important step in this
To accomplish our goals, four people (Ph.D. students, co- stage was removing duplicated papers from this initial table. We
authors of this manuscript) executed the defined protocol dur- found that 115 papers were duplicated. So, we have reached a
ing a time-frame of approximately five months. After defining total of 1046 pre-selected articles. After this initial removal of
research questions and elaborating the search strings for each papers, we applied the defined inclusion and exclusion criteria
source database, the searches were performed resulting in 1186 at the fields of the results table, in a “mechanical” way. This
scientific papers representing candidates for primary studies. means that we use some formulas, macros and other tools to fil-
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 449

Initial number of studies Distribution per year

1186 70
63
Search anomalies = 25
60
1161 Initial
filter

Step 1
Duplicates = 115 50

1046 40
EC1=181, EC5=2,
Exclusion
EC7=4, EC8=24, 30
EC9=334 criteria 30
501
EC1=4, EC2=142,

Step 2
Exclusion 20
EC3=91, EC5=3,
12 13
EC8=1, EC10=1, criteria
223 EC11=36 10 6 6
EC1=1, EC2=2,
1
EC3=17, EC5=6, Exclusion
0
EC8=4, EC11=8, criteria
Step 3
EC12=6 2012 2013 2014 2015 2016 2017 2018
178

QC = 48
Quality Fig. 3. Distribution of the primary studies per year.
criteria
131
Primary studies B.2 Data Extraction and Synthesis
After the selection of the 131 primary studies, the Stage 6 of
Fig. 2. Number of papers excluded per step. the SLR consists of data extraction and progress monitoring. So,
the data of these papers were included in a data extraction table
for the subsequent steps. The attributes for the data extraction
table are described below:
ter the fields of the result table searching for papers that would
• Id, Paper, Year, Journal, Main Challenges, App Domain,
be removed without a detailed analysis. For instance, the rel-
Authentication (Technique, Tier), Access Control (Technique,
evance criterion (defined at the Section II.A.3) was applied on
Tier), Data Protection (Technique, Tier), Trust (Technique,
the records of the result table through a formula in the work-
Tier), Notes, Status, Quality Criteria (QC1, QC2, QC3, Sum).
sheet. This way, after applying the inclusion and exclusion cri-
teria through this mechanical process, we reached a number of The last stage of the conducting phase consists of synthesiz-
501 pre-selected papers. It’s very important to mention that all ing the extracted data from the primary studies to answer the
tasks described at the step 1 of the Fig. 2 were applied by all of research questions, that will be detailed in Sections III and IV.
the four researchers together, without distribution of the work.
The intention was to reach a common understanding and felling C. Documenting
about the application of the protocol by each researcher and de- Finally, the last two stages to the SLR are the Stage 8 - reports
crease potential biases. and recommendations and Stage 9 - results presentation. Both
• Step 2: Resuming the selection process to identify the pri-
steps consist of preparing the reports and the results in a paper
mary studies, the 501 articles selected in the previous phase (or report) format. In the next sections, we present the obtained
were distributed between the four researches. Thus, the step two results, the analysis, the discussion and conclusions.
of our selection process consists of applying the exclusion cri-
teria in the abstract, title, and keywords (screening), reaching a
total of 223 selected articles. So, until the present step, it was III. STATISTICS FOR SELECTED STUDIES
not necessary reading the whole paper.
After selecting and analyzing the primary studies, in this sec-
• Step 3: The 223 selected articles in step two were divided
tion, quantitative descriptions of frequencies were used to eval-
among the researchers once again, totaling about 56 articles
uate and synthesize the primary studies in terms of their publi-
per researcher. In this step, the papers were fully analyzed by
cation year, application domain and the architectural tier of the
means of their complete reading. After this more detailed anal-
deployment security technique.
ysis, from the initial set of 223 articles, 45 were eliminated by
exclusion criteria. The studies were evaluated from the perspec-
A. Distribution over the Years
tive of the quality criteria (described in Section II.A.4) in which
48 articles were removed of our analysis. Thus, after perform- It is well-known that security and data privacy are major chal-
ing all the steps responsible for the elimination of articles, 131 lenges in the context of IoT. The distribution of primary studies
primary studies were identified. At this stage, we finished the over the years (Fig. 3) corroborates that these challenges have
Stage 4 (papers selection) and the Stage 5 (quality evaluation) of been the subject of study and with strong growth in recent years.
the SLR process (Fig. 1). The remaining articles were selected It is important to note that more than 70% of the primary studies
to be further analyzed and their data extracted for future eval- have been published in the last two years (2017 and 2018). If
uation and implementation of the last phase of the SLR (docu- we consider the fact that this extraction was carried out in May
menting). 2018, it is not risky to suggest that the year of 2018 might exceed
2017 in the number of related publications.
450 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO. 5, OCTOBER 2019

Application domain 100 All

Cloud
80 28 Fog + Cloud
Fog + Cloud

Count of articles
Healthcare 17% 60 5 Thing + Cloud
5
Smart cities 9 Thing + Fog
11
Industrial IoT 40 Thing
8
Education 29 9
11%
Image processing 20 9
14 6
9
Military 15 5
61% 8 11 6
Social networks 7% 0
Authentication Access control Data protection Trust
Wearable devices
Any
Fig. 5. Distribution of studies per tiers.

Fig. 4. General application domain.

