IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO.

5, OCTOBER 2017 1125

A Survey on Internet of Things: Architecture,

Enabling Technologies, Security and
Privacy, and Applications
Jie Lin, Wei Yu, Nan Zhang, Xinyu Yang, Hanlin Zhang, and Wei Zhao

Abstract—Fog/edge computing has been proposed to be services for customers or applications in the space between
integrated with Internet of Things (IoT) to enable computing networking central servers and end-users [16], [147]. In
services devices deployed at network edge, aiming to improve fog/edge computing, the massive data generated by different
the user’s experience and resilience of the services in case of fail-
ures. With the advantage of distributed architecture and close to kinds of Internet of Things (IoT) devices can be processed at
end-users, fog/edge computing can provide faster response and the network edge instead of transmitting it to the centralized
greater quality of service for IoT applications. Thus, fog/edge cloud infrastructure due to bandwidth and energy consumption
computing-based IoT becomes future infrastructure on IoT devel- concerns [103], [116]. Because fog/edge computing is orga-
opment. To develop fog/edge computing-based IoT infrastructure, nized as distributed architecture and can process data and store
the architecture, enabling techniques, and issues related to IoT
should be investigated first, and then the integration of fog/edge data in networking edge devices, which is close to end-users,
computing and IoT should be explored. To this end, this paper fog/edge computing can provide services with faster response
conducts a comprehensive overview of IoT with respect to system and greater quality, in comparison with cloud computing [147].
architecture, enabling technologies, security and privacy issues, Thus, fog/edge computing is more suitable to be integrated
and present the integration of fog/edge computing and IoT, and with IoT to provide efficient and secure services for a large
applications. Particularly, this paper first explores the relation-
ship between cyber-physical systems and IoT, both of which play number of end-users, and fog/edge computing-based IoT can
important roles in realizing an intelligent cyber-physical world. be considered as the future IoT infrastructure [16].
Then, existing architectures, enabling technologies, and security To design and deploy fog/edge computing-based IoT, the
and privacy issues in IoT are presented to enhance the under- concept and features of IoT should be investigated first. IoT
standing of the state of the art IoT development. To investigate the can connect ubiquitous devices and facilities with various
fog/edge computing-based IoT, this paper also investigate the rela-
tionship between IoT and fog/edge computing, and discuss issues networks to provide efficient and secure services for all appli-
in fog/edge computing-based IoT. Finally, several applications, cations anytime and anywhere [9], [80]. Based on the afore-
including the smart grid, smart transportation, and smart cities, mentioned definition, two features are required in IoT. First,
are presented to demonstrate how fog/edge computing-based IoT IoT is the extension of the net or Internet [10], meaning that,
to be implemented in real-world applications. in IoT, various networks should coexist, and the interoperabil-
Index Terms—Applications, enabling technologies, fog/edge ity among these networks is critical for information delivery
computing, Internet of Things (IoT), security and privacy. and supporting applications [7], [87]. Interconnection is a crit-
ical architecture issue in IoT [131]. Second, things connected
I. I NTRODUCTION in IoT are no longer limited to devices or objects, but can
OG/EDGE computing is an architecture organized by the
F networking edge devices or clients to provide computing
also be information, human behaviors, etc. [119], [123]. Thus,
IoT should include mechanisms that handle the connection of
objects in a broader manner.
Manuscript received September 27, 2016; revised February 13, 2017; There have been a number of research efforts devoted to
accepted February 28, 2017. Date of publication March 15, 2017; date of cur-
rent version October 9, 2017. This work was supported in part by the National developing IoT prototypical systems [131], [136]. Nonetheless,
Science Foundation (NSF) under Grant CNS 1350145 and in part by the USM most of the systems that focus on specific applications
Wilson H. Elkins Professorship fund. Any opinions, findings and conclusions are implemented within extranet or intranet, and have no
or recommendations expressed in this material are those of the authors and
do not necessarily reflect the views of the agencies. (Corresponding author: interaction with each other. Based on the features of IoT that
Wei Yu.) interconnection is a critical architecture issue, strictly speak-
J. Lin and X. Yang are with the Department of Computer Science ing, these systems or applications are not “IoT,” but the “Net
and Technology, Xi’an Jiaotong University, Xi’an 710049, China (e-mail:
[email protected]; [email protected]). of Things,” or can even be considered as “Net of Devices,”
W. Yu is with the Department of Computer and Information Sciences, and the interactions between these extranets and intranets
Towson University, Towson, MD 21252 USA (e-mail: [email protected]). were missed [123], [131]. Thus, IoT should cover all things
N. Zhang is with Department of Computer Science, George Washington
University, Washington, DC 20052 USA (e-mail: [email protected]). in large-scale networks, in which various networks should
H. Zhang is with the Department of Computer Science and Technology, coexist, and are able to interact with each other via various
Qingdao University, Qingao 266061, China (e-mail: [email protected]). gateways and middlewares, supported by the complex control
W. Zhao is with the Department of Computer and Information Science,
University of Macau, Macau, China (e-mail: [email protected]). plane [87]. One vision is that a generalized network infras-
Digital Object Identifier 10.1109/JIOT.2017.2683200 tructure that integrates various networks should be designed,
2327-4662 c 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1126 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

and all IoT-based systems or applications can provide their ser- understand the concept and features of IoT. Then, to provide a
vices by efficiently sharing network resources and information better understanding of the state of the art in IoT development,
across the generalized network infrastructure. For example, in the architectures, enabling technologies, and challenges in IoT
smart cities [14], [155], if a generalized network infrastructure are clearly presented. We consider IoT as multilayer archi-
can be implemented and is able to cover all regions in a city, tectures, divided into the perception layer, networking layer,
applications (smart grid, smart transportation, smart health- service layer, and application layer. Based on the multilayer
care, etc.) can share their individual network infrastructures to architecture, enabling technologies and open issues in each
enable data collection and information delivery. In this vision, layer are then presented. After that, security vulnerabilities
everything that is interconnected in the network can be realized and challenges are discussed, and the security issues with
because all applications can interact with each other easily and respect to confidentiality, integrity, availability, as well as pri-
share the resources effectively. The implementation of gener- vacy issues in IoT are discussed. In addition, the integration of
alized network infrastructure can reduce the cost of network IoT and fog/edge computing and related issues are presented
deployment as well [155]. to enable the design and deployment of fog/edge computing-
To have a generalized network infrastructure, the develop- based IoT. Finally, several applications (smart grid, smart
ment of IoT with respect to architectures, enabling technolo- transportation, and smart cities) are presented to illustrate
gies, and possible challenges should be studied first. In recent how fog/edge computing-based IoT are to be implemented in
years, several published survey papers reviewed the IoT tech- real-world IoT-based systems.
nologies from different aspects. For example, the survey work This paper is organized as follows. We introduce rela-
done by Atzori et al. [10] presented the enabling communica- tion between CPS and IoT in Section II. We present the
tion technologies and different visions of IoT, which can help architectures of IoT in Section III. We present the enabling
those who want to approach this field have a primary under- technologies and challenges of IoT in Section IV. We present
standing of IoT. The survey work done by Al-Fuqaha et al. [7] the security and privacy issues of IoT in Section V. The
presented the enabling technologies, protocols, and possible integration of IoT and fog/edge computing is presented in
applications of IoT, in which the horizontal overview of IoT Section VI. Finally, we conclude this paper in Section VIII.
was provided and the key IoT challenges were presented to
point out the future directions. There have been also a num- II. CPSs AND IoT
ber of research efforts devoted to security and privacy issues in
In this section, the relation between CPS and IoT is clarified.
IoT. For instance, the survey work done by Andrea et al. [8]
In the following, we first give the overview of CPS and then
presented the security vulnerabilities and challenges in IoT
discuss the key differences between CPS and IoT.
from the view of applications, networks, and physical systems,
and considered the security and privacy issues in technolo-
gies associated with physical systems, networking, software, A. Overview of CPS
and encryption. The survey work done by Sha et al. [115] Generally speaking, CPS is referred to as the system that
presented challenges issues and opportunities in IoT. In addi- can efficiently integrate both cyber and physical components
tion to the aforementioned survey papers, Botta et al. [18] through the integration of the modern computing and com-
considered the integration of cloud computing and IoT. Also, munication technologies [5], [130], aiming to changing the
Wu and Zhao [131] proposed a novel IoT infrastructure, method of interaction among the human, cyber and physical
namely WInternet, which can be designed and realized by cur- worlds. CPS emphasizes the interactions between cyber and
rent Internet technologies, and meets various requirements of physical components and has a goal of making the monitor-
IoT. Although a number of efforts have been conducted, most ing and control of physical components secure, efficient, and
existing surveys have only focused on specific aspects of IoT. intelligent by leveraging cyber components [23].
This calls for a comprehensive survey of IoT to help newcom- In CPS, “cyber” means using the modern sensing, comput-
ers have a general understanding of the complex discipline of ing, and communication technologies to effectively monitor
this emergent research area. and control the physical components, while “physical” means
To full the gap, this paper first reviews the existing efforts the physical components in real world, and “system” reflects
on IoT and then present the integration of fog/edge computing the complexity and diversity. Based on the clarification, we
and IoT and related issues. In particular, this paper conduct can see that a CPS consists of multiple heterogeneous dis-
a comprehensive overview of IoT with respect to architec- tributed subsystems [50]. Similar to the development of IoT,
tures, enabling technologies, security and privacy issues, and CPS has been developed in numerous areas [50], [72], [73],
present the foundation of fog/edge computing-based IoT and including smart grid, smart transportation, etc.
applications. Meanwhile, possible open issues and challenges As shown in [23], the CPS is the integration of physi-
in IoT are presented as well. Particularly, the relation between cal components, sensors, actuators, communication networks,
cyber-physical systems (CPSs) and IoT is explored first. Notice and control centers, in which sensors are deployed to mea-
that both CPS and IoT emphasize the interactions between sure and monitor the status of physical components, actuators
the cyber world and the physical world, and are easily con- are deployed to ensure the desirable operations on physi-
fused with one another. In addition, the difference between cal components, and communication networks are used to
CPS and IoT has not been clearly distinguished before. The deliver measured data and feedback comments among sen-
detailed relation between CPS and IoT can help newcomers to sors, actuators, and control centers. The control centers are

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1127

various networks is not only limited to physical connections.

Control plane (interfaces, middleware, protocols, etc.) should
be designed to ensure that data can be efficiently delivered
across different kinds of networks and shared. For instance,
in a smart city, networks of smart weather forecasting, smart
transportation, and smart grid should be interconnected and
interact with each other. Data from smart transportation and
smart weather forecasting should be processed and extracted
and used by the smart grid to determine the states and bright-
ness of street-lamps to ensure efficient use of energy resources,
Fig. 1. Integration of IoT and CPS. as well as traffic safety at night.
Actually, control plane in IoT is more complex than that
used to analyze measured data and sent feedback com- in Internet and has been ignored by most if not all. Recently,
mands to actuators, ensuring the system operate in desired some efforts have been focused on the control plane in IoT.
states [23], [143]. For example, Wu and Zhao [131] proposed an IoT architecture,
namely WInternet, which focuses on interconnecting various
B. Difference Between CPS and IoT Net of Things into a large-scale global network. In WInternet,
Based on the overview of CPS, we know that both CPS and the internal architecture of nodes was innovated with embed-
IoT aim to achieve the interaction between cyber world and ded computing capability to ensure that critical applications
physical world [99]. Particularly, CPS and IoT can measure can interaction with physical space. Also, protocols (netlet
the state information of physical components via smart sensor computation and pipe communication protocol) were designed
devices without human’s input. Meanwhile, in both CPS and to meet requirements of IoT applications.
IoT, the measured state information can be transmitted and To summarize, the basic difference between CPS and IoT
shared through wired or wireless communication networks. is that, CPS is considered as a system, while IoT is consid-
After the analysis of measured state information, both CPS ered as “Internet.” The common requirements for both CPS
and IoT can provide secure, efficient, and intelligent services and IoT are real-time, reliable, and secure data transmission.
to applications. The existing efforts on CPS applications and The distinct requirements for CPS and IoT are listed as fol-
IoT applications have been expanded to similar areas (smart lows: for CPS, effective, reliable, accurate, real-time control
grid, smart transportation, smart city, etc.). is primary goal, while for IoT, resource sharing and manage-
Because of the similarities between CPS and IoT, it is an ment, data sharing and management, interface among different
urgent need to clarify the difference between CPS and IoT nets, massive-scale data and big data collection and storage,
so that newcomers may enter this complex discipline easily. data mining, data aggregation and information extraction, and
Nonetheless, few existing efforts clearly identify the difference high quality of network quality of service (QoS) are important
between CPS and IoT, and several efforts have even considered services.
the CPS and IoT to be the same concept. Thus, to fulfil this In fact, one of the most representative applications that inte-
gap, the difference of CPS and IoT is clarified below. grate CPS and IoT is smart cities, in which several CPS appli-
As mentioned above, the essence of CPS is the system and cations operate simultaneously, including smart gird, smart
the main objective of CPS is to measure the state information transportation, smart healthcare, etc. As shown in Fig. 1, the
of physical devices and ensure the secure, efficient, and intelli- communication layers of all applications are interconnected as
gent operation on physical devices. In CPS, the sensor/actuator a unified network to provide service for smart cities.
layer, communication layer, and application (control) layer are
present. The sensor/actuator layer is used to collect real-time III. A RCHITECTURE
data and execute commands, communication layer is used to
In this section, we show several existing architectures
deliver data to upper layer and commands to lower layer, and
for IoT.
application (control) layer is used to analyze data and make
decisions. Fig. 1 illustrates the three layers in CPS. From this
figure, we can see that CPS is a vertical architecture. A. Three-Layer Architecture
In contrast, IoT is a networking infrastructure to connect a Typically, the architecture of IoT is divided into three
massive number of devices and to monitor and control devices basic layers [83]: 1) application layer; 2) network layer; and
by using modern technologies in cyber space. Thus, the key 3) perception layer, which are further described below.
of IoT is “interconnection.” The main objective of IoT is 1) Perception Layer: It is also known as the sensor layer,
to interconnect various networks so that the data collection, is implemented as the bottom layer in IoT architecture [11].
resource sharing, analysis, and management can be carried The perception layer interacts with physical devices and com-
out across heterogeneous networks. By doing so, reliable, effi- ponents through smart devices (RFID, sensors, actuators, etc.).
cient, and secure services can be provided. Thus, IoT is a Its main objectives are to connect things into IoT network, and
horizontal architecture, which should integrate communica- to measure, collect, and process the state information associ-
tion layers of all CPS applications to achieve interconnection, ated with these things via deployed smart devices, transmitting
as shown in Fig. 1. Notice that, the interconnection of the processed information into upper layer via layer interfaces.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1128 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

