Edge-computing-driven Internet of Things: A Survey

LINGHE KONG, JINLIN TAN, JUNQIN HUANG, GUIHAI CHEN, and

SHUAITIAN WANG, Shanghai Jiao Tong University, China
XI JIN and PENG ZENG, Shenyang Institute of Automation, Chinese Academy of Sciences, China
MUHAMMAD KHAN, King Saud University, Kingdom of Saudi Arabia
SAJAL K. DAS, Missouri University of Science and Technology, USA

The Internet of Things (IoT) is impacting the world’s connectivity landscape. More and more IoT devices are
connected, bringing many benefits to our daily lives. However, the influx of IoT devices poses non-trivial
challenges for the existing cloud-based computing paradigm. In the cloud-based architecture, a large amount
of IoT data is transferred to the cloud for data management, analysis, and decision making. It could not only
cause a heavy workload on the cloud but also result in unacceptable network latency, ultimately undermining
the benefits of cloud-based computing. To address these challenges, researchers are looking for new comput-
ing models for the IoT. Edge computing, a new decentralized computing model, is valued by more and more
researchers in academia and industry. The main idea of edge computing is placing data processing in near-
edge devices instead of remote cloud servers. It is promising to build more scalable, low-latency IoT systems.
Many studies have been proposed on edge computing and IoT, but a comprehensive survey of this crossover
area is still lacking.
In this survey, we first introduce the impact of edge computing on the development of IoT and point
out why edge computing is more suitable for IoT than other computing paradigms. Then, we analyze the
necessity of systematical investigation on the edge-computing-driven IoT (ECDriven-IoT) and summarize
new challenges occurring in ECDriven-IoT. We categorize recent advances from bottom to top, covering six
aspects of ECDriven-IoT. Finally, we conclude lessons learned and propose some challenging
CCS Concepts: • Computer systems organization → Distributed architectures; Real-time operating sys-
tems; • Networks → Network protocols; • Security and privacy; • Computing methodologies → Dis-
tributed computing methodologies;
Additional Key Words and Phrases: Edge computing, Internet of Things

This work was supported in part by NSFC Grants No. 62141220, No. 61972253, No. U1908212, No. 72061127001, No. 62172276,
and No. 61972254, the Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher
Learning, and Open Research Projects of Zhejiang Lab No. 2022NL0AB01.
174
Authors’ addresses: L. Kong (corresponding author), J. Tan, J. Huang, G. Chen, and S. Wang, Shanghai Jiao Tong
University, No. 800, Dongchuan Road, Shanghai, 200240; emails: {linghe.kong, jinlintan, junqin.huang}@sjtu.edu.cn,
[email protected], [email protected]; X. Jin and P. Zeng, Shenyang Institute of Automation, Chinese Academy of
Sciences, No. 135, Chuangxin Road, Hunnan District, Shenyang City, Liaoning Province, 110169; emails: {jinxi, zp}@sia.cn;
M. Khan, King Saud University, Kingdom of Saudi Arabia, Po Box 92144, Riyadh 11653; email: [email protected]; S.
K. Das, Missouri University of Science and Technology, Rolla, MO 65409, USA; email: [email protected].
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee
provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and
the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored.
Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires
prior specific permission and/or a fee. Request permissions from [email protected].
© 2022 Association for Computing Machinery.
0360-0300/2022/12-ART174 $15.00
https://doi.org/10.1145/3555308

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:2 L. Kong et al.

ACM Reference format:

Linghe Kong, Jinlin Tan, Junqin Huang, Guihai Chen, Shuaitian Wang, Xi Jin, Peng Zeng, Muhammad Khan,
and Sajal K. Das. 2022. Edge-computing-driven Internet of Things: A Survey. ACM Comput. Surv. 55, 8, Arti-
cle 174 (December 2022), 41 pages.
https://doi.org/10.1145/3555308