B. Application Domain
sirable, raising privacy and security concerns. In recent years,
As shown in Fig. 4, the concern with security and data pri- the frameworks of edge computing [23] or fog computing [24]
vacy issues is transversal because it embraces many applica- have emerged as a solution for these above-mentioned issues. In-
tion domains. Most of primary studies are focused on solutions stead of performing all the computations in the remote cloud,
that can be applied to any application domain (61%), followed edge/fog computing proposes the decentralization of the com-
by 17% studies that addressed healthcare and 11% of smart putation through the offloading of some specific tasks to nodes
cities applications. The rest of publications are fairly distributed at the edge of the network, which are closer to the data sources.
among specific domains, with a slight predominance for Indus- Edge/fog nodes are devices not so computational powerful as
trial IoT (7%). These numbers indicate that our research object cloud data centers, but are able to provide location-aware ser-
is a general concern in the context of IoT. Efficient approaches vices and fast response to time-sensitive applications.
for handling security would have great potential to be adopted in By adopting the edge computing approach, an intermediate
a multitude of application domains and would represent a major tier of computing is introduced between the physical/IoT de-
advance in the field. vices and the cloud, giving rise to a 3-tier IoT ecosystem. There-
fore, the execution of both the workload generated by the appli-
C. Deployment of Security Techniques at the Architectural Tiers cations and by infrastructure-level components, such as those
Considering the heterogeneity of IoT environments, compu- responsible for the implementation of security aspects, can oc-
tational resources of things vary from constrained devices that cur in one of these three tiers. The need arises to make decisions
have limited CPU, memory, and power resources to more power- about the most appropriate tier to implement and deploy each se-
ful ones. When the workload generated by an application is very curity solution in an IoT System. Some security techniques re-
demanding in terms of computational resources, it is common to quire more extensive computing resources, while others need to
move part of the computational effort from constrained IoT de- be used as close to the data source as possible to ensure their pri-
vices to other places with more resources. This process is known vacy. Therefore, we consider that analyzing the retrieved studies
as computational offload and the natural candidate to accommo- from the prism of the architectural tier in which the presented
date the processing and long-term storage of the huge amount techniques were implemented can help guiding the design deci-
of IoT data is the cloud. By integrating IoT with the cloud, an sions of developers.
ecosystem composed of two architectural tiers emerges, with the We assume in this paper an architecture for IoT systems that
physical/things tier at the bottom and the cloud as the upper tier. comprises three distinct tiers: Thing, fog, and cloud, as pro-
However, cloud-based IoT systems suffer from a number of posed in [25]. The goal of considering where the techniques
limitations. The most prominent one is to deliver results in real presented in the studies are implemented, in such architecture,
time, as required by several IoT applications. Because clouds are is that the analysis carried out provide indications as to the ten-
clusters of data centers located in specific geographic locations, dencies with respect to such aspect.
the network overhead of moving data into and outside the core of However, as shown in Fig. 5, we did not observe a trend with
the cloud needs to be accounted for. Strict latency requirements regard to the placement of the proposed techniques, since there
are often incompatible with the unpredictable performance of was a balanced distribution of solutions over all tiers. The only
cloud-based analytics or controllers [22]. Another drawback of exception refers to the Authentication aspect, where the imple-
using traditional cloud platforms for IoT applications is the na- mentation in the “Thing + Fog” and “All” tiers had more articles.
ture of the data generated by many of these applications. As We believe that this behavior happens because authentication in-
we discuss throughout this paper, sensors embedded in the envi- volves at least two entities, so for example, devices on Thing tier
ronment often collect data of an extremely sensitive nature and authenticate either with another device or with a member of su-
send these data directly to remote data centers in the cloud with- perior tiers.
out any kind of treatment or anonymization, which is not de-
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 451

Security aspects ies), Industrial IoT (4 studies), followed by education, image

processing, military and wearable devices with 1 study each.
23 With this number, it is possible to affirm that authentication
is an aspect widely studied and of great interest of the scientific
community within IoT. There are some techniques that emerge
Authentication as trends, with curious emphasis on two based on mechanisms
42 92 originally designed to address other aspects of security: Elliptic-
Access control
curve cryptography (ECC) and datagram transport layer security
Data protection
(DTLS), as can be seen in Table 4. While the former is essen-
Trust tially an encryption technique, the latter falls within the group
of communication protocols.
The studies that implemented these techniques share the
49 same challenge of dealing with resource-constrained devices.
Lightweight approaches for these techniques were observed to
Fig. 6. Distribution of studies by security aspect. fit the universe of wireless sensors networks and IoT. In [28]
security keys using elliptic curve digital signature algorithm
IV. SECURITY ASPECTS ANALYSIS (ECDSA) with hash message digest are assigned to IoT nodes.
A DTLS handshake step at the edge (gateway) [29] and a re-
In this section we discuss the data extracted from primary fined and lightweight version DTLS [27], [30] are good exam-
studies with the perspective of security aspects considered in ples of these novel approaches. Besides these two, four other
this article to identify trends and research gaps. techniques originally designed for authentication also emerge
as trends. These, in turn, can be classified into two groups by
A. General Aspects Analysis similarity: OAuth and OpenID, which are widely diffused as
Fig. 6 shows the number of studies that addressed each se- standards for web and distributed applications in general; and
curity aspect analyzed in this SLR. Most of papers covers tech- biometrics and physical unclonable function (PUF), techniques
niques related to authentication, thus signaling a great interest of strongly related to physical devices.
research in techniques that implement authentication in an IoT It is important to remark that “Author Solution” stands for
context. On the other hand, few studies have investigated trust proprietary techniques that were proposed in the respective stud-
issues, which is the aspect least addressed in the primary studies, ies but not widely spread or adopted yet.
although it is obviously of great relevance in the IoT context. Contrary to the trends identified for the choice of techniques,
With the tendency in IoT to adopt distributed transactions and the implementation tier of them does not present any clear pref-
more relaxed business models, without formal contracts, trust erence in the selected studies. Although there is a slight pre-
between the parties is fundamental. Thus, we consider the study dominance for "Thing + Fog" tiers, it is reasonable to say that
and development of techniques that implement trust for the IoT Authentication solutions are distributed and can be present in all
context as a research gap that deserves further investigation. tiers of the reference architecture.
Regarding access control and data protection, we verified a
reasonable amount of studies that deal with these aspects, al- C. Access Control
though not as significant as authentication. Still, in Fig. 6, we Access control is the process of controlling requests to some
emphasize that absolute numbers were used instead of percent- resource from an authorized entity, allowing or denying access
ages since many papers addressed more than one security as- according to specified rules [69]. This aspect was unfolded as
pect. Thus, the number of articles that addressed each technique the second most found among primary studies counting 49 ar-
is higher than the number of selected primary studies. ticles. Most of studies concentrates in no specific application,
classified as “Any” (25 studies), followed by healthcare appli-
B. Authentication cations (12 studies), smart cities (6 studies), Industrial IoT (4
Authentication is the process of identifying a legitimate en- studies), and education and military (1 study each).
tity of a particular application [26]. As the authors in [27] state, Due to the number of studies that cover access control, we
device authentication has been shown to offer significant bene- might say that this is a quite relevant research area and re-
fits to IoT security architecture and it became one of the most veals tendency regarding the adoption of some techniques. Ta-
indispensable elements in IoT security ecosystems, due the fact ble 5 shows the most relevant techniques, presenting blockchain
that authentication is essential to offer other security aspects. (BC) as a promising one. The adoption of BC allows removing
Moreover, authentication acts as a first step of security for a sys- the need for a third-party institution that intermediate all trans-
tem, which reduces its exposure to malicious software. Our per- actions between entities (things). Considering that IoT takes
formed literature review reinforces this assertion because, from advantage of opportunistic interactions (random or serendipity
the four aspects chosen for evaluation, this is the most frequently encounters), establishing trust between devices is challenging
addressed on the primary studies, counting 92 related papers. since the devices might not be known by each other and they can
Due to its high relevance, its distribution among the applica- potentially be tampered. This way, BC approach could help to
tion domains does not differ significantly from the general view: deal with this challenge, providing an infrastructure that devices
Any (58 studies), healthcare (16 studies), smart cities (10 stud- could check the trustability of the devices. BC was originally
452 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO. 5, OCTOBER 2019

Table 4. Top techniques for authentication grouped by tiers.