2) Network Layer: It is also known as the transmission that, in some existing efforts, the service layer is divided into
layer, is implemented as the middle layer in IoT architec- two sublayers, namely service composition sublayer and ser-
ture [68]. The network layer is used to receive the processed vice management sublayer. In addition, the business layer is
information provided by perception layer and determine the extracted from the application layer and works as the upper
routes to transmit the data and information to the IoT hub, layer of the application layer to provide complex service
devices, and applications via integrated networks. The network requests.
layer is the most important layer in IoT architecture, because In the four-layer SoA-based IoT architecture, the percep-
various devices (hub, switching, gateway, cloud comput- tion layer is performed as the bottom layer of the architecture,
ing perform, etc.), and various communication technologies and used to measure, collect, and extract the data associ-
(Bluetooth, Wi-Fi, long-term evolution, etc.) are integrated ated with physical devices [52]. The network layer is used
in this layer. The network layer should transmit data to or to determine routes and provide data transmission support
from different things or applications, through interfaces or via integrated heterogeneous networks [10], [47]. The service
gateways among heterogeneous networks, and using various layer is located between network layer and application layer,
communication technologies and protocols. providing services to support the application layer [10]. The
3) Application Layer: It is also known as the business layer, service layer consists of service discovery, service composi-
is implemented as the top layer in IoT architecture [7]. The tion, service management, and service interfaces. Here, service
application layer receives the data transmitted from network discovery is used to discover desired service requests, service
layer and uses the data to provide required services or oper- composition is used to interact with the connected objects,
ations. For instance, the application layer can provide the and divide or integrate services to meet service requests in
storage service to backup received data into a database, or an efficient way, service management is used to manage and
provide the analysis service to evaluate the received data for determine the trust mechanisms to meet service requests, and
predicting the future state of physical devices. A number of service interfaces are used to support interactions among all
applications exist in this layer, each having different require- provided services. The application layer is used to support
ments. Examples include smart grid, smart transportation, the service requests by users. The application layer can sup-
smart cities, etc. [124], [132]. port a number of applications, including smart grid, smart
The three-layer architecture is basic for IoT and has been transportation, smart cities, etc.
designed and realized in a number of systems [132]. Yet,
despite the simplicity of the multilayer architecture of IoT,
functions and operations in the network and application lay- IV. E NABLING T ECHNOLOGIES AND
ers are diverse and complex. For example, the network layer C HALLENGES IN D IFFERENT L AYERS
not only needs to determine routes and transmit data, but Based on the architectures mentioned above, IoT can be
also provide data services (data aggregation, computing, etc.). realized with several enabling technologies. In this section, the
The application layer not only needs to provide services to four-layer SoA-based IoT architecture is taken as an example
customers and devices, it must also provide data services to present the relevant enabling technologies and challenges
(data mining, data analytics, etc.). Thus, to establish a generic in each layer.
and flexible multilayer architecture for IoT, a service layer
should be developed between network layer and application
layer to provide the data services in IoT. Based on this con- A. Perception Layer
cept, service-oriented architectures (SoAs) have recently been In the perception layer, the main function is to identify
developed to support IoT [7], [136]. and track objects. To achieve this function, the following
technologies can be implemented.
1) RFID: Generally speaking, RFID, as a noncontact com-
B. SoA-Based Architecture munication technology, is used to identify and track objects
Generally speaking, SoA is a component-based model, without contact. It support data exchange via radio signals
which can be designed to connect different functional units over a short distance [8], [162]. The RFID-based system con-
(also known as services) of an application via interfaces and sists of RFID tag, RFID reader, and antenna [62]. RFID
protocols [10], [86], [135]. SoA can focus on designing the tag can be a microchip attached to an antenna. Each RFID
workflow of coordinated services, and enable the reuse of soft- tag is attached in an object and has its unique identification
ware and hardware components, improving the feasibility of number. An RFID reader can identify an object and obtain
SoA for use in designing IoT architecture [10], [136]. Thus, the corresponding information by querying to the attached
SoA can be easily integrated into IoT architecture, in which RFID tag through appropriate signals [64]. An antenna is used
data services provided by the network layer and the appli- to transmit signals between RFID tag and RFID reader. In
cation layer in the traditional three-layer architecture can be comparison with other technologies, RFID has the following
extracted and form a new layer, namely the service layer (also benefits [51], [123] (fast scanning, durability, reusability, large
known as the interface layer or middleware layer). Thus, in an storage, noncontact reading, security, small size, low cost,
SoA-based IoT architecture, four layers exist and interact with etc.). Because of these benefits, RFID can be useful in the per-
each other [122], these being the perception layer, network ception layer of IoT to identify and track objects and exchange
layer, service layer, and application layer. It is worth noting information.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1129

2) Wireless Sensor Networks: Wireless sensor network 2) 6LoWPAN: Low-power WPANs (LoWPANs) are orga-
(WSN) can play a very important role in IoT [34], [66], [85], nized by a large number of low-cost devices connected via
[94], [133], [150]. WSN can monitor and track the status of wireless communications [123]. In comparison with other
devices, and transmit the status data to the control center or types of networks, LoWPAN has a number of advantages
sink nodes via multiple hops [6], [69]. Thus, WSN can be (small packet sizes, low power, low bandwidth, etc.) [123]. As
considered as the further bridge between the real world and an enhancement, 6LoWPAN protocol was designed by com-
the cyber world [130]. In comparison with other technolo- bining IPv6 and LoWPAN. In 6LoWPAN, IPv6 packets can
gies, WSN has a number of benefits, including scalability, be transmitted over IEEE 802.15.4 networks [98]. Because
dynamic reconfiguration, reliability, small size, low cost, and of the low cost and low energy consumption, 6LoWPAN is
low energy consumption. All these benefits help WSN to be suitable to IoT, in which a large number of low cost devices
integrated in various areas with diverse requirements. are included. 6LoWPAN have several advantages, including a
Notice that both RFID and WSN can be used for data acqui- great connectivity and compatibility with legacy architectures,
sition in IoT, and the difference is that RFID is mainly used low-energy consumption, ad-hoc self-organization, etc.
for object identification, while WSN is mainly used for the 3) ZigBee: ZigBee is a wireless network technology,
perception of real-world physical parameters associated with designed for short-term communication with low-energy con-
the surrounding environment [123]. sumption [99]. In ZigBee protocol, five layers are included:
3) Others: Barcode, also denoted 1-D code, stores the the physical layer, the MAC layer, the transmission layer, the
information in several black lines and white spacings. These network layer, and the application layer [123]. The advantages
lines and spacings have different widths, organized in a lin- of ZigBee networks include low energy consumption, low cost,
ear or 1-D direction, and are arranged with special encoding low data rate, low complexity, reliability, and security. ZigBee
rules [49]. The information included in the barcode can be network can support multiple topologies, including star, tree,
read by a machine that scans the barcode with an infrared and mesh topologies [13].
beam [93]. 4) Z-Wave: Z-wave is a short-term wireless communica-
A 2-D code records the information by using black and tion technology with the advantages of low cost, low energy
white pixels laid out on the plane, in which black pixel rep- consumption, and great reliability [99]. The main objective of
resents a binary of “1” and white pixel represents a binary Z-wave is to provide reliable transmission between a control
of “0” [49]. With special encoding rules, the black and unit and one or more end-devices, and Z-wave is suitable for
white pixels can store a significant amount of information. the network with low bandwidth. Notice that no more than
In comparison with barcode, 2-D code has the benefit of 232 nodes (slaves) can be included in a Z-wave network, and
high information content, high reliability, high robustness, all nodes (slaves) would be controlled by the controller and
etc. [123]. have routing capability [99], [123]. Z-wave network supports
In addition, RFID sensor network (RSN) is an integration of the dynamic routing technology, and each slave stores a route
RFID system and sensor network. In an RSN, sensor network list in its memory, which is updated by the controller [41].
can cooperate with RFID system to identify and track the sta- Although both of ZigBee and Z-wave support the short-
tus of objects [138]. In an RSN, small RFID-based sensing range wireless communication with low cost and low energy
devices and RFID reader are implemented, where the RFID consumption, there are some differences between them. The
reader works as a sink node to generate data and provides main difference between ZigBee and Z-wave is the frequency
power for network operations. band operated in by the physical layer. In ZigBee, the
frequency band of the physical layer is normally 2.4 GHz,
B. Network Layer while the frequency band in Z-wave is less than 1 GHz (908.42
The network layer is used to determine routing, and provide ∼ 868.42 MHz) [123]. The ZigBee network can support end-
data transmission support through integrated heterogeneous devices (slaves) up to 65 000, while the Z-wave network can
networks. In the following, some protocols that can enable only support 232 end-devices (slaves) [123]. In comparison
the reliable and secure communication in IoT are presented. with ZigBee architecture, Z-wave is simple in implementation.
1) IEEE 802.15.4: IEEE 802.15.4 is a protocol designed for 5) Message Queue Telemetry Transport: Using the pub-
the physical layer and the MAC layer in wireless personal area lish/subscribe technique, message queue telemetry transport
networks (WPANs) [7], [37]. The goal of IEEE 802.15.4 is to (MQTT) is a messaging protocol, which is used to collect
focus on low-rate WPANs, providing the low rate connections measured data on remote sensors and transmit the data to
of all things in a personal area with low energy consumption, servers [7]. MQTT is a simple and lightweight protocol, and
low rate transmission, and low cost [4]. IEEE 802.15.4 proto- supports the network with low bandwidth and high latency.
col stack is based on open system interconnection model, in MQTT can be implemented in various platforms to connect
which each layer only implements parts of transmission func- things in IoT into the Internet, and thus MQTT can be used as
tions, and lower layers can provide service to upper layers. a messaging protocol between sensors/actuators and servers,
IEEE 802.15.4 can support bands of 868/915M and 2.4 GHz, making MQTT play an important role in IoT.
and the data transmission rate on these bands can achieve 20, 6) Constrained Application Protocol: Constrained applica-
40, and 250 Kb/s, respectively, [7]. IEEE 802.15.4 is a basis tion protocol (CoAP) is a messaging protocol based on repre-
for many wireless communication technologies and protocols, sentational state transfer (REST) architecture [7], [17], [38].
such as ZigBee [63], WirelessHART [59], etc. Because most of devices in IoT are resources constrained

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1130 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

(i.e., small storage and low computing capability), HTTP can- deployed at resource-constrained nodes to determine routes
not be used in IoT, due to its complexity. To overcome the over low power and lossy links [7], [128], [151]. Although
issue, CoAP was proposed to modify some HTTP functions these protocols can be integrated into IoT, enhanced protocols
to meet the requirements for IoT. Generally speaking, CoAP with more security, reliability, and interoperability capabilities
is the application layer protocol in the 6LoWPAN protocol are required to advance the development of IoT.
stack, and aims to enable resources constrained devices to
achieve RESTful interactions. The group communication and
push notification are supported by CoAP, but broadcasting C. Service Layer
is not. Resource observation, block-wise resource transport, As described above, the service layer is located between
resource discovery, interaction with HTTP, and security are the network layer and the application layer, and provides effi-
all the important features provided by CoAP [7], [38]. cient and secure services to objects or applications. In the
7) Extensible Messaging and Presence Protocol: service layer, the following enabling technologies should be
Extensible messaging and presence protocol (XMPP) is an included to ensure that the service can be provided efficiently:
instant messaging protocol based on XML streaming proto- interface technology, service management technology, mid-
cols [7], [111]. XMPP inherits features of XML protocol, so dleware technology, and resource management and sharing
that XMPP has great scalability, addressing, and security capa- technology.
bilities, and can be used for multiparty chatting, voice and 1) Interface: The interface technology must be designed
video streaming, and tele-presence. In XMPP, the following in the service layer to ensure the efficient and secure infor-
three roles are included: 1) client; 2) server; and 3) gateway, as mation exchange for communications among devices and
well as bidirectional communication is supported between two applications. In addition, the interface should efficiently man-
parties of these three roles. Particularly, the server can achieve age the interconnected devices, including device connection,
the functionality of link management and message routing, the device disconnection, device communication, and device oper-
gateway is used to support the stable communication among ation [136].
heterogeneous systems, and the client can be connected to the To support applications in IoT, an interface profile (IFP)
server based on TCP/IP protocol and transmit context based on can be considered as a service standard, which can be used to
XML streaming protocol. Thus, XMPP can be used in IoT to facilitate the interactions among services provided by various
support the object to object communication with XML-based devices or applications. To achieve an efficient IFP, univer-
text messages. sal plug and play should be implemented [36], [45], [136].
8) Data Distribution Service: Data distribution service As the development of IoT, a number of efforts on the
(DDS) is a publish/subscribe protocol for supporting high interface have been performed. For instance, SOCRADES
performance device-to-device communication [7], [44]. DDS integration architecture can be used to provide effective inter-
was developed by object-manage-group [44] and is a data actions between applications and services [45], [107]. As the
centric protocol, in which multicasting can be supported to development of SoA-IoT, service provisioning process has the
achieve great QoS and high reliability. The broker-less pub- functionality of providing interactions with applications and
lish/subscribe architecture makes DDS suitable to real-time services [136], [166]. Although a number of interface tech-
constrained IoT and device-to-device communications [7]. In nologies have been developed for IoT, implementing more
addition, DDS can achieve great scalability. effective, secure, and scalable interface technologies with
9) Advanced Message Queuing Protocol: Advanced mes- low cost remains a great challenge in future research to
sage queuing protocol (AMQP) is an open standard support IoT.
message queuing protocol used to provide message service 2) Service Management: Service management can effec-
(queuing, routing, security, reliability, etc.) in the applica- tively discover the devices and applications, and schedule effi-
tion layer [7], [40]. AMQP focuses on the message-oriented cient and reliable services to meet requests. A service can be
environments and can be considered as a message-oriented considered as a behavior, including collection, exchanging, and
middleware protocol. Using AMQP, clients can achieve sta- storage of data, or an association of these behaviors to achieve
ble communication with message middlewares, even if these a special objective [10], [86]. In IoT, some requirements can
clients and middlewares are produced by different programming be met by only one service, while other requirements have
languages. In addition, AMQP implements various kinds of to be met by the integration of multiple services. Thus, the
message exchange architectures, including store and forward, service can be divided into two categories in IoT: 1) primary
publish and subscribe, message distribution, message queuing, service and 2) secondary service [136]. The primary service,
context-based routing, and point-to-point routing [113]. also known as the basic service, can expose the primary func-
10) Others: In addition to the transmission protocols, tionalities at devices or applications. In contrast, the secondary
communication protocols, and messaging protocols, other pro- service can achieve the auxiliary functionalities based on the
tocols can play important roles in IoT as well. For example, primary service or other secondary service.
multicast DNS (mDNS) can support the name resolution in IoT To hide the implementation detail of services and make
applications [7], [55]. DNS service discovery can be used by these services be compatibly implemented in heterogeneous
clients to discover desired services in a special network via devices and applications, SoA has been used to integrate ser-
mDNS [7], [30]. Routing protocol for low power and lossy vices. Through this, the reliability and consistence of services
networks is a link-independent routing protocol, which can be can be provided [78], [136]. For example, OSGi platform

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1131

established by a dynamic SoA architecture is an effective mod- acquisition are typical examples. Pervasive middleware is
ular platform to deploy services. To deploy an SoA-based designed for the pervasive computing environment, and pro-
service, the service composition platforms should be devel- vides services on multiple and heterogeneous platforms [92].
oped first, and then the functionalities and communication To integrate middleware into IoT, the following challenges
capabilities of devices should be abstracted. Finally, a com- need to be addressed [25].
mon set of services should be provisioned [10], [136]. In 1) Interoperability challenge is to connect heterogeneous
SoA-based service, each service offered by a device or appli- devices in communication and information exchange.
cation can be considered as a standard service, which can be 2) Scalability challenge is to be effectively operated in
effectively and easily used in various heterogeneous devices either small-scale environment or large-scale environ-
and applications without any change. In this way, require- ment that could involve a massive number of objects.
ments in SoA-based IoT can be satisfied more quick and 3) Abstraction provision challenge is to provide abstrac-
efficient [136]. tions at various levels.
3) Middleware: Middleware is a software or service pro- 4) Spontaneous interaction challenge is to provide the
gramming that can provide an abstraction interposed between reliable service for spontaneous events.
IoT technologies and applications [46], [153]. In middleware, 5) Infixed infrastructure challenge is to provide reliable
the details of different technologies are hidden, and the stan- services without requesting a fixed infrastructure.
dard interfaces are provided to enable developers to focus on 6) Multiplicity challenge is to support simultaneously com-
the development of applications without considering the com- munication among devices and to select or schedule the
patibility between applications and infrastructures [10]. Thus, most suitable services for devices from a massive set of
by using middleware, devices and applications with different services.
interfaces can exchange information and share resources with The middleware for IoT should achieve trust, security, and
each other. privacy.
Middleware has the following benefits [10]: 1) middle- 4) Resource Management and Sharing: Various heteroge-
ware can support various applications; 2) middleware can run neous networks are integrated to provide data delivery for
on various operating systems and platforms; 3) middleware all applications in IoT (smart transportation, smart grid, etc.).
can support the distributed computing and the interaction of To reduce the cost, some applications can share part of the
services among heterogeneous networks, devices, and appli- network resources to increase its utilization. In this case,
cations; 4) middleware can support standard protocols; and ensuring that information requested by various applications
5) middleware can provide standard interfaces, providing is delivered on time is a challenging issue in IoT. Existing
portability and standard protocols to enable interoperability, resource sharing mechanisms primarily focus on the spec-
and making middleware play an important role in standard- trum sharing, which is used to efficiently coordinate multiple
ization [25]. Middleware can also provide a stable high-level networks in the same frequency to maximize the utilization
interface for applications. With stable interfaces, applications of network resources [77], [126], [164]. The spectrum shar-
can work independently on hardware and operating system. ing can be divided into three dimensions, including time,
This feature makes middleware suitable for IoT, because a frequency, and space. While most of the existing schemes
large number of heterogeneous devices and networks are inte- were developed for machine-to-machine or device-to-device
grated, and these devices and networks would be changed or communications, IoT focuses on thing-to-thing networks, in
updated often. which “thing” not only refers to devices or machines, but also
A number of research efforts on middleware have been refers to human behaviors, and other objects. Thus, design-
developed, and can be divided into five categories [25], [102], ing an effective resource sharing scheme across heterogeneous
including: 1) message-oriented middleware; 2) semantic Web- networks that is suitable for IoT environment is a significant
based middleware; 3) location-based service and surveillance challenge for future development.
middleware; 4) communication middleware; and 5) perva- In addition, raw data in IoT are collected by smart devices
sive middleware. Particularly, message-oriented middleware (RFID, sensors, etc.), and most of these smart devices are
can provide the reliable information exchange among var- resource-constrained and cannot harvest energy from environ-
ious platforms, and communication protocols (e.g., AMQP, ment. Thus, an energy saving scheme should be considered
DDS, MQTT, and XMPP) [7], [25]. Semantic Web-based in resource management [108]. There have been a number
middleware can provide the interactions and interoper- of efforts on energy conservation and energy management
ability among various sensor networks. Examples of this in sensor networks, including schemes to enhance the life
category includes the SoA-based middleware [118], task of sensors via harvesting energy from distributed energy
computing-based middleware [43], etc. Location-based ser- resources [21], schemes to reduce the energy of sensors via
vice and surveillance middleware integrates the locations of duty-cycle scheme [97], energy-based routing protocols to bal-
devices and other information to provide integrated value ser- ance the energy consumption and to increase the life of the
vices [109]. Communication middleware can provide reliable sensor network [150], [151], etc. Although these efforts can
communications among heterogeneous devices and applica- work well on energy saving and management, a scheme that is
tions. In communication middleware, RFID-based middle- suitable for IoT network infrastructures comprised of heteroge-
ware (Fosstrak [2], etc.), sensor network-based middleware neous networks is an unresolved challenge for future research
(TinyREST [81], etc.) and the supervisory control and data as well.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1132 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

V. S ECURITY AND P RIVACY services, and people sharing the same communication network
In this section, the security features of IoT are presented in IoT.
first. Then, the security and privacy issues, and possible 6) Trust: Trust can ensure the aforementioned security
solutions are discussed in detail. and privacy objectives to be achieved during the interactions
among different objects, different IoT layers, and different
applications. The objectives of trust in IoT can be divided as
A. Security Features of IoT trust between each IoT layer, trust between devices, and trust
between devices and applications [8]. With trust, security, and
1) Confidentiality: Confidentiality can ensure that the data
privacy can be enforced. Trust management systems should be
is only available to authorized users throughout the process,
designed to implement these trust objectives in IoT.
and cannot be eavesdropped or interfered by nonauthorized
users. In IoT, confidentiality is an important security princi-
ple, because a large number of measurement devices (RFID, B. Security
sensors, etc.) can integrated in IoT. Thus, it is critical to In this section, security challenges in each layer of IoT
ensure that the data collected by a measurement device will architecture are presented in detail. In SoA-based IoT, the ser-
not reveal secure information to its neighboring devices. To vice layer is established via extracting the functionality of data
achieve great confidentiality, enhanced techniques, including services in the network layer and the application layer. Thus,
secure key management mechanisms, and others should be security challenges in the service layer can be attributed to
developed and used [22]. challenges in the network and the application layers. In the
2) Integrity: Integrity can ensure that the data cannot be following, only security challenges in the perception layer, the
tampered by intended or un-intended interference during the network layer, and the application layer are presented.
data delivery in communication networks, ultimately providing 1) Perception Layer: As the main purpose of the perception
the accurate data for authorized users. Integrity is important layer in IoT is to collect data, the security challenges in this
for IoT, because if IoT applications receive forged data or layer focus on forging collected data and destroying perception
tampered data, erroneous operation status can be estimated devices, which are presented below.
and wrong feedback commands can be made, which could a) Node capture attacks: In a node capture attack, the
further disrupt the operation of IoT applications. To achieve adversary can capture and control the node or device in IoT
acceptable integrity, enhanced secure data integrity mecha- via physically replacing the entire node, or tampering with the
nisms (false data filtering schemes, etc.) should be developed hardware of the node or device [162]. If a node is compro-
and applied [143]. mised by the node capture attack, the important information
3) Availability: Availability can ensure that the data and (group communication key, radio key, matching key, etc.) can
devices are available for authorized users and services when- be exposed to the adversary. The adversary can also copy the
ever the data and devices are requested. In IoT, services are important information associated with the captured node to
commonly requested in real-time fashion, and services can- a malicious node, and then fake the malicious node as an
not be scheduled and provided if the requested data cannot authorized node to connect to the IoT network or system. This
be delivered in a timely manner. Thus, availability is also an attack is denoted as the node replication attack. A node capture
important security principle. One of the most serious threats to attack can incur a serious impact on the network. To defend
availability is the denial-of-service (DoS) attack, and enhanced against the node capture attack, effective schemes to monitor
techniques (secure and efficient routing protocols, etc.) should and detect malicious nodes need to be studied [15].
be studied and applied to ensure availability in IoT [82]. b) Malicious code injection attacks: In addition to the
4) Identification and Authentication: Identification can node capture attack, the adversary can control a node or a
ensure that nonauthorized devices or applications cannot be device in IoT by injecting malicious code into the memory of
connected to IoT, and authentication can ensure that the data the node or device, which is denoted as the malicious code
delivered in networks are legitimate, and the devices or appli- injection attack [142]. The injected malicious code not only
cations that request the data are legitimate as well. In IoT, can perform specific functions, but can also grant the adver-
identifying and authenticating each data and object is diffi- sary access into the IoT system, and even gain the full control
cult, because a large number of diverse objects comprise an of the IoT system. To defend against the malicious code injec-
IoT. Thus, designing efficient mechanisms to deal with the tion attack, effective code authentication schemes need to be
authentication of objects or things is critical in IoT [32]. designed and integrated into IoT [114], [142].
5) Privacy: Privacy can ensure that the data can only be c) False data injection attacks: With the captured node
controlled by the corresponding user, and that no other user or device in IoT, the adversary can inject false data in place
can access or process the data. Unlike confidentiality, which of normal data measured by the captured node or device,
aims to encrypt the data without being eavesdropped and inter- and transmit the false data to IoT applications [143]. After
fered by nonauthorized users, privacy ensures that the user receiving the false data, IoT applications can return erroneous
can only have some specific controls based on received data feedback commands or provide wrong services, which fur-
and cannot infer other valuable information from the received ther affects the effectiveness of IoT applications and networks.
data [20], [106], [144], [159]. Privacy is considered as one of To defend against such a malicious attack, techniques (false
important security principles due to a large number of devices, data filtering schemes, etc.), which can efficiently detect and