Technique Tier Studies
Elliptic-curve cryptography (ECC) Thing; Thing + Fog; Thing + Cloud; Fog; All [26]–[28], [31]–[37]
Author solution Thing + Fog; Fog; Cloud; All [38]–[47]
OAuth Thing + Fog; Fog; Fog + Cloud; Cloud; All [48]–[53]
Datagram transport layer security (DTLS) Fog; All [29], [45], [54]–[59]
Biometric Thing; Thing + Fog [60]–[64]
OpenId Thing + Fog; Fog + Cloud [52], [53], [65]
Physical unclonable functions (PUF) Thing + Fog; All [66]–[68]

proposed for cryptocurrencies applications, but it rapidly spread and Industrial IoT applications tend to implement access control
to several other applications, serving as distributed database to outside things. This approach unfolds an evident concern about
store transactions in a tamper-proof manner. As an emerging having lightweight solutions to keep the devices alive for as long
technology, BC appears as a solution for access control in quite as possible, which make sense specially in these types of appli-
recent studies. For instance, in [70], the author presents an archi- cations. Within each tier of implementation there is no tendency
tecture based on BC to provide a fully distributed access control revealed for a specific technique, in the same way that within
system and arbitrate roles and permissions for IoT applications. each application domain.
Ouaddah et al. [71] propose an access control framework called
FairAccess which uses transactions on BC to grant, get, dele- D. Data Protection
gate and revoke access. Authors in [72] present four BC’s, each
one responsible for (i) interactions between things; (ii) context; As presented in [88], data protection essentially concerns
(iii) accountability; and (iv) access rules, that intents to general- the accountability to enable individual control over personal
ize different access control models. BC solutions do not focus data flow, making a commitment from the data collection step
on any specific application domain, and most of them are im- through the data dissemination, data processing, and data stor-
plemented in the fog or in cloud, not at the things. This makes age, i.e., a total fine-grained data management. Most of the tech-
sense because it is too expensive for a thing to contain the entire niques found in the primary studies are based on encryption to
chain of transactions and to participate in the mining process. provide data protection. This might be justified by the fact that
Other techniques that stand out are attribute-based access the ECC technique and others are used. For example, according
control (ABAC) and the ones that we name “Author Solution” to [89], the elliptical curve digital signature algorithm (ECDSA)
meaning proprietary solutions proposed by respective authors technique uses ECC to create a digital signature of data in order
of analyzed studies. Regarding ABAC, this technique basically to allow verifying its authenticity without compromising over-
uses attributes from users, environment, among others, to create all performance. The size of the key used in ECDSA is only
policies to control access. The authors in [76] propose an exten- 160 bits, which is very small as compared to pure DSA. This re-
sion for message queue telemetry transport (MQTT) to provide duces the communication overhead and improves the privacy of
a secure publish/subscribe system, not only considering authen- communication. The article also points out that ECDSA retains
tication techniques, but also policy enforcement using ABAC. the privacy of communication and improves security against at-
This technique is interesting for security since it provides ac- tackers, adding less number of bits to the original message as
cess control based on information beyond user’s authentication, compared to common used encryption schemes (e.g. Rivest-
that is, for example, the need to comply with specific attributes, Shamir-Adleman – RSA). This reduces the packet size and de-
like being in a given place, having a certain age, among others, creases the communication overhead, which contributes to a bet-
that must be in accordance with applications requirements to al- ter performance of the security technique.
low some user operation. In general, attribute-based approaches The authors in [90] point out specifically that public key cryp-
are relevant for IoT since they provide more flexibility for ap- tography (PKC) is considered unfeasible for wireless sensor net-
plications by allowing direct addressing data instead of commu- works in the early stages, so symmetric key-based security ar-
nication endpoints, which complies with data-centric systems chitectures were designed for this field. In [91], the authors men-
(as IoT). Besides, these techniques provide context-aware char- tion that with the rapid development of Industrial IoT (IIoT), a
acteristics, which makes access control challenging since more large number of resource-constrained devices with limited com-
granular rules will be needed to control access to more available munication, energy, and bandwidth is being used, which re-
data. Then, the analyzed studies cover what makes access con- quires the search for lightweight security methods, which cer-
trol a challenge, for instance, the mobility of nodes, the number tainly makes data protection a research challenge. The authors
of connected nodes and their vulnerabilities to be tampered, as implement an authentication and key agreement mechanism
well as their opportunistic interactions. based on implicit certificate, that is proposed based on open
In terms of the implementation place, access control tech- platform communications unified architecture (OPC UA) secu-
niques were spread implemented, revealing implementations at rity model. In the process of establishing a secure channel, the
the Things, Fog, and Cloud, tiers. It means that there is no trend lightweight ECC encryption algorithm is adopted to ensure the
about the most suitable place to deploy access control solutions. security of data transmission in the communication process. So,
Nevertheless, most of studies involving healthcare, smart cities, the authors show that this mechanism could be used in resource-
constrained environments. According to [52], ECC, unlike other
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 453

Table 5. Top techniques for access control grouped by tiers.

Technique Tier Studies
Blockchain Fog; Cloud; ALL [70]–[74]
Attribute-based access control (ABAC) Thing+ Fog; Fog + Cloud; Cloud [49], [75]–[77]
Author solutions Fog; Cloud [42], [45], [71]
Biometric Thing; ALL [78]–[80]
Smart card Thing; Thing + Fog [81]–[83]
Registration/subscription Thing + Fog; ALL [75], [84]
DCapBAC Thing + Fog; Fog + Cloud [52], [65]
Smart contracts Fog [70], [74]
Rule-based access control (RBAC) FOG; Cloud [49], [85]
SecIIot Thing + Fog; Fog [86], [87]

cryptographic schemes, requires lower computing and memory covered trust, data protection and access control techniques.
resources as well as smaller keys. With these studies we can see Regarding the place of the trust techniques implementation
that it is possible to have data protection even on end devices, in terms of the architectural tiers considered in this study, we
which is our suggestion if they are the origin of the data. verified the existence of a balanced distribution, since 6 studies
The implementation tier with more occurrences of data pro- implemented their technique in the Cloud, 6 in the Thing tier, 5
tection solutions was Thing + Fog. In general, this occurred Thing + Fog, 3 in all tiers and 3 implemented in the Fog. Consid-
due to the use of cryptographic techniques between these tiers, ering this even distribution of the implementation of these tech-
considering that data should be protected from its origin, as niques among the tiers, it was not possible to verify any trend
commented before. In this way, appropriated techniques to the of trust techniques regarding the placement of their code. Like-
computational constraints in each tier of the proposed architec- wise, as presented in [102], considering subjective user proper-
ture should be used, which consider restricted devices and a ties related to the concept of trust, it becomes difficult to iden-
resource-rich edge, approximating data protection implementa- tify where trust techniques should be implemented, making the
tion to the place from where data is generated. choice of location of the implementation dependent on the ap-
In Table 6, the most relevant techniques found in the primary plication features. For instance, in [103] authors choose to im-
studies cover cryptographic solutions, with prominence to ECC- plement the trust solution in Cloud since the application require-
based approaches. This behavior suggests a direct relation be- ment for classifying nodes has heavy computational effort to be
tween data protection and authentication techniques. The three placed on another tier. Nguyen et al. [104] present a lightweight
application domains with more occurrences, excluding the gen- trust solution implemented on things to deliver an initial trust
eral domain (Any), were healthcare, smart cities and Industrial level for devices that have not previously encountered each other
IoT. This observation reveals a correspondence to other aspects within the range of a personal area network (PAN). So, the so-
and to the general case. Also, we can observe that ECC and AES lution was implemented at the Thing tier.
are commonplace techniques for this security aspect. With respect to the techniques themselves, no highlights
could be observed. Of the 23 papers, two used blockchain-based
E. Trust solutions, while the other papers applied different techniques, as
shown in Table 7, which depicts the techniques used by the 10
The concept of trust is different for human community and best articles according to the quality criteria stipulated for this
IoT devices. Among people, trust is built based not only on oth- secondary study. It is worth mentioning that these quality crite-
ers’ behavior and reputation, but also on the experience, knowl- ria do not relate to the best articles in general, but to those that
edge, and other attributes of the observer. For instance, the trust best fit the criteria that this study aimed to verify.
level assigned by an observer to another person can be influ- A relevant fact related to the techniques employed was the
enced by cultural and environmental aspects of the observer. On use of blockchain to guarantee trust. Because it is an emerg-
the other hand, for IoT devices A and B, the trust level assigned ing technology, blockchain still presents some shortcomings,
to B by A is based only on the behavior of B, i.e., the behavior among them issues related to scalability [109]. In [109], authors
of A does not influence the evaluation of trust of B. propose a food supply chain traceability system for real-time
According to [102], trust is a complicated concept, which food tracking based on the implementation of blockchain. As
comprises many aspects such as confidence, belief, integrity, described in Section IV.C, authors in [71] proposed FairAccess,
among others. Considering this fact along with our analysis, we a framework for access control in IoT based on Blockchain, that,
come to a concept of trust that comprises identity assurance, de- besides giving access control solution, gives trust solution also.
vice’s behavior, data integrity and protection. Therefore, trust Finally, based on the lack of standardization of techniques to
spreads through other aspects considered in this study such as implement trust and the small number of studies that have pre-
authentication and data protection. sented proposals for this aspect, we consider trust a research
Out of the 131 primary studies selected, only 23 papers ad- gap. A more detailed study of such aspect would be important
dressed techniques related to trust, so this was the aspect least and necessary, with a particular focus on standardization and
addressed by the studies. It‘s important to note that some stud- convergence of definitions and techniques.
ies discussed more than one security aspect, such as [84] that
454 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO. 5, OCTOBER 2019

Table 6. Top techniques for data protection grouped by tiers.

Technique Tier Studies
Elliptic-curve cryptography (ECC) Thing; Thing + Fog; Cloud; All [27], [37], [52], [83], [90]–[94]
Advanced encryption standard (AES) Thing + Fog; Cloud [45], [93], [95], [96]
Cryptography Fog; Thing + Fog; All [48], [60], [84], [97]
Attribute-based encryption (ABE) Thing + Fog; Fog + Cloud [98], [99]
Datagram transport layer security (DTLS) Thing + Fog [99], [100]
Elliptic-curve digital signature algorithm (ECDSA) Fog; Thing + Fog [71], [89]
OpenSSL Thing + Fog [42], [45]
XOR operations Thing; Fog [78], [79]
Blockchain Fog [73]
One time password (OTP) Fog + Cloud [101]

Table 7. Top techniques for trust grouped by tiers.

Technique Tier Studies
Couterfeit detection, traceability and authentication (CDTA) Thing [92]
Author solution Thing [28], [104]
MRC scheme with TM security algorithm Thing [105]
Knowledge base systems (KBS) Cloud [93]
Social-based trusted solution Cloud [106]
Malleable signature schemes (MSS) Thing [107]
Fuzzy Thing + Fog [52]
Remote attestation model Thing + Fog [108]
Machine learning Cloud [103]
Elastic slide window Cloud [103]

V. DISCUSSION years, which shows an increasing interest on IoT security by

the research community. This also indicates that more attention
Through the analysis of the obtained results, we could iden- might be expected to this area on the upcoming years. Among
tify that trust in IoT is a research gap with few studies exploring eliminated studies, a significant number of those that lack from
this aspect. Also, a considerable number of studies that dealt experiments that would evaluate their approaches and solutions,
with authentication aspect, using ECC, “Author Solutions” and which helps readers to confirm their feasibility and efficiency.
OAuth-based techniques. The ECC technique is reasonable to We would like to highlight the importance of having a holis-
appear in this aspect since it is well-established. The “Author tic view of security to protect IoT building blocks and provide
Solutions” is a mix of different solutions proposed by the au- security by design. Finally, there is a lack of a well-defined ar-
thors of the primary studies. And what catches some attention is chitecture that considers security aspects that could serve as a
the adoption of OAuth-based techniques that was thought to be reference architecture to develop IoT security solutions.
an open standard for API access delegation.
Concerning the decision regarding the deployment of security
solutions (in a 3-tier IoT architecture), our analysis showed that, VI. CONCLUDING REMARKS
for authentication, a minimum of two entities are involved in the
In order to obtain the studies to be analyzed in this paper, we
authentication process. The entities might belong to a specific
followed a meticulous and systematic process. We presented our
tier or to more than one tier. This way, the place of authentica-
research methodology through an SLR and showed the details
tion implementation should be the same of those entities’ tier.
of the searches and the obtained results, besides providing some
For access control, the deployment of solutions regarding this
numerical data. As future studies, we aim to consider other se-
security aspect should consider lightweight implementations at
curity aspects (e.g., privacy and secure middleware), so as to
the Thing tier whenever possible and pervade all other tiers in
present different perspectives of security in IoT. We also intend
which an IoT application resides. Concerning data protection,
to follow the results of our study to pursue the revealed trends
the implementation of this aspect should be done from the data
and propose new techniques to overcome research gaps.
generation to all involved tiers, traversing the entire life cycle of
data processing. Similar to access control, the implementation
of trust techniques should consider all tiers covered by an IoT REFERENCES
application. These results underscore the holistic nature of se-
[1] L. Atzori, A. Iera, and G. Morabito, “The internet of things: A survey,”
curity solutions for IoT. In order to achieve the true potential of Comput. Netw., vol. 54, no. 15, pp. 2787–2805, Oct 2010.
this technology, it is essential to develop systems where security [2] Ericsson, “Ericsson mobility report,” Tech. Rep., Ericsson, 2017.
permeates all tiers from the moment a sensitive data gains some [3] A. Al-Fuqaha et al, “Internet of things: A survey on enabling technolo-
gies, protocols, and applications,” IEEE Commun. Surveys Tuts., vol. 17,
context to its consumption by the user’s applications. no. 4, pp. 2347–2376, Fourthquarter 2015.
Almost all analyzed studies were published in the last three [4] K. T. Nguyen, M. Laurent, and N. Oualha, “Survey on secure commu-
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 455