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1133

drop the false data before the data is received by the IoT a) DoS attacks: DoS attacks can consume all of the
applications, need to be designed [71], [72]. available resources in IoT by attacking network protocols or
d) Replay attacks (or freshness attacks): In IoT, the bombarding the IoT network with massive traffic, rendering
adversary can use a malicious node or device to transmit the services of IoT systems unavailable [83]. The DoS attack
to the destination host with legitimate identification infor- is considered to be one of the most common attacks, and rep-
mation, which has been received by the destination host, in resents an attack category, which can result in the services of
order to make the malicious node or device obtain the trust IoT systems being unavailable. Thus, DoS attacks can be gen-
of IoT [89], [162]. Replay attack is commonly launched in erated by attack schemes, including Ping of Death, TearDrop,
authentication process to destroy the validity of certification. UDP flood, SYN flood, Land Attack, etc. To defend against
To mitigate the replay attack, techniques (secure time stamp DoS attacks, attacking schemes need to be carefully investi-
schemes, etc.) should be designed and developed in IoT [31]. gated first, and then the efficient defensive schemes to mitigate
e) Cryptanalysis attacks and side channel attacks: A attacks need be developed to secure IoT systems [82].
cryptanalysis attack can use the obtained ciphertext or plain- b) Spoofing attacks: The purpose of spoofing attacks is
text to infer the encryption key being used in the encryption for the adversary to gain full access to the IoT system, and
algorithm [157]. Nonetheless, the efficiency of cryptanalysis send malicious data into the system [8]. In IoT, examples of
attack is low. To improve the efficiency, new attacks, namely spoofing attacks include IP spoofing [91], RFID spoofing [88],
the side channel attacks, can be introduced by the adver- etc. In an IP spoofing attack, the adversary can spoof and
sary. For example, in the side channel attack investigated in record the valid IP address of other authorized devices in the
IoT [137], the adversary could deploy some techniques on IoT, and then access the IoT system to send malicious data
the encryption devices in IoT to obtain the encryption key, with the obtained valid IP address, making malicious data
which is used in IoT for encrypting data and decrypting data. appear to be valid. In an RFID spoofing attack, the adver-
One of the typical side channel attacks is the timing attack, sary can spoof and record the information of a valid RFID
in which the adversary can obtain the encryption key by ana- tag, and then send malicious data with this valid tag ID to
lyzing the time information required to execute the encryption the IoT system. Secure trust management, identification and
algorithm. To mitigate the side channel attack, efficient and authentication can be possible solutions to defend against the
secure encryption algorithms and key management schemes spoofing attack [28], [32].
need to be developed in IoT [22]. c) Sinkhole attacks: In a sinkhole attack, a compro-
f) Eavesdropping and interference: Because most of mised device or node claims exceptional capabilities of power,
devices in IoT will communicate via wireless networks, vul- computation, and communication, such that more neighboring
nerability lies in the fact that information delivered in wireless devices or nodes will select the compromised device or node
links can be eavesdropped by nonauthorized users [42], [163]. as the forwarding node in data routing process because of the
To deal with eavesdropping, secure encryption algorithms and appealing capabilities [117]. By doing this, the compromised
key management schemes are required. The adversary can device or node can increase the amount of data obtained before
also send noise data or signal to interfere with the infor- its delivered in the IoT system. Notice that a sinkhole attack
mation delivered in wireless links. To ensure the accuracy not only can break the confidentiality of delivered data, but
and timely delivery of data, effective secure noise filtering also can be a fundamental step to launch additional attacks
schemes are required to filter the noise data and restore original (DoS attack, etc.). To defend against the sinkhole attack, tech-
data [90]. niques such as secure multiple routing protocols need to be
g) Sleep deprivation attacks: In IoT, most devices or studied and applied [57].
nodes have low power ability. To extend the life cycle of the d) Wormhole attacks: Wormhole attack can be launched
devices and nodes, devices or nodes are programmed to follow by two cooperative malicious devices or nodes in IoT, in which
a sleep routine to reduce the power consumption [8], [112]. the two malicious devices in different locations can exchange
Nonetheless, the sleep deprivation attack can break the pro- routing information with private links to achieve a false one-
grammed sleep routines and keep device or nodes awake all hop transmission between them, even if they are located far
the time until they are shut down. To extend the life cycle away from each other [67]. In a wormhole attack, because
of these devices and nodes, the energy harvest scheme can the forwarding hops are reduced, more data will be delivered
be one possible solution, in which devices and nodes can har- through these two malicious devices or nodes. With access
vest energy from the external environment (solar, etc. [21]). In to more delivered data, the wormhole attack can lead to the
addition, other techniques, like secured duty-cycle mechanism similar damage as sinkhole attack. To defend against worm-
to mitigate the sleep deprivation attack, need to be studied hole attack, there are some possible defensive techniques.
in IoT. One technique is to modify the routing protocols to enhance
2) Network Layer: As the main purpose of the network the security in the route selection process [26], while other
layer in IoT is to transmit collected data, the security chal- techniques involve deploying secure hardware (GPS, directed
lenges in this layer focus on the impact of the availability of antenna, etc.).
network resources. Also, most devices in IoT are connected e) Man in the middle attack: In a man in the mid-
into IoT networks via wireless communication links. Thus, dle attack, a malicious device controlled by the adversary
most security challenges in this layer are related to wireless can be virtually located between two communicating devices
networks in IoT. in IoT [96]. By stealing the identify information of the

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1134 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

two normal devices, the malicious device can be a middle users via the infected e-mails and phishing websites [8], [54].
device to store and forward all data, which is communicated Secure authorization access, and identification and authentica-
between these two normal devices, while the two normal tion can mitigate phishing attacks [8]. Nonetheless, the most
devices cannot detect the existence of the malicious device, efficient way is for users themselves to always be vigilant
and instead believe that they directly communicate with each while surfing online. This becomes an issue as most of devices
other. The man in the middle attack can violate the con- in IoT are machines, which may lack of such intelligence.
fidentiality, integrity, and privacy of restricted data in IoT b) Malicious virus/worm: A malicious virus/worm is
through monitoring, eavesdropping, tampering, and controlling another challenges to IoT applications [8], [127], [154]. The
the communication between the two normal devices. Unlike adversary can infect the IoT applications with malicious self-
malicious node capture attacks that need to physically tamper propagation attacks (worms, Trojan Horse, etc.), and then
with the hardware of devices, the man in middle attack can obtain or tamper with confidential data. Reliable firewall,
be launched by only relying on the communication protocols virus detection, and other defensive mechanisms need to be
used in IoT networks. Secure communication protocols and deployed to combat malicious virus/worm attacks in IoT
key management schemes, which can ensure the identify and applications [110].
key information of normal devices not be leaked to the adver- c) Malicious scripts: Malicious scripts represent the
sary, can be efficient defense techniques to protect against the scripts that are added to software, modified in software, and
attack [22], [82]. deleted from software with the purpose of harming the system
f) Routing information attacks: Routing information functions of IoT [8]. Because all IoT applications are con-
attacks focus on the routing protocols in IoT systems, in which nected to the Internet, the adversary can easily fool the
the routing information can be manipulated and resent by the customers in running malicious scripts (java attack applets,
adversary to create route loops in the data transmission of active-x scripts, etc.) when requesting services through the
the network, leading to the extension of source paths and the Internet. Malicious scripts can pose the leakage of confidential
increase of end-to-end delay in IoT networks [8]. To defend data and even a complete system shut down. To defend against
against the routing information attack, secure routing protocols malicious scripts, effective script detection techniques, includ-
and trust management to establish secure links among devices ing honeypot techniques, static code detection, and dynamic
in IoT and ensure the identifying information and IP addresses action detection, need to be deployed in IoT systems.
not to be leaked to the adversary are possible techniques to
be used.
g) Sybil attacks: In a sybil attack, a malicious device, C. Privacy
namely a sybil device, can claim several legitimate identi- In general, all of the massive data collected and used in
ties and impersonate them in IoT systems [8], [95], [158]. IoT should go through the following three steps: 1) data col-
Because a sybil device has several legitimate identities, false lection; 2) data aggregation [129]; and 3) data mining and
data sent by the sybil device can be easily accepted by their analytics [125], [165]. Particularly, data collection is enacted
benign neighboring devices. Also, routes that select sybil to sense and collect the status data of objects in IoT, data
devices as forwarding nodes may consider that several dif- aggregation integrates an amount of related data into a com-
ferent intersected paths are determined, but, in fact only prehensive information, and data mining and analytics extract
one path is determined and all transmitted data needs to go the potential value of integrated comprehensive information
through the sybil device, in which jamming and DoS can for special applications in IoT [125]. Although data collection,
be used. To defend against sybil attacks, secure identification data aggregation, and data mining and analytics can provide
and authentication mechanisms need to be developed for IoT a number of services to our daily lives, the privacy issues of
systems [32]. the data in these steps are raised in IoT as well. Privacy, as a
h) Unauthorized access: RFID is an important enabling new challenge in IoT, can lead to property loss, and even com-
technology in IoT. Nonetheless, as a large number of RFID- promise human safety [106], [144]. For example, in the smart
based devices are integrated in IoT, and most of the RFID grid, if the adversary obtains the private data of the energy con-
tags lack proper authentication mechanisms, RFID tags can be sumption of customers, he or she can infer the time when users
accessed and the information stored in tags can be obtained, are in the home or out of home, and conduct theft or other
modified, and deleted by the adversary [8], [60]. Thus, autho- damage to users with a probability. Thus, privacy-preserving
rization access and authentication mechanisms for RFID-based mechanisms need to be developed to ensure private data not
devices in IoT is a challenge in need of further develop- to be leaked to the adversary in IoT.
ment [56]. Based on different data processing steps, privacy-preserving
3) Application Layer: The main purpose of the application mechanisms can be divided into three categories: 1) privacy
layer is to support services requested by users. Thus, chal- preservation in data collection [65]; 2) privacy preservation in
lenges in the application layer focus on the software attacks. data aggregation; and 3) privacy preservation in data mining
Here, several possible challenges in the application layer of and analytics [20], [134]. As the privacy in data collection,
IoT are presented below. data mining, and data analytics can be greatly preserved
a) Phishing attack: In phishing attacks, the adversary by various techniques (encryption, key management, etc.), a
can obtain the confidential data of users, such as identification majority of the existing efforts on privacy preservation in IoT
and passwords, by spoofing the authentication credentials of focus on data privacy in data aggregation.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1135

In data aggregation, the relevant data could be processed in

several different locations, and thus it is difficult to achieve pri-
vacy preservation through traditional encryption mechanisms.
Thus, several privacy-preserving mechanisms have been devel-
oped that focused on data aggregation, and can be divided into
the following categories: 1) anonymity-based privacy preserva-
tion [104]; 2) encryption-based privacy preservation [39]; and
3) perturbation-based privacy preservation [48], [100], [101].
Particularly, in anonymity-based privacy preservation, sev-
eral related anonymity techniques (K-anonymity, L-diversity,
Fig. 2. Fog/edge computing.
T-closeness, etc.) were used in the data aggregation process
to preserve the privacy of identification information [105]. In
addition, traffic analysis techniques could affect anonymous
communication systems [75], [76], [148]. In encryption-based As IoT is becoming the next technology revolution, it will
privacy preservation, several encryption techniques (homo- affect big data in aspects of data storage, data processing, and
morphic encryption, commitment mechanism, secret sharing, analytics. In IoT, continuous streams of data will affect the
zero-knowledge proof, etc.) were used in the data aggregation data storage capacity in various organizations. Additional data
to ensure data not to be eavesdropped by adversaries [39]. centers will be needed to deal with the load of data collected
Nonetheless, existing encryption techniques can only achieve from IoT applications. One possible solution is to move the
the confidentiality on data transmission and may not work well data to the cloud by leveraging the platform as a service. When
on privacy preservation. In perturbation-based privacy preser- an organization selects a technology for performing big data
vation, perturbation-based techniques (data customization, data processing and analytics, the nature of the IoT data needs to be
sharing, random noise injection, etc.) were used in data aggre- considered. Hadoop and Hive can be used to handle big data.
gation to perturb raw data, achieving privacy preservation [48], Nonetheless, for data collected by IoT application, NoSQL
however, the utilization of data could hinder the application document databases (the Apache CouchDB, etc.) may be suit-
of this technique in the IoT. able [33]. This is because the NoSQL document databases can
Due to the great performance by directly operating on provide high throughput and low latency. In addition, Apache
raw data, perturbation-based privacy preserving schemes Kafka is one IoT tool for intermediate message brokering It
are highly popular techniques used in IoT. Nonetheless, can be used for the real-time stream processing. The security
most of perturbation-based privacy preserving achieves great of big data will also be affected by IoT [156].
performance via reducing the utility of the data. With low
utility, data may not, or may only partially, support ser-
vices requested by IoT applications. Thus, the design of B. Fog/Edge Computing-Based IoT
privacy preserving schemes with great data utility remains Cloud computing is now a mature technology used to pro-
great challenges on data privacy preservation in IoT for future vide computing services or data storage over the Internet,
research. and most of the big IT companies (Amazon, IBM, Google,
etc.) are hosting cloud services. Cloud computing provides
VI. I NTEGRATION OF IoT AND F OG /E DGE C OMPUTING the benefits of flexibility, efficiency, and ability to store and
use data. Nonetheless, when cloud computing is used in IoT,
In this section, we present how to integrate IoT with
new challenges will appear. In many IoT/CPS applications,
fog/edge computing.
data from a massive number of things and objects spanning a
large geographical area need to be stored, processed, and ana-
A. Overview lyzed efficiently. To fulfill the gap, fog/edge computing is able
The information generated by the things requires big data to extend cloud computing to be closer to the things it sup-
to collect and process all of the information that is produced ports [120]. Instead of doing all the computation in the center
and gathered, and turn it into something that is useful. Big of the cloud, fog/edge computing can provide computing and
data requires the support of IoT because of the challenges of storage service to devices (nodes) at the edge of the network.
massive sensing and actuating data supported by IoT (smart A fog/edge computing node can be any network device with
grid, smart transportation, etc.). In addition, the data collected the capability of storage, computing, and network connectivity
in IoT applications are generally unstructured data, and need (routers, switches, video surveillance cameras, servers, etc.),
further analysis to extract useful information. The IoT and big as shown in Fig. 2. These devices can be deployed at any
data can work well with each other. One real-world example place with a network connection, and collect the data from
is United Parcel Service (UPS), which is one of the largest IoT devices associated with IoT applications. Different types
shipping companies in the world [79]. UPS deploys sensors to of IoT data can be directed to the proper place for further
collect data (which is the IoT application) and conduct the big- analysis based on performance requirements. The high priority
data analysis to reduce cost and improve delivery efficiency. data that needs to be addressed immediately can be processed
The sensors are deployed on the delivery vehicles and collect on fog/edge computing nodes, which are the closest to the IoT
the tracking the information (mileage, speed, fuel cost, etc.). devices that generate the data. The low priority data, which is