nication protocols for the internet of things,” Ad Hoc Netw., vol. 32, [32] J. Shen, S. Chang, J. Shen, Q. Liu, and X. Sun, “A lightweight multi-
pp. 17–31, 2015. layer authentication protocol for wireless body area networks,” Future
[5] K. Zhao and L. Ge, “A survey on the internet of things security,” in Proc. Generation Comput. Syst., vol. 78, pp. 956–963, Jan. 2018.
CISIS, Dec. 2013, pp. 663–667. [33] M. Turkanović, B. Brumen, and M. Hölbl, “A novel user authentication
[6] Q. Jing, A. V. Vasilakos, J. Wan, J. Lu, and D. Qiu, “Security of the and key agreement scheme for heterogeneous ad hoc wireless sensor net-
internet of things: Perspectives and challenges,” Wireless Netw., vol. 20, works, based on the internet of things notion,” Ad Hoc Netw., vol. 20,
pp. 2481–2501, Nov. 2014. pp. 96–112, Sept. 2014.
[7] S. Sicari, A. Rizzardi, L. Grieco, and A. Coen-Porisini, “Security, privacy [34] K.-H. Wang, C.-M. Chen, W. Fang, and T.-Y. Wu, “A secure authen-
and trust in internet of things: The road ahead,” Comput. Netw., vol. 76, tication scheme for internet of things,” Pervasive and Mobile Comput.,
pp. 146–164, Jan. 2015. vol. 42, pp. 15–26, Dec. 2017.
[8] I. Ali and Z. Ullah, “Internet of things security, device authentication and [35] X. Zhang, S. Poslad, and Z. Ma, “A semi-outsourcing secure data privacy
access control: A review,” International J. Comput. Science Inf. security, scheme for IoT data transmission,” in Proc. IEEE PIMRC, Oct. 2017,
vol. 14, no. 8, Aug. 2016, pp. 456–466. pp. 1–5.
[9] F. A. Alaba, M. Othman, I. A. T. Hashem, and F. Alotaibi, “Internet of [36] Z. Mahmood et al, “Secure authentication and prescription safety proto-
things security: A survey,” J. Netw. Comput. Appl., vol. 88, pp. 10–28, col for telecare health services using ubiquitous IoT,” Applied Sciences,
June 2017. vol. 7, p. 1069, Oct. 2017.
[10] D. E. Kouicem, A. Bouabdallah, and H. Lakhlef, “Internet of things se- [37] S. Ullah, B. Rinner, and L. Marcenaro, “Smart cameras with onboard
curity: A top-down survey,” Comput. Netw., vol. 141, pp. 199–221, Aug. signcryption for securing IoT applications,” in Proc. IEEE GIoTS, June
2018. 2017, pp. 1–6.
[11] M. Ammar, G. Russello, and B. Crispo, “Internet of things: A survey on [38] R. Amin, N. Kumar, G. Biswas, R. Iqbal, and V. Chang, “A light weight
the security of IoT frameworks,” J. Inf. Security Appl., vol. 38, pp. 8–27, authentication protocol for IoT-enabled devices in distributed cloud
Feb. 2018. computing environment,” Future Generation Comput. Syst., vol. 78,
[12] M. A. Khan and K. Salah, “IoT security: Review, blockchain solu- pp. 1005–1019, Jan. 2018.
tions, and open challenges,” Future Generation Comput. Syst., vol. 82, [39] F. Wu et al, “A lightweight and privacy-preserving mutual authentication
pp. 395–411, May 2018. scheme for wearable devices assisted by cloud server,” Comput. Electri-
[13] A. R. Sfar, E. Natalizio, Y. Challal, and Z. Chtourou, “A roadmap for cal Engineering, vol. 63, pp. 168–181, Oct. 2017.
security challenges in the internet of things,” Digital Commun. Netw., [40] F. Wu et al, “A lightweight and robust two-factor authentication scheme
vol. 4, no. 2, pp. 118–137, Apr. 2018. for personalized healthcare systems using wireless medical sensor net-
[14] M. Irshad, “A Systematic Review of Information Security Frameworks in works,” Future Generation Comput. Syst., vol. 82, pp. 727–737, May
the internet of things,” in Proc. IEEE HPCC, Dec. 2016, pp. 1270–1275. 2018.
[15] B. Kitchenham et al, “Systematic literature reviews in software engineer- [41] Z. Wang, “A privacy-preserving and accountable authentication protocol
ing – a systematic literature review,” Inf. Software Tech., vol. 51, no. 1, for IoT end-devices with weaker identity,” Future Generation Comput.
pp. 7–15, Jan. 2009. Syst., vol. 82, pp. 342–348, May 2018.
[16] P. Brereton, B. A. Kitchenham, D. Budgen, M. Turner, and M. Khalil, [42] M. Tao, K. Ota, M. Dong, and Z. Qian, “AccessAuth: Capacity-aware
“Lessons from applying the systematic literature review process within security access authentication in federated-IoT-enabled V2G networks,”
the software engineering domain,” J. Syst. Software, vol. 80, no. 4, J. Parallel Distributed Comput., vol. 118, no. 1, pp. 107–117, Aug. 2018.
pp. 571–583, Apr. 2007. Software Performance. [43] M. S. Farash, M. Turkanović, S. Kumari, and M. Hölbl, “An efficient
[17] E. Cavalcante et al, “On the interplay of internet of things and cloud user authentication and key agreement scheme for heterogeneous wire-
computing: A systematic mapping study,” Comput. Commun., vol. 89-90, less sensor network tailored for the internet of things environment,” Ad
pp. 17 – 33, Sept. 2016. Hoc Netw., vol. 36, no. 1, pp. 152–176, Jan. 2016.