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1136 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

not delay-sensitive, can be directed to some aggregation nodes allocated for end-users i, and pi is the priority level for end-
for further processing and analysis. user i. Based on (2) and (3), a fog/edge node can allocate its
In addition to the benefits that fog/edge computing can resources to all end-device while achieving maximum overall
contribute, there are some challenges to integrate fog/edge satisfaction.
computing with IoT. One possible challenge is how to effi- In a fog/edge computing-based IoT, a number of fog/edge
ciently manage fog/edge computing infrastructure and allocate nodes are connected, if a fog/edge node does not have enough
available resources to IoT devices. At each time, a large num- resources to provide the requested services from nearby end-
ber of services can be requested by IoT devices, and each users while its neighboring nodes have spare resources, the
fog/edge service node only has limited computing and storage fog/edge node can move some local data to its neighboring
capability. In this case, all fog/edge nodes should be opti- nodes to be processed and stored data. By doing this, ser-
mally managed and allocated for IoT devices (or a set of IoT vices for its local end-users can be provided. This is related
devices in a cluster) to provide requested services efficiently. to the resource allocation among fog/edge nodes, which will
Another challenge is how to efficiently manage fog/edge com- be described below.
puting resources. While the previous challenge focuses on the 2) Resource Allocation Among Fog/Edge Nodes: As the
interface between fog/edge nodes and IoT services, this chal- distributed architecture of fog/edge computing-based IoT, all
lenge focuses on the resource management among fog/edge fog/edge nodes can be connected with each other via the
nodes. network connections and share their computing and storing
When fog/edge nodes are allocated to provide services, dif- resources to provide service for end-users. In this scenario,
ferent requirements need to be considered, including service if a fog/edge node does not have enough resource to pro-
availability, energy consumption, and even revenue. Thus, how vide local requested services, the fog/edge node can move
to optically map the fog/edge service nodes to IoT devices to some requested services with low priority level to be pro-
meet requirements of IoT applications remains a compelling cessed in its neighboring fog/edge nodes, which have spare
issue. In addition, security and privacy issues (authentication, resources. The spare resources of a fog/edge node can be
access control, intrusion detection, trust management, etc.) represented as
in fog/edge computing infrastructures that integrate with IoT 
n
f
remain also challenging [27], [29], [35], [152]. The security Rspare = Rf − rimax (4)
and privacy issues can be mitigated by countermeasure tech- i=1
nologies mentioned in Section V, and thus the challenges
where, Rf is the resource that fog/edge node f has, and rimax is
in resource allocation are discussed below. The challenges f
the maximum resource needed by end-user i. Thus, if Rspare is
on resources allocation in fog/edge computing-based IoT can
less than “0,” fog/edge node f does have enough resource and
be divided as resources allocation between end-devices and
needs assist from neighboring nodes, and the fog/edge node
fog/edge node and resources allocation among fog/edge nodes.
can be denoted as resource-poor node. Otherwise, fog/edge
1) Resource Allocation Between End-Devices and Fog/Edge
node f has spare resource to help other fog/edge nodes, and
Node: Because computing and storing resources are limited
the node can be denoted as resource-rich node in the fog/edge
in a fog/edge node, it is difficult to totally satisfy all ser-
computing infrastructure.
vices requested by end-users simultaneously. To address this
In the resource allocation among fog/edge nodes, a resource-
issue, each end-users may have a satisfaction function to assess
poor fog/edge node may not care about which resource-node
the allocated resources to provide its requested service. The
helps it to provide computing services, and a resource-rich
satisfaction function can be represented by
node does not care about data from which that it processes.

log(r + 1), 0  r < rmin The only one all fog/edge nodes care about is to achieve the
S(r) = (1)
log(rmax + 1), r  rmin minimum cost (minimum delay, etc.) in the overall fog/edge
computing infrastructure. By taking the objective of the min-
where S is the satisfaction function, r is the allocated imum delay as an example, we have
resources, and rmax is the maximum resource, which is ⎛ ⎞
required to provide the requested service. 1 
Objective. min⎝Costall = · Rspare · Costfg ⎠
fg
With this satisfaction function, the main objective of 2
Lfg ∈L
fog/edge node is to maximize the overall satisfaction of all
end-users, which can be represented as S.t. ⎧

⎪
⎪ ∀f ∈ NS ,
fi f
Rspare ≤ Rspare
⎪
⎪
Objective. max{Soverall } (2) ⎪
⎪ i∈Nf

⎪
⎨ ∀g ∈ Ng , jg g
S.t. ⎧ Rspare = Rspare
 j∈Ng (5)
⎪
⎪ Soverall = ni=1 {pi · Si (ri )} ⎪
⎪
⎨ ⎪
⎪ ∀Lfg ∈ L,
fg gf
Rspare = −Rspare
r1 + r2 + · · · + r n  R ⎪
⎪
(3) ⎪
⎩ ∀Lfg ∈ L, fg
⎪
⎪ p1 + p2 + · · · + pn = 1 Rspare ≤ Constraintsfg
⎩
r1 , r2 , . . . , rn  0
where Costall is the total cost, Costfg is the cost of delivering
where Soverall is the overall satisfaction of all end-users, R data on link Lfg between fog/edge node f and g, Nf is the set
jg
is the resource that a fog/edge node has, ri is the resource of neighboring nodes of fog/edge node f , Rspare is the data

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1137

moved from fog/edge node j to node g, and Constraintsjg is B. Smart Transportation

the constraints of link Lfg (bandwidth, etc.). Based on this Smart transportation, also known as intelligent transporta-
formalization, resource allocation among fog/edge nodes with tion systems, is another typical IoT-CPS-based application, in
the minimum cost in fog/edge computing infrastructure can be which intelligent transportation management, control system,
realized. communication networks, and computing techniques are inte-
grated to make transportation systems reliable, efficient, and
VII. A PPLICATIONS secure [73]. In the smart transportation system, a large num-
In the following, several applications, including the smart ber of smart vehicles are included and connected with each
grid, smart transportation, and smart cities, are presented other through wireless networks [58], [61]. Smart vehicles can
to demonstrate how fog/edge computing-based IoT to be efficiently perceive and share traffic data and schedule drivers’
implemented in real-world applications. travels with great efficiency, reliability, and safety. In the recent
past, smart vehicles (Google’s Self-Driving car, etc.) have been
designed and tested. Those smart vehicles can detect objects
A. Smart Grid
around them and safely manage speed during traveling without
In integrating IoT and CPS, the smart grid has been devel- the operation of drivers [3].
oped to replace traditional power grid to provide reliable and In the smart transportation system, each smart vehicle is
efficient energy service to consumers [1]. In the smart gird, deployed with a number of electronic control units (ECUs) to
distributed energy generators are introduced to improve the monitor and control subsystems in the vehicles. These ECUs
utilization of distributed energy resources and electric vehi- are organized as an internal network to share the collected
cles are introduced to improve the capability of energy storage data within the vehicle [121]. In addition, each smart vehicle is
and reduce emission of CO2 , and smart meters and bidi- deployed with communication interfaces to connect to the out-
rectional communication networks are introduced to achieve side network. With these communication interfaces, vehicles
the interactions between customers and utility providers. With can carry out vehicle-to-vehicle communication and vehicle-
these techniques, the smart grid can achieve great reliability, to-infrastructure communication [58]. In this way, all vehicles
efficiency, safety, and interactivity [71], [72]. can be connected into the smart transportation system, namely
By integrating with IoT, a large number of smart meters the vehicular network, and exchange and share massive data
can be deployed in houses and buildings connected in smart of current traffic status, and ultimately offer the most efficient
grid communication networks [74]. Smart meters can mon- and secure travels to customers. The massive collected data
itor energy generation, storage, and consumption, and can can be further stored and processed in the fog/edge computing
interact with utility providers to report energy demand infor- infrastructure, enabling efficient service to drivers and system
mation of customers and receive real-time electricity pricing operators.
for customers [71], [160]. With the aid of fog/edge computing Because all the traffic status data are shared by vehic-
infrastructure, the large amount of data collected from smart ular networks, the adversary may intrude into the system
meters can be stored and processed so that the effective oper- and control ECUs in vehicles by launching malicious attacks
ations of the smart grid can be supported. With the interaction against vehicle networks and fog/edge computing nodes in the
information, utility providers can optimize the energy dispatch fog/edge computing-based IoT infrastructure, sharing mislead-
of the grid, and customers can optimize their energy consump- ing traffic status data with other vehicles via communication
tion, resulting in the improvement of resource utilization and interfaces deployed in the compromised vehicle [12], [140].
the reduction of cost. In this case, the confidentiality, integrity, and privacy of traffic
Lastly, because a large number of smart meters are deployed status data can be compromised by the adversary, and seri-
in the smart grid, and communicate with each other via ous damage to the transportation system can be caused (the
wireless communication links and processed in fog/edge com- increase number of congested roads, increase time spent to
puting infrastructure, adversaries can easily capture these complete travels, etc.). Thus, in order to deploy an efficient
smart meters, nodes in fog/edge computing infrastructure, and and secure smart transportation system, techniques that can
obtain or modify the data collected [72], [142]. The con- support services in the aforementioned eight main categories
fidentiality and privacy of energy consumption information and related security issues need be carefully investigated in
can be available to adversaries. With the modified data, util- future research.
ity providers may incorrectly estimate the energy supply and
demand of the gird, and can feedback erroneous energy dis-
patch decisions, leading to imbalance on energy supply and C. Smart Cities
demand in the grid and even posing large-scale outages [72]. Smart cities can be considered a complex IoT paradigm,
In addition, key function components in the smart grid can which aims to manage public affairs via introducing infor-
be disrupted. Examples include state estimation [139], [141], mation and communication technology (ICT) solutions [155].
energy routing [70], [72], energy price [71], [145], [161], opti- Smart cities can use public resources in more efficient ways,
mal power flow [140], etc. Thus, efficient security mechanisms resulting in the improvement of the QoSs provided to users
that can preserve data privacy and integrity in the data collec- and the reduction of operational costs to public administra-
tion and transmission processes need to be developed for the tors [53], [155]. For instance, one practical implementation of
smart grid [146], [149]. smart cities, namely Padova Smart City, has been realized in

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1138 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

the city of Padova in Italy, which can select open data and [7] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, and
ICT solutions for public administrators as early as possible to M. Ayyash, “Internet of Things: A survey on enabling technologies,
protocols, and applications,” IEEE Commun. Surveys Tuts., vol. 17,
achieve the best use of public resources [19], [24]. no. 4, pp. 2347–2376, 4th Quart., 2015.
Smart cities, as a complex CPS/IoT application, may consist [8] I. Andrea, C. Chrysostomou, and G. Hadjichristofi, “Internet of Things:
of several subapplications or services [84], [155], includ- Security vulnerabilities and challenges,” in Proc. IEEE Symp. Comput.
Commun. (ISCC), Larnaca, Cyprus, Jul. 2015, pp. 180–187.
ing the smart gird, smart transportation, the structural health [9] A. P. Athreya and P. Tague, “Network self-organization in the Internet
of buildings, waste management, environmental monitoring, of Things,” in Proc. IEEE Int. Conf. Sens. Commun. Netw. (SECON),
smart health, smart lighting, etc. All these subapplications, Jun. 2013, pp. 25–33.
[10] L. Atzori, A. Iera, and G. Morabito, “The Internet of Things: A survey,”
or services, should be supported by a unified communication Comput. Netw., vol. 54, no. 15, pp. 2787–2805, Oct. 2010.
network infrastructure, or communication networks designed [11] L. Atzori, A. Iera, G. Morabito, and M. Nitti, “The social Internet of
for these subapplications or services should be interconnected Things (SIoT)—When social networks meet the Internet of Things:
Concept, architecture and network characterization,” Comput. Netw.,
to establish a large-scale interconnected heterogenous network vol. 56, no. 16, pp. 3594–3608, Nov. 2012.
for IoT/CPS applications, with the aim of achieving the best [12] S. Azadegan, W. Yu, H. Liu, A. Sistani, and S. Acharya, “Novel anti-
use of public resources in cities. To enable effective smart forensics approaches for smart phones,” in Proc. 45th Hawaii Int. Conf.
Syst. Sci. (HICSS), 2012, pp. 5424–5431.
cities, all enabling technologies discussed in Section IV and [13] P. Baronti et al., “Wireless sensor networks: A survey on the state of
security and privacy issues discussed in Section V should be the art and the 802.15.4 and ZigBee standards,” Comput. Commun.,
carefully investigated and integrated. In addition, the fog/edge vol. 30, no. 7, pp. 1655–1695, May 2007.
[14] J. Bélissent, “Getting clever about smart cities: New opportunities
computing-based IoT can enable efficient subapplications and require new business models,” Forrester Res., Cambridge, MA, USA,
services in smart cities. Tech. Rep., Nov. 2010. [Online]. Available: https://www.forrester.com/
report/Getting+Clever+About+Smart+Cities+New+Opportunities+
Require+New+Business+Models/-/E-RES56701?aid=AST127312
VIII. C ONCLUSION [15] M. V. Bharathi, R. C. Tanguturi, C. Jayakumar, and K. Selvamani,
In this paper, a comprehensive review of IoT has been “Node capture attack in wireless sensor network: A survey,” in Proc.
IEEE Int. Conf. Comput. Intell. Comput. Res. (ICCIC), Coimbatore,
presented, including architectures, enabling technologies, and India, Dec. 2012, pp. 1–3.
security and privacy issues, as well as the integration of [16] F. Bonomi, R. A. Milito, J. Zhu, and S. Addepalli, “Fog computing and
fog/edge computing and IoT to support diverse applications. its role in the Internet of Things,” in Proc. 1st Edition MCC Workshop
Mobile Cloud Comput., Helsinki, Finland, Aug. 2012, pp. 13–16.
Particularly, the relationship and difference between IoT and [17] C. Bormann, A. P. Castellani, and Z. Shelby, “CoAP: An application
CPS has been clarified at the outset. Possible architectures protocol for billions of tiny Internet nodes,” IEEE Internet Comput.,
for IoT have been discussed, including the traditional three- vol. 16, no. 2, pp. 62–67, Mar./Apr. 2012.
[18] A. Botta, W. de Donato, V. Persico, and A. Pescapé, “On the integra-
layer architecture and the SoA-based four-layer architecture. tion of cloud computing and Internet of Things,” in Proc. Int. Conf.
Based on the SoA-based IoT architecture, enabling technolo- Future Internet Things Cloud (FiCloud), Barcelona, Spain, Aug. 2014,
gies in layers (perception layer, network layer, and service pp. 23–30.
[19] N. Bressan et al., “The deployment of a smart monitoring system using
layer) have been detailed, respectively. In addition, to secure wireless sensor and actuator networks,” in Proc. 1st IEEE Int. Conf.
IoT, potential security and privacy issues that could affect Smart Grid Commun. (SmartGridComm), Gaithersburg, MD, USA,
the effectiveness of IoT, and their potential solutions, have Oct. 2010, pp. 49–54.
[20] Z. Cai, Z. He, X. Guan, and Y. Li, “Collective data-sanitization
been presented. To investigate the fog/edge computing-based for preventing sensitive information inference attacks in social
IoT, the relationship between IoT and fog/edge computing and networks,” IEEE Trans. Depend. Secure Comput., to be published,
related issues have been discussed. Furthermore, several appli- doi: 10.1109/TDSC.2016.2613521.
[21] A. Cammarano, C. Petrioli, and D. Spenza, “Pro-energy: A novel
cations, including the smart grid, smart transportation, and energy prediction model for solar and wind energy-harvesting wire-
smart cities, are presented to show how fog/edge computing- less sensor networks,” in Proc. IEEE 9th Int. Conf. Mobile Ad-Hoc
based IoT to be implemented in real-world applications. The Sensor Syst. (MASS), Las Vegas, NV, USA, Oct. 2012, pp. 75–83.
[22] S. Capkun, L. Buttyan, and J.-P. Hubaux, “Self-organized public-
main purpose of this survey is to provide a clear, comprehen- key management for mobile ad hoc networks,” IEEE Trans. Mobile
sive, and deep understanding of IoT and its integration with Comput., vol. 2, no. 1, pp. 52–64, Jan./Mar. 2003.
fog/edge computing, outlining the breadth of topics that IoT [23] A. A. Cardenas, S. Amin, and S. Sastry, “Secure control: Towards
survivable cyber-physical systems,” in Proc. 28th Int. Conf. Distrib.
entails, and highlighting areas that remain unresolved, in an Comput. Syst. Workshops, Beijing, China, Jun. 2008, pp. 495–500.
effort to further promote the development of IoT. [24] P. Casari et al., “The ‘wireless sensor networks for city-wide
ambient intelligence (WISE-WAI)’ project,” Sensors, vol. 9, no. 6,
pp. 4056–4082, May 2009.
R EFERENCES [25] M. A. Chaqfeh and N. Mohamed, “Challenges in middleware solutions
[1] NIST & The Smart Grid. Accessed on Sep. 21, 2016. [Online]. for the Internet of Things,” in Proc. Int. Conf. Collaboration Technol.
Available: http://www.nist.gov/smartgrid/nistandsmartgrid.cfm Syst. (CTS), Denver, CO, USA, May 2012, pp. 21–26.
[2] Fosstrak: Open Source RFID Software Platform. [Online]. Available: [26] U. K. Chaurasia and V. Singh, “MAODV: Modified wormhole detec-
https://fosstrak.github.io/ tion AODV protocol,” in Proc. 6th Int. Conf. Contemp. Comput. (IC3),
[3] Google Self-Driving Car. [Online]. Available: http://www.google.com/ Noida, India, Aug. 2013, pp. 239–243.
selfdrivingcar/how/ [27] F. Chen, T. Xiang, X. Fu, and W. Yu, “User differentiated verifiable file
[4] IEEE Standard for Local and Metropolitan Area Networks: Overview search on the cloud,” IEEE Trans. Services Comput., to be published,
and Architecture, IEEE Standard 802-2001, pp. 1–48, Feb. 2002. doi: 10.1109/TSC.2016.2589245.
[5] S. H. Ahmed, G. Kim, and D. Kim, “Cyber physical system: [28] I.-R. Chen, J. Guo, and F. Bao, “Trust management for service compo-
Architecture, applications and research challenges,” in Proc. IFIP sition in SOA-based IoT systems,” in Proc. IEEE Wireless Commun.
Wireless Days (WD), Valencia, Spain, Nov. 2013, pp. 1–5. Netw. Conf. (WCNC), Istanbul, Turkey, Apr. 2014, pp. 3444–3449.
[6] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “A sur- [29] Z. Chen et al., “A cloud computing based network monitoring and
vey on sensor networks,” IEEE Commun. Mag., vol. 40, no. 8, threat detection system for critical infrastructures,” Big Data Res.,
pp. 102–114, Aug. 2002. vol. 3, pp. 10–23, Apr. 2016.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1139

[30] S. Cheshire, “DNS-based service discovery,” INTERNET-DRAFT [53] A. Laya, V.-I. Bratu, and J. Markendahl, “Who is investing in machine-
draft-cheshire-dnsext-dns-sd-04.txt, 2011. to-machine communications?” in Proc. 24th Eur. Regional Conf. Int.
[31] C.-H. Cho, K.-H. Do, J.-W. Kim, and M.-S. Jun, “Design of RFID Telecommun. Soc., Florence, Italy, Oct. 2013, pp. 1–21.
mutual authentication protocol using time stamp,” in Proc. 4th [54] T. N. Jagatic, N. A. Johnson, M. Jakobsson, and F. Menczer, “Social
Int. Conf. Comput. Sci. Convergence Inf. Technol. (ICCIT), Seoul, phishing,” Commun. ACM, vol. 50, no. 10, pp. 94–100, Oct. 2007.
South Korea, Nov. 2009, pp. 1047–1051. [55] A. J. Jara, P. Martinez-Julia, and A. Skarmeta, “Light-weight multicast
[32] M.-C. Chuang and J.-F. Lee, “TEAM: Trust-extended authentication DNS and DNS-SD (lmDNS-SD): IPv6-based resource and service dis-
mechanism for vehicular ad hoc networks,” IEEE Syst. J., vol. 8, no. 3, covery for the Web of Things,” in Proc. 6th Int. Conf. Innov. Mobile
pp. 749–758, Sep. 2014. Internet Services Ubiquitous Comput. (IMIS), Palermo, Italy, Jul. 2012,
[33] C. Doukas and F. Antonelli, “COMPOSE: Building smart & context- pp. 731–738.
aware mobile applications utilizing IoT technologies,” in Proc. 5th [56] B.-Z. Jing et al., “RFID access authorization by face recognition,” in
IEEE Glob. Inf. Infrastruct. Netw. Symp., Trento, Italy, Oct. 2013, Proc. Int. Conf. Mach. Learn. Cybern., vol. 1. Jul. 2009, pp. 302–307.
pp. 1–6. [57] G. Kalnoor and J. Agarkhed, “QoS based multipath routing for intru-
[34] X. Du, Y. Xiao, M. Guizani, and H.-H. Chen, “An effective key man- sion detection of sinkhole attack in wireless sensor networks,” in Proc.
agement scheme for heterogeneous sensor networks,” Ad Hoc Netw., Int. Conf. Circuit Power Comput. Technol. (ICCPCT), Mar. 2016,
vol. 5, no. 1, pp. 24–34, 2007. pp. 1–6.
[35] X. Fu, Z. Ling, W. Yu, and J. Luo, “Cyber crime scene investi- [58] M. Khanjary and S. M. Hashemi, “Route guidance systems: Review
gations (C2SI) through cloud computing,” in Proc. IEEE 30th Int. and classification,” in Proc. 6th Euro Amer. Conf. Telematics Inf.
Conf. Distrib. Comput. Syst. Workshops (ICDCSW), Genoa, Italy, 2010, Syst. (EATIS), Valencia, Spain, May 2012, pp. 1–7.
pp. 26–31. [59] A. N. Kim, F. Hekland, S. Petersen, and P. Doyle, “When HART
[36] K. Gama, L. Touseau, and D. Donsez, “Combining heterogeneous goes wireless: Understanding and implementing the wirelessHART
service technologies for building an Internet of Things middleware,” standard,” in Proc. IEEE Int. Conf. Emerg. Technol. Factory Autom.,
Comput. Commun., vol. 35, no. 4, pp. 405–417, Feb. 2012. Hamburg, Germany, Sep. 2008, pp. 899–907.
[37] G. Gan, Z. Lu, and J. Jiang, “Internet of Things security analysis,” [60] D. S. Kim, T.-H. Shin, and J. S. Park, “Access control and authorization
in Proc. Int. Conf. Internet Technol. Appl. (iTAP), Wuhan, China, for security of RFID multi-domain using SAML and XACML,” in Proc.
Aug. 2011, pp. 1–4. Int. Conf. Comput. Intell. Security, vol. 2. Nov. 2006, pp. 1587–1590.
[38] W. Gao, J. Nguyen, W. Yu, C. Lu, and D. Ku, “Assessing performance [61] R. Kim, H. Lim, and B. Krishnamachari, “Prefetching-based data dis-
of constrained application protocol (CoAP) in MANET using emula- semination in vehicular cloud systems,” IEEE Trans. Veh. Technol.,
tion,” in Proc. ACM Int. Conf. Rel. Convergent Syst. (RACS), Odense, vol. 65, no. 1, pp. 292–306, Jan. 2015.
Denmark, 2016, pp. 103–108.
[62] D. M. K. Finkenzeller, RFID Handbook: Fundamentals and
[39] J. Girao, D. Westhoff, and M. Schneider, “CDA: Concealed data aggre-
Applications in Contactless Smart Cards, Radio Frequency
gation for reverse multicast traffic in wireless sensor networks,” in Proc.
Identification and Near-Field Communication, 3rd ed. Chichester,
IEEE Int. Conf. Commun. (ICC), vol. 5. Seoul, South Korea, May 2005,
U.K.: Wiley, Aug. 2010.
pp. 3044–3049.
[63] W. Kluge et al., “A fully integrated 2.4-GHz IEEE 802.15.4-compliant
[40] R. Godfrey, D. Ingham, and R. Schloming, OASIS Advanced Message
transceiver for ZigBeeTM trade applications,” IEEE J. Solid-State
Queuing Protocol (AMQP) Version 1.0; OASIS Standard,” Oct. 2012.
Circuits, vol. 41, no. 12, pp. 2767–2775, Dec. 2006.
[41] C. Gomez and J. Paradells, “Wireless home automation networks:
A survey of architectures and technologies,” IEEE Commun. Mag., [64] S. Lahiri, RFID Sourcebook. Upper Saddle River, NJ, USA: IBM Press,
vol. 48, no. 6, pp. 92–101, Jun. 2010. 2005.
[42] G. Gomez, F. J. Lopez-Martinez, D. Morales-Jimenez, and [65] N. D. Lane et al., “A survey of mobile phone sensing,” IEEE Commun.
M. R. McKay, “On the equivalence between interference and eaves- Mag., vol. 48, no. 9, pp. 140–150, Sep. 2010.
dropping in wireless communications,” IEEE Trans. Veh. Technol., [66] T. N. Le, W. Yu, X. Bai, and D. Xuan, “A dynamic geographic hash
vol. 64, no. 12, pp. 5935–5940, Dec. 2015. table for data-centric storage in sensor networks,” in Proc. IEEE Int.
[43] A. Gómez-Goiri and D. López-de Ipiña, “A triple space-based semantic Conf. Comput. Netw. Commun. (WCNC), Las Vegas, NV, USA, 2007,
distributed middleware for Internet of Things,” in Proc. Int. Conf. Web pp. 2168–2174.
Eng., Vienna, Austria, Jul. 2010, pp. 447–458. [67] P. Lee, A. Clark, L. Bushnell, and R. Poovendran, “A passivity
[44] Data Distribution Service (DDS), Version 1.2, Object Manag. Group, framework for modeling and mitigating wormhole attacks on net-
Nov. 2016. worked control systems,” IEEE Trans. Autom. Control, vol. 59, no. 12,
[45] D. Guinard, V. Trifa, S. Karnouskos, P. Spiess, and D. Savio, pp. 3224–3237, Dec. 2014.
“Interacting with the SOA-based Internet of Things: Discovery, query, [68] M. Leo, F. Battisti, M. Carli, and A. Neri, “A federated architecture
selection, and on-demand provisioning of Web services,” IEEE Trans. approach for Internet of Things security,” in Proc. Euro Med Telco
Services Comput., vol. 3, no. 3, pp. 223–235, Jul./Sep. 2010. Conf. (EMTC), Naples, Italy, Nov. 2014, pp. 1–5.
[46] S. Hadim and N. Mohamed, “Middleware: Middleware challenges and [69] J. Lin, X. Yang, W. Yu, and X. Fu, “Towards effective en-route filter-
approaches for wireless sensor networks,” IEEE Distrib. Syst. Online, ing against injected false data in wireless sensor networks,” in Proc.
vol. 7, no. 3, p. 1, Mar. 2006. IEEE Glob. Telecommun. Conf. (GLOBECOM), Houston, TX, USA,
[47] C. Han, J. M. Jornet, E. Fadel, and I. F. Akyildiz, “A cross-layer com- Dec. 2011, pp. 1–5.
munication module for the Internet of Things,” Comput. Netw., vol. 57, [70] J. Lin et al., “On distributed energy routing protocols in the smart
no. 3, pp. 622–633, Feb. 2013. grid,” in Software Engineering, Artificial Intelligence, Networking
[48] W. He, X. Liu, H. Nguyen, K. Nahrstedt, and T. Abdelzaher, “PDA: and Parallel/Distributed Computing (Studies in Computational
Privacy-preserving data aggregation in wireless sensor networks,” Intelligence), vol. 492, R. Lee, Ed. Heidelberg, Germany: Springer,
in Proc. 26th IEEE Int. Conf. Comput. Commun. (INFOCOM), 2013.
Anchorage, AK, USA, May 2007, pp. 2045–2053. [71] J. Lin, W. Yu, and X. Yang, “Towards multistep electricity prices in
[49] X. He, “The two-dimensional bar code application in book manage- smart grid electricity markets,” IEEE Trans. Parallel Distrib. Syst.,
ment,” in Proc. Int. Conf. Web Inf. Syst. Min. (WISM), vol. 1. Sanya, vol. 27, no. 1, pp. 286–302, Jan. 2016.
China, Oct. 2010, pp. 409–411. [72] J. Lin, W. Yu, X. Yang, G. Xu, and W. Zhao, “On false data injec-
[50] R. G. Helps and S. J. Pack, “Cyber-physical system concepts for tion attacks against distributed energy routing in smart grid,” in Proc.
IT students,” in Proc. 14th Annu. ACM SIGITE Conf. Inf. Technol. IEEE/ACM 3rd Int. Conf. Cyber-Phys. Syst. (ICCPS), Beijing, China,
Educ. (SIGITE), Orlando, FL, USA, Oct. 2013, pp. 7–12. Apr. 2012, pp. 183–192.
[51] D. Ćika, M. Draganić, and Z. Šipuš, “Active wireless sensor with radio [73] J. Lin et al., “A novel dynamic en-route decision real-time route
frequency identification chip,” in Proc. 35th Int. Conv., Opatija, Croatia, guidance scheme in intelligent transportation systems,” in Proc. IEEE
May 2012, pp. 727–732. 35th Int. Conf. Distrib. Comput. Syst. (ICDCS), Columbus, OH, USA,
[52] E. Ilie-Zudor, Z. Kemény, F. van Blommestein, L. Monostori, and Jun. 2015, pp. 61–72.
A. van der Meulen, “Survey paper: A survey of applications and [74] M. B. Line, I. A. Tøndel, and M. G. Jaatun, “Cyber security challenges
requirements of unique identification systems and RFID techniques,” in smart grids,” in Proc. 2nd IEEE PES Int. Conf. Exhibit. Innov. Smart
Comput. Ind. Eng., vol. 62, no. 3, pp. 227–252, Apr. 2011. Grid Technol. (ISGT Europe), Manchester, U.K., Dec. 2011, pp. 1–8.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1140 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