[18] Y. Yang, L. Wu, G. Yin, L. Li, and H. Zhao, “A survey on security and pri- [44] W.-L. Tai, Y.-F. Chang, and W.-H. Li, “An IoT notion–based authenti-
vacy issues in internet-of-things,” IEEE Internet Things J., vol. 4, no. 5, cation and key agreement scheme ensuring user anonymity for heteroge-
pp. 1250–1258, Oct. 2017. neous ad hoc wireless sensor networks,” J. Inf. Security Appl., vol. 34,
[19] B. Kitchenham and S. Charters, “Guidelines for performing systematic no. 2, pp. 133–141, June 2017.
literature reviews in software engineering,” Tech. Rep. EBSE2007-01, [45] S. R. Moosavi et al, “End-to-end security scheme for mobility enabled
2007. healthcare internet of things,” Future Generation Comput. Syst., vol. 64,
[20] C. M. D. Farias et al, “A systematic review of shared sensor networks,” pp. 108–124, Nov. 2016.
ACM Comput. Surv., vol. 48, no. 4, pp. 1–50, Feb. 2016. [46] X.-J. Lin, L. Sun, and H. Qu, “Insecurity of an anonymous authentication
[21] C. Wohlin et al, Experimentation in Software Engineering: An Introduc- for privacy-preserving IoT target-driven applications,” Comput. Security,
tion. Norwell, MA, USA: Kluwer Academic Publishers, 2000. vol. 48, pp. 142–149, Feb. 2015.
[22] B. Zhang et al, “The cloud is not enough: Saving iot from the cloud,” in [47] Y.-Y. Deng, C.-L. Chen, W.-J. Tsaur, Y.-W. Tang, and J.-H. Chen, “In-
Proc. USENIX Workshop, July 2015. ternet of things (IoT) based design of a secure and lightweight body area
[23] P. Garcia Lopez et al, “Edge-centric computing: Vision and chal- network (BAN) Healthcare System,” Sensors, vol. 17, no. 12, p. 2919,
lenges,”SIGCOMM Comput. Commun. Rev., vol. 45, no. 5, pp. 37–42, Dec. 2017.
Sept. 2015. 2015. [48] L. A. B. Pacheco, E. Alchieri, and P. A. Barreto, “Enhancing and eval-
[24] F. Bonomi, R. Milito, J. Zhu, and S. Addepalli, “Fog computing and its uating an architecture for privacy in the integration of internet of things
role in the internet of things,” in Proc. ACM MCC Workshop, Aug. 2012, and cloud computing,” in Proc. IEEE NCA, Jan. 2017, pp. 1–8.
pp. 13–16. [49] A. Álonso, F. Fernández, L. Marco, and J. Salvachúa, “IAACaaS: IoT
[25] W. Li et al, “System modelling and performance evaluation of a three-tier application-scoped access control as a service,” Future Internet, vol. 9,
cloud of things,” Future Generation Comput. Syst., vol. 70, pp. 104–125, no. 4, pp. 1–14, July 2017.
May 2017. [50] S. Kinikar and S. Terdal, “Implementation of open authentication proto-
[26] S. Kalra and S. K. Sood, “Secure authentication scheme for IoT and col for IoT based application,” in Proc. IEEE ICICT, Aug. 2016, pp. 1–4.
cloud servers,” Pervasive and Mobile Comput., vol. 24, pp. 210–223, [51] S. Cirani, M. Picone, P. Gonizzi, L. Veltri, and G. Ferrari, “IoT-OAS: An
Dec. 2015. OAuth-based authorization service architecture for secure services in IoT
[27] K. H. Yeh, “A Secure IoT-based healthcare system with body sensor net- scenarios,” IEEE Sensors J., vol. 15, pp. 1224–1234, Feb. 2015.
works,” IEEE Access, vol. 4, pp. 10288–10299, Dec. 2016. [52] J. Bernal Bernabe, J. L. Hernandez Ramos, and A. F. Skarmeta Gomez,
[28] M. B. Krishna and P. Lorenz, “Delay aware secure hashing for op- “TACIoT: multidimensional trust-aware access control system for the in-
portunistic message forwarding in internet of things,” in Proc. IEEE ternet of things,” Soft Comput., vol. 20, no. 5, pp. 1763–1779, May 2016.
GLOBECOM Workshops, Dec. 2017, pp. 1–6. [53] S. H. Lee, K. W. Huang, and C. S. Yang, “TBAS: Token-based authoriza-
[29] A. Rajagopalan, M. Jagga, A. Kumari, and S. T. Ali, “A DDoS prevention tion service architecture in internet of things scenarios,” International J.
scheme for session resumption SEA architecture in healthcare IoT,” in Distributed Sensor Netw., vol. 13, no. 7, July 2017.
Proc. IEEE CICT, Feb. 2017, pp. 1–5. [54] T. Kothmayr, C. Schmitt, W. Hu, M. Brunig, and G. Carle, “A DTLS
[30] S. L. Keoh, S. S. Kumar, and H. Tschofenig, “Securing the internet of based end-to-end security architecture for the internet of things with two-
things: A standardization perspective,” IEEE Internet Things J., vol. 1, way authentication,” in Proc. IEEE LCN Workshops, Oct. 2012, pp. 956–
no. 3, June 2014. 963.
[31] S. C. Cha, J. F. Chen, C. Su, and K. H. Yeh, “A blockchain connected [55] G. Lessa dos Santos, V. T. Guimaraes, G. da Cunha Rodrigues, L. Z.
gateway for BLE-based devices in the internet of things,” IEEE Access, Granville, and L. M. R. Tarouco, “A DTLS-based security architecture
vol. 6, pp. 24639–24649, Jan. 2018. for the internet of things,” in Proc. IEEE ISCC, July 2015, pp. 809–815.
456 JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 21, NO. 5, OCTOBER 2019

[56] M. A. ElAffendi and A. L. Alamudy, “Could virtualization be the ulti- [81] F. Wu et al, “An efficient authentication and key agreement scheme for
mate solution for IoT resource constrained devices problem? A multi- multi-gateway wireless sensor networks in IoT deployment,” J. Netw.
level security framework based on device virtualization,” in Proc. ICCA, Comput. Appl., vol. 89, pp. 72–85, July 2017.
Sept. 2017, pp. 232–237. [82] F. Wu, L. Xu, S. Kumari, and X. Li, “A privacy-preserving and prov-
[57] R. Hummen, H. Shafagh, S. Raza, T. Voig, and K. Wehrle, “Delegation- able user authentication scheme for wireless sensor networks based on
based authentication and authorization for the IP-based internet of internet of things security,” J. Ambient Intelligence Humanized Comput.,
things,” in Proc. SECON, June 2014, pp. 284–292. vol. 8, no. 1, pp. 101–116, Feb. 2017.
[58] T. Kothmayr, C. Schmitt, W. Hu, M. Brünig, and G. Carle, “DTLS based [83] T. D. P. Bai and S. A. Rabara, “Design and development of integrated,
security and two-way authentication for the internet of things,” Ad Hoc secured and intelligent architecture for internet of things and cloud com-
Netw., vol. 11, pp. 2710–2723, Nov. 2013. puting,” in Proc. IEEE FiCloud, Aug. 2015, pp. 817–822.
[59] S. R. Moosavi et al., “SEA: A secure and efficient authentication and au- [84] M. Taylor, D. Reilly, and B. Lempereur, “An access control management
thorization architecture for IoT-based healthcare using smart gateways,” protocol for internet of things devices,” Netw. Security, vol. 2017, no. 7,
Procedia Comput. Science, vol. 52, pp. 452–459, Jan. 2015. pp. 11–17, July 2017.
[60] X. Li et al, “A three-factor anonymous authentication scheme for wire- [85] X. Huang, P. Craig, H. Lin, and Z. Yan, “SecIoT: A security framework
less sensor networks in internet of things environments,” J. Netw. Com- for the internet of things,” Security and Commun. Netw., vol. 9, pp. 3083–
put. Appl., vol. 103, no. 1, pp. 194–204, Feb. 2017. 3094, Nov. 2016.
[61] P. Peris-Lopez, L. González-Manzano, C. Camara, and J. M. de Fuentes, [86] G. Chen and W. S. Ng, “An efficient authorization framework for secur-
“Effect of attacker characterization in ECG-based continuous authenti- ing industrial internet of things,” in Proc. IEEE TENCON, Nov. 2017,
cation mechanisms for internet of things,” Future Generation Comput. pp. 1219–1224.
Syst., vol. 81, pp. 67–77, Apr. 2018. [87] G. Chen and W. S. Ng, “An efficient authorization framework for secur-
[62] Y. Lu et al, “Exploring finger vein based personal authentication for se- ing industrial internet of things,” in Proc. IEEE TENCON, Dec. 2017,
cure IoT,” Future Generation Comput. Syst., vol. 77, pp. 149–160, Dec. pp. 1219–1224.
2017. [88] A. Crabtree, T. Lodge, J. Colley, C. Greenghalgh, and R. Mortier, “Ac-
[63] J. Srinivas, S. Mukhopadhyay, and D. Mishra, “Secure and efficient user countable internet of things? Outline of the IoT databox model,” in Proc.
authentication scheme for multi-gateway wireless sensor networks,” Ad IEEE WoWMoM, June 2017, pp. 1–6.
Hoc Netw., vol. 54, pp. 147–169, Jan. 2017. [89] M. B. Tamboli and D. Dambawade, “Secure and efficient CoAP based
[64] M. W. Condry and C. B. Nelson, “Using smart edge IoT devices for authentication and access control for internet of things (IoT),” in Proc.
safer, rapid response with industry IoT control operations,” Proc. IEEE, IEEE RTEICT, May 2016, pp. 1245–1250.
vol. 104, no. 5, May 2016, pp. 938–946. [90] M. Lavanya and V. Natarajan, “Lightweight key agreement protocol
[65] M. Hossain, S. M. Islam, F. Ali, K. S. Kwak, and R. Hasan, “An inter- for IoT based on IKEv2,” Comput. Electrical Engineering, vol. 64,
net of things-based health prescription assistant and its security system pp. 580–594, Nov. 2017.
design,” Future Generation Comput. Syst., vol. 82, pp. 422–439, May [91] M. Wei, S. Zhang, P. Wang, and K. Kim, “An authentication and key
2018. agreement mechanism for OPC unified architecture in industrial internet
[66] U. Chatterjee, R. S. Chakraborty, and D. Mukhopadhyay, “A PUF-based of things,” International J. Distributed Sensor Netw., vol. 14, pp. 1–11,
secure communication pProtocol for IoT,” ACM Trans. Embedded Com- Jan. 2018.
put. Syst., vol. 16, no. 3, pp. 1–25, Apr. 2017.
[92] K. Yang, D. Forte, and M. M. Tehranipoor, “Cdta: A comprehensive
[67] W. Feng, Y. Qin, S. Zhao, and D. Feng, “AAoT: Lightweight attesta-
solution for counterfeit detection, traceability, and authentication in the
tion and authentication of low-resource things in IoT and CPS,” Comput.
iot supply chain,” ACM Trans. Des. Autom. Electron. Syst., vol. 22, no. 3,
Netw., vol. 134, pp. 167–182, Apr. 2018.
pp. 1–31, Apr. 2017.
[68] C. Huth, J. Zibuschka, P. Duplys, and T. Guneysu, “Securing systems on
[93] a. J. Jara, M. a. Zamora-Izquierdo, and a. F. Skarmeta, “Interconnection
the internet of things via physical properties of devices and communica-
framework for mHealth and remote monitoring based on the internet of
tions,” in Proc. IEEE SysCon, Apr. 2015, pp. 8–13.
things,” IEEE J. Selected Areas Commun., vol. 31, no. 9, pp. 47–65, Aug.
[69] R. Sandhu and P. Samarati, “Access control: Principle and practice,”
2013.
IEEE Commun. Mag., vol. 32, no. 9, pp. 40–48, Sept. 1994.
[94] M. Saadeh, A. Sleit, K. E. Sabri, and W. Almobaideen, “Lightweight
[70] O. Novo, “Blockchain Meets IoT: An architecture for scalable ac-
identity based signature for mobile object authentication in the inter-
cess management in IoT,” IEEE Internet Things J., vol. 5, no. 2,
net of things,” J. Theoretical and Applied Inf. Technol., vol. 96, no. 3,
pp. 1184–1195, Mar. 2018.
pp. 788–798, Feb. 2018.
[71] A. Ouaddah, A. Abou Elkalam, and A. Ait Ouahman, “FairAccess:
A new blockchain-based access control framework for the internet of [95] M. A. Jan, F. Khan, M. Alam, and M. Usman, “A payload-based mutual
things,” Security and Commun. Netw., vol. 9, no. 18, pp. 5943–5964, authentication scheme for internet of things,” Future Generation Comput.
Dec. 2016. Syst., vol. 92, pp. 1028–1039, Mar. 2019.
[72] O. J. A. Pinno, A. R. A. Gregio, and L. C. De Bona, “ControlChain: [96] P. Hu et al, “Security and privacy preservation scheme of face identifica-
Blockchain as a central enabler for access control authorizations in the tion and resolution framework using fog computing in internet of things,”
IoT,” in Proc. IEEE GLOBECOM, Dec. 2018, pp. 1–6. IEEE Internet Things J., vol. 4, no. 5, pp. 1143–1155, Oct. 2017.
[73] N. Baracaldo et al, “Securing data provenance in internet of things (IoT) [97] B. Ndibanje et al, “A Secure and efficient mutual authentication hand-
systems,” Lecture Notes in Computer Science (including subseries Lec- off protocol for sensors devices support in internet of things,” Sensors
ture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Materials, vol. 29, no. 7, p. 1, Jan. 2017.
vol. 10380, pp. 92–98, Oct. 2017. [98] S. Zebboudj et al, “Big data source location privacy and access control in
[74] N. Rifi, E. Rachkidi, N. Agoulmine, and N. C. Taher, “Towards using the framework of IoT,” in Proc. ICEE-B, Oct. 2017, pp. 1–5.
blockchain technology for eHealth data access management,” in Proc. [99] J. Choi, Y. In, C. Park, S. Seok, H. Seo, and H. Kim, “Secure IoT frame-
IEEE ICABME, Oct. 2017, pp. 1–4. work and 2D architecture for end-To-end security,” J. Supercomputing,
[75] A. Alshehri and R. Sandhu, “Access control models for virtual object vol. 74, no. 8, pp. 3521–3535, Aug. 2018.
communication in cloud-enabled IoT,” in Proc. IEEE IRI, Jan. 2017, [100] A. Haroon, S. Akram, M. A. Shah, and A. Wahid, “E-Lithe: A
pp. 16–25. lightweight secure DTLS for IoT,” in Proc. IEEE VTC, Sept. 2017, pp. 1–
[76] A. Rizzardi, S. Sicari, D. Miorandi, and A. Coen-porisini, “AUPS : An 5.
open source authenticated publish / subscribe system for the internet of [101] S. Lee et al, “Design and implementation for data protection of energy iot
things,” Inf. Syst., vol. 62, pp. 29–41, Dec. 2016. utilizing otp in the wireless mesh network,” Energy Procedia, vol. 141,
[77] U. Salama, L. Yao, X. Wang, H. Y. Paik, and A. Beheshti, “Multi-level pp. 540–544, Dec. 2017.
privacy-preserving access control as a service for personal healthcare [102] Z. Yan, P. Zhang, and A. V. Vasilakos, “A survey on trust management
monitoring,” in Proc. IEEE ICWS, June 2017, pp. 878–881. for internet of things,” J. Netw. Comput. Appl., vol. 42, pp. 120–134, June
[78] N. Kadham and K. Sreenivasa Ravi, “A lightweight one time password 2014.
(OTP) based smart learning in internet of things,” International J. Engi- [103] J. Caminha, A. Perkusich, and M. Perkusich, “A smart trust management
neering Tech., vol. 7, no. 2.7, pp. 480–483, Mar. 2018. method to detect on-off attacks in the internet of things,” Security Com-
[79] P. K. Dhillon and S. Kalra, “A lightweight biometrics based remote user mun. Netw., vol. 2018, pp. 1–10, Apr. 2018.
authentication scheme for IoT services,” J. Inf. Security Appl., vol. 34, [104] T. Nguyen, D. Hoang, D. Nguyen, and A. Seneviratne, “Initial trust es-
no. 2, pp. 255–270, June 2017. tablishment for personal space iot systems,” in Proc. IEEE INFOCOM
[80] D. Shah and V. Haradi, “IoT based biometrics implementation on rasp- Workshop, May 2017, pp. 784–789.
berry Pi,” Procedia Comput. Science, vol. 79, pp. 328–336, Dec. 2016. [105] J. I. Chen and D. J. Chuang, “Embedding of mrc in tm to increase the
MACEDO et al.: ON THE SECURITY ASPECTS OF INTERNET OF THINGS: A ... 457