[75] Z. Ling et al., “A new cell counter based attack against tor,” in [99] H. B. Pandya and T. A. Champaneria, “Internet of Things: Survey and
Proc. 16th ACM Conf. Comput. Commun. Security (CCS), Chicago, case studies,” in Proc. Int. Conf. Elect. Electron. Signals Commun.
IL, USA, 2009, pp. 578–589. Optim. (EESCO), Visakhapatnam, India, Jan. 2015, pp. 1–6.
[76] Z. Ling et al., “A new cell-counting-based attack against tor,” [100] A. Pingley, W. Yu, N. Zhang, X. Fu, and W. Zhao, “CAP: A context-
IEEE/ACM Trans. Netw., vol. 20, no. 4, pp. 1245–1261, Aug. 2012. aware privacy protection system for location-based services,” in Proc.
[77] H. Liu, S. Hua, X. Zhuo, D. Chen, and X. Cheng, “Cooperative spec- IEEE Int. Conf. Distrib. Comput. Syst. (ICDCS), Montreal, QC, Canada,
trum sharing of multiple primary users and multiple secondary users,” 2009, pp. 49–57.
Digit. Commun. Netw., vol. 2, no. 4, pp. 191–195, 2016. [101] A. Pingley, W. Yu, N. Zhang, X. Fu, and W. Zhao, “A context-aware
[78] Y. Liu and G. Zhou, “Key technologies and applications of scheme for privacy-preserving location-based services,” Comput. Netw.,
Internet of Things,” in Proc. 5th Int. Conf. Intell. Comput. Technol. vol. 56, no. 11, pp. 2551–2568, 2012.
Autom. (ICICTA), Zhangjiajie, China, Jan. 2012, pp. 197–200. [102] R. Prodan and S. Ostermann, “A survey and taxonomy of infrastructure
[79] S. Lohr, “The age of big data,” New York Times, Feb. 2012. as a service and Web hosting cloud providers,” in Proc. 10th IEEE/ACM
[80] P. López, D. Fernández, A. J. Jara, and A. F. Skarmeta, “Survey of Int. Conf. Grid Comput., Banff, AB, Canada, Oct. 2009, pp. 17–25.
Internet of Things technologies for clinical environments,” in Proc. 27th [103] Q. Pu et al., “Low latency geo-distributed data analytic,” in Proc. ACM
Int. Conf. Adv. Inf. Netw. Appl. Workshops (WAINA), Barcelona, Spain, SIGCOMM, London, U.K., Aug. 2015, pp. 421–434.
Mar. 2013, pp. 1349–1354. [104] K. P. N. Puttaswamy, R. Bhagwan, and V. N. Padmanabhan,
[81] T. Luckenbach, P. Gober, S. Arbanowski, A. Kotsopoulos, and K. Kim, “Anonygator: Privacy and integrity preserving data aggregation,” in
“Tinyrest-a protocol for integrating sensor networks into the Internet,” Proc. ACM/IFIP/USENIX 11th Int. Conf. Middleware, Bengaluru,
in Proc. Real-World Wireless Sensor Netw. (REALWSN), Jun. 2005, India, 2010, pp. 85–106.
pp. 101–105. [105] F. Qiu, F. Wu, and G. Chen, “Privacy and quality preserving multimedia
[82] S. U. Maheswari, N. S. Usha, E. A. M. Anita, and K. R. Devi, “A novel data aggregation for participatory sensing systems,” IEEE Trans.
robust routing protocol RAEED to avoid DoS attacks in WSN,” in Proc. Mobile Comput., vol. 14, no. 6, pp. 1287–1300, Jun. 2015.
Int. Conf. Inf. Commun. Embedded Syst. (ICICES), Chennai, India, [106] X. Ren, X. Yang, J. Lin, Q. Yang, and W. Yu, “On scaling pertur-
Feb. 2016, pp. 1–5. bation based privacy-preserving schemes in smart metering systems,”
[83] R. Mahmoud, T. Yousuf, F. Aloul, and I. Zualkernan, “Internet in Proc. 22nd Int. Conf. Comput. Commun. Netw. (ICCCN), Nassau,
of Things (IoT) security: Current status, challenges and prospec- The Bahamas, Jul. 2013, pp. 1–7.
tive measures,” in Proc. 10th Int. Conf. Internet Technol. Secured [107] D. Romero et al., RESTful Integration of Heterogeneous Devices
Trans. (ICITST), London, U.K., Dec. 2015, pp. 336–341. in Pervasive Environments. Heidelberg, Germany: Springer, 2010,
[84] S. Mallapuram, N. Ngwum, F. Yuan, C. Lu, and W. Yu, “Smart city: pp. 1–14.
The state of the art, datasets, and evaluation platforms,” in Proc. 16th [108] L. Roselli et al., “Review of the present technologies concurrently con-
IEEE/ACIS Int. Conf. Comput. Inf. Sci. (ICIS), 2017. tributing to the implementation of the Internet of things (IoT) paradigm:
[85] L. D. Mello and L. T. Kubota, “Review of the use of biosensors as RFID, green electronics, WPT and energy harvesting,” in Proc. IEEE
analytical tools in the food and drink industries,” Food Chem., vol. 77, Topical Conf. Wireless Sensors Sensor Netw. (WiSNet), San Diego, CA,
no. 2, pp. 237–256, 2002. USA, Jan. 2015, pp. 1–3.
[86] D. Miorandi, S. Sicari, F. De Pellegrini, and I. Chlamtac, “Internet of [109] A. Roxin, C. Dumez, N. Cottin, J. Gaber, and M. Wack, “TransportML:
Things,” Ad Hoc Netw., vol. 10, no. 7, pp. 1497–1516, Sep. 2012. A middleware for location-based services collaboration,” in Proc. 3rd
[87] M. H. Miraz, M. Ali, P. S. Excell, and R. Picking, “A review on Internet Int. Conf. New Technol. Mobility Security, Cairo, Egypt, Dec. 2009,
of Things (IoT), Internet of everything (IoE) and Internet of Nano pp. 1–6.
Things (IoNT),” in Proc. Internet Technol. Appl. (ITA), Wrexham, U.K., [110] A. K. Sahoo, A. Das, and M. Tiwary, “Firewall engine based on graph-
Sep. 2015, pp. 219–224. ics processing unit,” in Proc. Int. Conf. Adv. Commun. Control Comput.
[88] A. Mitrokotsa, M. R. Rieback, and A. S. Tanenbaum, “Classifying Technol. (ICACCCT), May 2014, pp. 758–763.
RFID attacks and defenses,” Inf. Syst. Front., vol. 12, no. 5, [111] P. Saint-Andre, “Extensible messaging and presence protocol (XMPP):
pp. 491–505, Nov. 2010. Core,” Internet Eng. Task Force, Fremont, CA, USA, RFC 6121,
[89] Y. Mo and B. Sinopoli, “Secure control against replay attacks,” Mar. 2011.
in Proc. 47th Annu. Allerton Conf. Commun. Control Comput., [112] M. Sarkar and D. B. Roy, “Prevention of sleep deprivation attacks using
Monticello, IL, USA, Sep./Oct. 2009, pp. 911–918. clustering,” in Proc. 3rd Int. Conf. Electron. Comput. Technol. (ICECT),
[90] J. R. Mohammed, “A new simple adaptive noise cancellation scheme vol. 5. Apr. 2011, pp. 391–394.
based on ALE and NLMS filter,” in Proc. 5th Annu. Conf. Commun. [113] S. Schneider, Understanding the Protocols Behind the Internet
Netw. Services Res. (CNSR), Fredericton, NB, Canada, May 2007, of Things, Electron. Design, Oct. 2013. [Online]. Available:
pp. 245–254. http://electronicdesign.com/iot/understanding-protocols-behind-
[91] A. Mukaddam, I. Elhajj, A. Kayssi, and A. Chehab, “IP spoofing detec- internet-things
tion using modified hop count,” in Proc. IEEE 28th Int. Conf. Adv. Inf. [114] A. Seshadri, A. Perrig, L. van Doorn, and P. Khosla, “SWATT:
Netw. Appl., Victoria, BC, Canada, May 2014, pp. 512–516. Software-based attestation for embedded devices,” in Proc. IEEE Symp.
[92] A. Mukherjee, D. Saha, and C. Biswas, “Present scenarios and future Security Privacy, Berkeley, CA, USA, May 2004, pp. 272–282.
challenges in pervasive middleware,” in Proc. 1st Int. Conf. Commun. [115] K. Sha, W. Wei, A. Yang, and W. Shi, “Security in Internet of Things:
Syst. Software Middleware, New Delhi, India, Jan. 2006, pp. 1–5. Opportunities and challenges,” in Proc. Int. Conf. Identification Inf.
[93] R. Muniz, L. Junco, and A. Otero, “A robust software barcode reader Knowl. Internet Things, Oct. 2016, pp. 49–50.
using the Hough transform,” in Proc. Int. Conf. Inf. Intell. Syst., [116] W. Shi, J. Cao, Q. Zhang, Y. Li, and L. Xu, “Edge computing: Vision
Bethesda, MD, USA, Oct. 1999, pp. 313–319. and challenges,” IEEE Internet Things J., vol. 3, no. 5, pp. 637–646,
[94] J. Nakamura, Image Sensors and Signal Processing for Digital Still Oct. 2016.
Cameras (Opt. Sci. Eng.). Boca Raton, FL, USA: CRC Press, [117] V. Soni, P. Modi, and V. Chaudhri, “Detecting sinkhole attack in wire-
Aug. 2005. less sensor network,” Int. J. Appl. Innov. Eng. Manag., vol. 2, no. 2,
[95] J. Newsome, E. Shi, D. Song, and A. Perrig, “The Sybil attack pp. 29–32, Feb. 2013.
in sensor networks: Analysis & defenses,” in Proc. 3rd IEEE Int. [118] P. Spiess et al., “Soa-based integration of the Internet of Things in
Symp. Inf. Process. Sensor Netw., Berkeley, CA, USA, Apr. 2004, enterprise services,” in Proc. IEEE Int. Conf. Web Services (ICWS),
pp. 259–268. Los Angeles, CA, USA, Jul. 2009, pp. 968–975.
[96] R. P. Padhy, M. R. Patra, and S. C. Satapathy, “Cloud computing: [119] J. A. Stankovic, “Research directions for the Internet of Things,” IEEE
Security issues and research challenges,” Int. J. Comput. Sci. Inf. Internet Things J., vol. 1, no. 1, pp. 3–9, Feb. 2014.
Technol. Security, vol. 1, no. 2, pp. 136–146, Dec. 2011. [120] I. Stojmenovic and S. Wen, “The fog computing paradigm: Scenarios
[97] M. R. Palattella et al., “On optimal scheduling in duty-cycled indus- and security issues,” in Proc. Federated Conf. Comput. Sci. Inf.
trial IoT applications using IEEE802.15.4e TSCH,” IEEE Sensors J., Syst. (FedCSIS), Warsaw, Poland, Sep. 2014, pp. 1–8.
vol. 13, no. 10, pp. 3655–3666, Oct. 2013. [121] I. Studnia et al., “Survey on security threats and protection mechanisms
[98] M. R. Palattella et al., “Standardized protocol stack for the Internet in embedded automotive networks,” in Proc. 43rd Annu. IEEE/IFIP
of (important) Things,” IEEE Commun. Surveys Tuts., vol. 15, no. 3, Conf. Depend. Syst. Netw. Workshop (DSN W), Budapest, Hungary,
pp. 1389–1406, 3rd Quart., 2013. Jun. 2013, pp. 1–12.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 LIN et al.: SURVEY ON IoTs: ARCHITECTURE, ENABLING TECHNOLOGIES, SECURITY AND PRIVACY, AND APPLICATIONS 1141