security for iot technologies,” in Proc. IEEE iCAST, Nov. 2017, pp. 199– Rui R. Mello Jr. received his B.Sc. degree in 2004
204. in Computer Science from Catholic University of
[106] L. Militano, A. Orsino, G. Araniti, and A. Iera, “NB-IoT for D2D- Petrópolis. He received a M. Sc. degree in Systems
enhanced content uploading with social trustworthiness in 5g systems,” Engineering and Computation from Federal Univer-
Future Internet, vol. 9, no. 3, July2017. sity of Rio de Janeiro, in 2015. He is Lieutenant
[107] H. C. Pöhls et al, “Rerum: Building a reliable iot upon privacy- and Commander from Brazilian Navy and member of the
security- enabled smart objects,” in Proc. IEEE WCNCW Workshop, Apr. Brazilian Navy Research Institute. He is currently
2014, pp. 122–127. a D.Sc. student in Architectures and Operating Sys-
[108] B. Gong, Y. Zhang, and Y. Wang, “A remote attestation mechanism for tems, System Engineering and Computation Program
the sensing layer nodes of the internet of things,” Future Generation at Federal University of Rio de Janeiro.
Comput. Syst., vol. 78, pp. 867–886, Jan. 2018.
[109] F. Tian, “A supply chain traceability system for food safety based on
haccp, blockchain amp;amp; internet of things,” in Proc. ICSSSM, June
2017, pp. 1–6.
Felipe M. G. França received his B.Sc. and M.Sc. de-
grees in Electronic Engineering and System Engineer-
ing and Computation, respectively, from Federal Uni-
versity of Rio de Janeiro (UFRJ), in 1982 and 1987.
He received Ph.D. degree in Neural Systems Engi-
neering from Imperial College of Science Technol-
Evandro L. C. Macedo received his B.S. degree on ogy and Medicine, in 1994. He is member of the In-
Computing and Information Technology from State stitute of Electrical and Electronics Engineers (IEEE)
University of Rio de Janeiro (UERJ), Brazil in 2011. and Professor at UFRJ.
He acquired his Master’s degree on Systems Engi-
neering and Computing from the Systems Engineering
and Computing Program , Federal University of Rio
de Janeiro (UFRJ), Brazil in 2015. He is currently a
Ph.D. student in Systems Engineering and Computing
Program, UFRJ. Flavia C. Delicato has a Ph.D. in Electrical and Com-
puter Engineering (2005) from the Federal Univer-
sity of Rio de Janeiro. She is an Associate Professor
of Computer Science at Federal University of Rio de
Janeiro, Brazil. She is the author of 2 Books, more
than 100 papers and participates in several research
projects with funding from International and Brazil-
Egberto A. R. de Oliveira received his B.S. degree on ian government agencies. She is a level 1 Researcher
Computing and Information Technology from State Fellow of the Brazilian National Council for Scientific
University of Rio de Janeiro (UERJ), Brazil in 2009. and Technological Development.
He is currently a M. SC. student in Systems Engineer-
ing and Computing Program at Federal University of
Rio de Janeiro (PESC/COPPE/UFRJ) and Senior IT
Analyst at Fundação Vale do Rio Doce de Seguridade
Social (Valia). He is also a Microsoft Certified So- José F. de Rezende received B.Sc. and M.Sc. degrees
lutions Expert: Cloud platform & infrastructure and in Electronics Engineering from Federal University of
AWS certified solutions architect - associate. Rio de Janeiro in 1988 and 1991, respectively. He
received the Ph.D. degree in Computer Science from
Universit’e Pierre et Marie Curie in 1997. Since 1998
he has been an associate professor at UFRJ. He has
served in the editorial board of Ad Hoc Networks from
Elsevier since 2006.
Fabio H. Silva received his Master degree in Com-
puting and Systems from Military Institute of Engi-
neering - IME (2010), graduate in Technology for In-
ternet Systems from Federal Center for Technological
Education of Rio de Janeiro - CEFET/RJ (2007). Cur- Luís F. M. de Moraes received a B.S. and M.S. de-
rently is a Ph.D. student in Systems Engineering and grees in Electrical Engineering and Telecommunica-
Computing Program at Federal University of Rio de tions from Pontifical Catholic University of Rio de
Janeiro - COPPE/UFRJ, technician in National Insti- Janeiro (PUC-Rio), in 1973 and 1976, respectively.
tute of Technology - INT, Assistant Professor in Cari- He received Ph.D. degree in Electrical Engineering
oca University Centre - Unicarioca. and System Science from University of California
at Los Angeles (UCLA) in 1981. He is Professor at
Federal University of Rio de Janeiro and technical-
scientific coordinator of the REDERIO backbone, a
pioneering Internet project in Rio de Janeiro, Brazil,
supported by FAPERJ, since 1992.