[122] H. Suo, J. Wan, C. Zou, and J. Liu, “Security in the Internet of Things: [146] X. Yang, P. Zhao, X. Zhang, J. Lin, and W. Yu, “Toward a Gaussian-
A review,” in Proc. Int. Conf. Comput. Sci. Electron. Eng. (ICCSEE), mixture model-based detection scheme against data integrity attacks in
vol. 3. Hangzhou, China, Mar. 2012, pp. 648–651. the smart grid,” IEEE Internet Things J., vol. 4, no. 1, pp. 147–161,
[123] J. Tan and S. G. M. Koo, “A survey of technologies in Internet Feb. 2017.
of Things,” in Proc. IEEE Int. Conf. Distrib. Comput. Sensor Syst., [147] S. Yi, C. Li, and Q. Li, “A survey of fog computing: Concepts, appli-
Marina Del Rey, CA, USA, May 2014, pp. 269–274. cations and issues,” in Proc. Workshop Mobile Big Data, Hangzhou,
[124] L. Tan and N. Wang, “Future Internet: The Internet of Things,” in China, Jun. 2015, pp. 37–42.
Proc. 3rd Int. Conf. Adv. Comput. Theory Eng. (ICACTE), vol. 5. [148] W. Yu, X. Fu, S. Graham, D. Xuan, and W. Zhao, “DSSS-based
Chengdu, China, Aug. 2010, pp. V5-376–V5-380. flow marking technique for invisible traceback,” in Proc. IEEE Symp.
[125] C.-W. Tsai, C.-F. Lai, M.-C. Chiang, and L. T. Yang, “Data mining for Security Privacy (S P), Oakland, CA, USA, 2007, pp. 18–32.
Internet of Things: A survey,” IEEE Commun. Surveys Tuts., vol. 16, [149] W. Yu, D. Griffith, L. Ge, S. Bhattarai, and N. Golmie, “An integrated
no. 1, pp. 77–97, 1st Quart., 2014. detection system against false data injection attacks in the smart grid,”
[126] S. Wang and C. Wang, “Joint optimization of spectrum and energy Security Commun. Netw., vol. 8, no. 2, pp. 91–109, 2015.
efficiency in cognitive radio networks,” Digit. Commun. Netw., vol. 1, [150] W. Yu, T. N. Le, D. Xuan, and W. Zhao, “Query aggregation for
no. 3, pp. 161–170, 2015. providing efficient data services in sensor networks,” in Proc. IEEE
[127] X. Wang, W. Yu, A. Champion, X. Fu, and D. Xuan, “Detecting Mobile Ad-Hoc Sensor Syst. (MASS), Fort Lauderdale, FL, USA, 2004,
worms via mining dynamic program execution,” in Proc. 3rd Int. Conf. pp. 31–40.
Security Privacy Commun. Netw., Nice, France, 2007, pp. 412–421. [151] W. Yu and J. Lee, “Efficient energy sensitive routing protocols in
[128] T. Winter et al., “RPL: Ipv6 routing protocol for low-power and lossy mobile ad-hoc networks,” in Proc. Int. Conf. Wireless Netw., Las Vegas,
networks,” Internet Eng. Task Force, Fremont, CA, USA, RFC 6550, NV, USA, 2002.
Mar. 2012. [152] W. Yu, G. Xu, Z. Chen, and P. Moulema, “Cyber crime scene investiga-
[129] D. Wu, B. Yang, and R. Wang, “Scalable privacy-preserving big tions (C2SI) through cloud computing,” in Proc. IEEE Conf. Commun.
data aggregation mechanism,” Digit. Commun. Netw., vol. 2, no. 3, Netw. Security (CNS), Washington, DC, USA, 2013, pp. 26–31.
pp. 122–129, 2016. [153] W. Yu, D. Xuan, B. Graham, S. Santhanam, R. Bettati, and W. Zhao,
[130] F.-J. Wu, Y.-F. Kao, and Y.-C. Tseng, “Review: From wireless sensor “An integrated middleware-based solution for supporting secured
networks towards cyber physical systems,” Pervasive Mobile Comput., dynamic-coalition applications in heterogeneous environments,” in
vol. 7, no. 4, pp. 397–413, Aug. 2011. Proc. IEEE Workshop Inf. Assurance Security, West Point, NY, USA,
[131] J. Wu and W. Zhao, “Design and realization of WInternet: From Net 2002, pp. 259–264.
of Things to Internet of Things,” ACM Trans. Cyber Phys. Syst., vol. 1, [154] W. Yu, N. Zhang, X. Fu, and W. Zhao, “Self-disciplinary worms and
no. 1, Feb. 2017, Art. no. 2. countermeasures: Modeling and analysis,” IEEE Trans. Parallel Distrib.
[132] M. Wu, T.-J. Lu, F.-Y. Ling, J. Sun, and H.-Y. Du, “Research on Syst., vol. 21, no. 10, pp. 1501–1514, Oct. 2010.
the architecture of Internet of Things,” in Proc. 3rd Int. Conf. Adv. [155] A. Zanella, N. Bui, A. Castellani, L. Vangelista, and M. Zorzi, “Internet
Comput. Theory Eng. (ICACTE), vol. 5. Chengdu, China, Aug. 2010, of Things for smart cities,” IEEE Internet Things J., vol. 1, no. 1,
pp. V5-484–V5-487. pp. 22–32, Feb. 2014.
[133] Y. Xiao et al., “A survey of key management schemes in wireless sensor [156] A. Zaslavsky, C. Perera, and D. Georgakopoulos, “Sensing as a service
networks,” J. Comput. Commun., vol. 30, nos. 11–12, pp. 2314–2341, and big data,” in Proc. Int. Conf. Adv. Cloud Comput. (ACC), Charlotte,
2007. NC, USA, Jul. 2012.
[134] L. Xu, C. Jiang, J. Wang, J. Yuan, and Y. Ren, “Information secu- [157] J. Zhang, D. Gu, Z. Guo, and L. Zhang, “Differential power cryptanaly-
rity in big data: Privacy and data mining,” IEEE Access, vol. 2, sis attacks against present implementation,” in Proc. 3rd Int. Conf. Adv.
pp. 1149–1176, Oct. 2014. Comput. Theory Eng. (ICACTE), vol. 6. Chengdu, China, Aug. 2010,
[135] L. D. Xu, “Enterprise systems: State-of-the-art and future trends,” IEEE pp. V6-61–V6-65.
Trans. Ind. Informat., vol. 7, no. 4, pp. 630–640, Nov. 2011. [158] K. Zhang, X. Liang, R. Lu, and X. Shen, “Sybil attacks and their
[136] L. D. Xu, W. He, and S. Li, “Internet of Things in industries: A survey,” defenses in the Internet of Things,” IEEE Internet Things J., vol. 1,
IEEE Trans. Ind. Informat., vol. 10, no. 4, pp. 2233–2243, Nov. 2014. no. 5, pp. 372–383, Oct. 2014.
[137] B. Yang, K. Wu, and R. Karri, “Scan based side channel attack on ded- [159] L. Zhang, Z. Cai, and X. Wang, “Fakemask: A novel privacy preserving
icated hardware implementations of data encryption standard,” in Proc. approach for smartphones,” IEEE Trans. Netw. Service Manag., vol. 13,
Int. Test Conf. (ITC), Charlotte, NC, USA, Oct. 2004, pp. 339–344. no. 2, pp. 335–348, Jun. 2016.
[138] H. Yang and S.-H. Yang, “RFID sensor network network architectures [160] X. Zhang, X. Yang, J. Lin, G. Xu, and W. Yu, “Towards efficient
to integrate RFID, sensor and WSN,” Meas. Control, vol. 40, no. 2, and secured real-time pricing in the smart grid,” in Proc. IEEE Glob.
pp. 56–59, Mar. 2007. Commun. Conf. (GLOBECOM), San Diego, CA, USA, Dec. 2015,
[139] Q. Yang, L. Chang, and W. Yu, “On false data injection attacks against pp. 1–6.
Kalman filtering in power system dynamic state estimation,” Security [161] X. Zhang, X. Yang, J. Lin, G. Xu, and W. Yu, “On data integrity attacks
Commun. Netw., vol. 9, no. 9, pp. 833–849, Jun. 2016. against real-time pricing in energy-based cyber-physical systems,”
[140] Q. Yang et al., “On data integrity attacks against optimal power flow in IEEE Trans. Parallel Distrib. Syst., vol. 28, no. 1, pp. 170–187,
power grid systems,” in Proc. Annu. IEEE Consumer Commun. Netw. Jan. 2017.
Conf. (CCNC), Las Vegas, NV, USA, 2017. [162] K. Zhao and L. Ge, “A survey on the Internet of Things security,”
[141] Q. Yang et al., “On false data-injection attacks against power system in Proc. 9th Int. Conf. Comput. Intell. Security (CIS), Dec. 2013,
state estimation: Modeling and countermeasures,” IEEE Trans. Parallel pp. 663–667.
Distrib. Syst., vol. 25, no. 3, pp. 717–729, Mar. 2014. [163] N. Zhao, F. R. Yu, M. Li, and V. C. M. Leung, “Anti-eavesdropping
[142] X. Yang et al., “Towards a low-cost remote memory attestation for the schemes for interference alignment (IA)-based wireless networks,”
smart grid,” Sensors, vol. 15, no. 8, pp. 20799–20824, Aug. 2015. IEEE Trans. Wireless Commun., vol. 15, no. 8, pp. 5719–5732,
[143] X. Yang et al., “A novel en-route filtering scheme against false data Aug. 2016.
injection attacks in cyber-physical networked systems,” IEEE Trans. [164] K. Zheng, F. Hu, W. Wang, W. Xiang, and M. Dohler, “Radio
Comput., vol. 64, no. 1, pp. 4–18, Jan. 2015. resource allocation in LTE-advanced cellular networks with m2m com-
[144] X. Yang, X. Ren, J. Lin, and W. Yu, “On binary decomposition munications,” IEEE Commun. Mag., vol. 50, no. 7, pp. 184–192,
based privacy-preserving aggregation schemes in real-time monitor- Jul. 2012.
ing systems,” IEEE Trans. Parallel Distrib. Syst., vol. 27, no. 10, [165] X. Zheng, Z. Cai, J. Li, and H. Gao, “Location-privacy-aware
pp. 2967–2983, Oct. 2016. review publication mechanism for local business service systems,” in
[145] X. Yang et al., “Data integrity attacks against the distributed Proc. 36th Annu. IEEE Int. Conf. Comput. Commun. (INFOCOM),
real-time pricing in the smart grid,” in Proc. IEEE Int. Perform. Atlanta, GA, USA, 2017, pp. 1–9.
Comput. Commun. Conf. (IPCCC), Las Vegas, NV, USA, 2016, [166] H. Zhou, The Internet of Things in the Cloud: A Middleware
pp. 1–8. Perspective, 1st ed. Boca Raton, FL, USA: CRC Press, Oct. 2012.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.
 1142 IEEE INTERNET OF THINGS JOURNAL, VOL. 4, NO. 5, OCTOBER 2017

Jie Lin received the B.S. and Ph.D. degrees from the Xinyu Yang received the B.S., M.S., Ph.D. degrees,
Department of Computer Science and Technology, and Diploma degree in computer science and
Xi’an Jiaotong University, Xi’an, China, in 2009 and technology from Xi’an Jiaotong University, Xi’an,
2013, respectively. China, in 1995, 1997, 2001, and 2001, respectively.
He is currently an Associate Professor with the He is currently a Professor with the Department of
Department of Computer Science and Technology, Computer Science and Technology, Xi’an Jiaotong
Xi’an Jiaotong University. His current research University. His current research interests include
interests include smart grid, cyberspace security, and wireless communication, mobile ad hoc networks,
computer networks. and network security.

Wei Yu received the B.S. degree in electrical engi-

neering from the Nanjing University of Technology, Hanlin Zhang received the B.S. degree in software
Nanjing, China, in 1992, the M.S. degree in elec- engineering from Qingdao University, Qingdao,
trical engineering from Tongji University, Shanghai, China, in 2010, and the M.S. degree in applied
China, in 1995, and the Ph.D. degree in computer information technology and Doctoral degree in infor-
engineering from Texas A&M University, College mation technology from Towson University, Towson,
Station, TX, USA, in 2008. MD, USA, in 2011 and 2016, respectively.
He was with Cisco Systems Inc., San Jose, CA, He is currently an Assistant Professor with
USA, for nine years. He is currently an Associate the Department of Computer Science, Qingdao
Professor with the Department of Computer and University. His current research interests include
Information Sciences, Towson University, Towson, information security, cloud security, mobile security,
MD, USA. His current research interests include cyberspace security and pri- and network security.
vacy, computer networks, cyber-physical systems, and distributed computing.
Dr. Yu was a recipient of the 2014 NSF CAREER Award, the 2015
University System of Maryland (USM) Regents’ Faculty Award for Excellence
in Scholarship, Research, or Creative Activity, the USM’s Wilson H. Elkins
Professorship Award in 2016, and the Best Paper Awards from IEEE ICC
2008, ICC 2013, and IEEE IPCCC 2016. Wei Zhao received the undergraduate program in
physics degree from Shaanxi Normal University,
Xi’an, China, in 1977, and the M.S. and Ph.D.
degrees in computer and information sciences from
the University of Massachusetts Amherst, Amherst,
MA, USA, in 1983 and 1986, respectively.
Nan Zhang received the B.S. degree in computer He was the Dean of the School of Science,
science from Peking University, Beijing, China, in Rensselaer Polytechnic Institute, Troy, NY, USA. He
2001, and the Ph.D. degree in computer science from is currently the Rector with the University of Macau,
Texas A&M University, College Station, TX, USA, Macau, China. From 2005 to 2006, he was the
in 2006. Director for the Division of Computer and Network
He is an Associate Professor of Computer Science Systems, U.S. National Science Foundation, Arlington, TX, USA, when he
with the George Washington University, Washington, was on leave from Texas A&M University, College Station, TX, USA, where
DC, USA. His current research interests include he was a Senior Associate Vice President for Research and a Professor of
databases, data analytics, and information pri- Computer Science. He was the Founding Director of the Texas A&M Center
vacy/security. for Information Security and Assurance, which has been recognized as a
Dr. Zhang was a recipient of several awards Center of Academic Excellence in Information Assurance Education by the
including the NSF CAREER Award in 2008, the Best Paper Award of IEEE National Security Agency. Since then, he has been a Faculty Member with
ICC 2013 and IEEE NAS 2010, the Best Student Paper Award of ACM Amherst College, Amherst, MA, USA, the University of Adelaide, Adelaide
CIKM 2013, the Best Paper Nomination from IEEE ISI 2015, and the GW SA, Australia, and Texas A&M University. His current research interests
Technology Transfer Innovation Price and the First Place finish at the GW include distributed computing, real-time systems, computer networks, and
Business Plan Competition, both in 2012. cyber space security.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL. Downloaded on February 05,2025 at 12:41:54 UTC from IEEE Xplore. Restrictions apply.