1 INTRODUCTION
The Internet of Things (IoT) is a revolutionary approach that interlinks uniquely addressable
physical and virtual devices through different communication protocols. According to the statis-
tics, the number of wireless-connected IoT devices will reach 50 billion by 2025 [51]. Potential
devices include smartphones, bio-nano things, body sensors, smart tags, wearable devices, embed-
ded objects, and traditional electronic gadgets [6]. These devices usually have a variety of sensors
inside for collecting environmental data, which are fundamental elements of data-driven intelli-
gence. Thus, massive deployed devices lead to an explosive data increase in the meantime. The
collected data need to be processed and analyzed before providing useful results for users. But
the computation ability of lightweight IoT devices is quite limited. One solution to this problem
is cloud computing. In the cloud-based paradigm, IoT data is first transferred to the cloud server
for processing, and then computing results will be sent back to devices. However, data transmis-
sion rate and network bandwidth could become bottlenecks to the further development of massive
IoT [162]. Moreover, as most IoT devices will generate personal and sensitive data, it is inappropri-
ate to send all IoT data to remote cloud servers, which will cause security and privacy concerns.
Edge computing is a new computing paradigm that directs computational data, applications,
and services away from cloud servers to the network edge. Content providers and application
developers can use edge computing to offer users services closer to them in geography, which
can accelerate the response speed of services. Edge computing is characterized as high bandwidth,
ultra-low latency, and real-time access to network information [86, 150]. And IoT applications usu-
ally require real-time response, privacy preservation, and massive data transmission. Compared
with cloud computing, edge computing has the potential to match large-scale IoT applications’
requirements.
The common goal of IoT and edge computing is to perform seamless computing anytime and
anywhere, but they act in different roles in the system. IoT focuses on endpoint sensing, while edge
computing focuses on near-field computation. Thus, it is promising for edge-computing-driven
IoT (ECDriven-IoT) systems to make these two technologies complement each other. Nowadays,
IoT has been widely used in many complex scenarios, such as smart homes, smart cities, smart
grids, virtual reality (VR), augmented reality (AR), and autonomous driving. ECDriven-IoT can
benefit these IoT applications from three aspects: (1) Real-time response and high quality of
services (QoS). Edge computing can provide shorter network latency than cloud computing, as
edge servers lie closer to IoT devices in geography. This superiority can support high-demand real-
time IoT applications better. Owing to a majority of data processed in edge servers, the amount
of data offloaded to the cloud can be largely reduced. Thus, ECDriven-IoT can bring higher QoS
for those real-time IoT applications. (2) Low energy consumption. Most IoT nodes are power-
limited devices, but synchronizing large amounts of sensing data to the remote cloud wastes much
power. With edge computing, IoT nodes only need to send data to local edge servers, so the energy
consumption of IoT nodes can be decreased to a lower level. Thus, ECDriven-IoT can extend the
lifetime of IoT nodes and reduce the maintenance overhead. (3) High scalability. One unavoid-
able challenge in cloud-based IoT systems is the large-scale access requirements. The cloud server
could be the system bottleneck due to large amounts of concurrent connections from IoT nodes.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:3

In ECDriven-IoT, edge servers (e.g., base stations) provide moderate computing resources in a dis-
tributed manner, so ECDriven-IoT can provide good scalability that satisfies the requirements of
large-scale IoT applications like smart cities or autonomous driving. Therefore, we believe edge
computing is indispensable for future IoT, and the study combining IoT and edge computing has
academic prospects.
Many surveys [57, 157] pointed out that recent advances related to IoT and edge computing have
made many efforts to satisfy these requirements. However, when combining edge computing and
IoT, there are still several new challenges regarding how to efficiently integrate these two technolo-
gies and bridge the difference between them. We summarize three new challenges in ECDriven-IoT
systems:
• Heterogeneity of edge computing and IoT. IoT devices are working everywhere and vary
across different scenarios. Thus, there are various hardware devices and communication pro-
tocols in IoT systems. For edge computing, the deployment architecture of edge nodes also
requires different solutions for different scenarios. Thus, combining edge computing with
IoT is faced with the challenge on how to unify the diversity of IoT and edge computing and
make them complement each other. To efficiently apply edge computing in heterogeneous
IoT systems, the cooperation architecture of ECDriven-IoT, hardware devices, and commu-
nication protocols need to be explored and form industry standards.
• Coordination between communication and computing. When combining edge com-
puting with IoT, the system is more complex than only IoT or edge computing-based ones.
The communication between edge servers and IoT devices will bring extra consumption to
the whole system. Besides, edge servers and IoT devices are limited in power and computing
capacity. For example, if IoT devices offload all workload to edge servers, then it must pose
a greater demand on communication cost and computing capacity of edge nodes. So, we
should explore how to allocate workload between edge servers and IoT devices to balance
the cost of communication and computation.
• Complicated security and privacy issues. How to guarantee systems’ security and pri-
vacy is always a significant challenge in IoT and edge computing. However, these issues
become more tricky due to the heterogeneity and limited computing capability of ECDriven-
IoT. IoT devices and edge servers are vulnerable to various attacks. Once any of these points
are compromised, the system will be in great danger. So, a qualified ECDriven-IoT system
should fully consider possible security threats and countermeasures in different application
scenarios.
There have been numerous studies from IoT to edge computing, covering every aspect of the
system. It is necessary to reveal what research has been done in this area and explore what the
future research direction is. Many excellent surveys have focused on either IoT or edge computing.
In the IoT aspect, a plethora of surveys have referred to architecture [57, 157], communication [106,
124], IoT application [61, 159] as well as security and privacy [10, 110]. As for edge computing
combined with IoT, there have also been several surveys from different perspectives as shown in
Table 1. However, these surveys:
• covered a limited number of research areas, and the system architecture of ECDriven-IoT
has not been discussed.
• revealed many challenges in edge computing or IoT, but those new challenges arising from
combining edge computing and IoT have not been explored.
ECDriven-IoT is a promising solution taking advantage of edge computing to build scalable
and efficient IoT systems. Both academia and industry need a survey to explain what happens

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:4 L. Kong et al.

Table 1. Comparison of Previous Surveys in IoT and Edge Computing

Edge IoT-Edge Operating Communication
Survey IoT Architecture Computing Application Challenge
Computing Architecture System Protocol
Rafique et al. [142] ✓ ✓ ✓ ✓ ✓ ✓ ✓
J. Pan et al. [130] ✓ ✓ ✓ ✓ ✓
Abbas et al. [2] ✓ ✓ ✓ ✓
Porambage et al. [138] ✓ ✓ ✓ ✓ ✓
Javed et al. [80] ✓ ✓ ✓ ✓
Mouradian et al. [118] ✓ ✓ ✓ ✓
Salman et al. [152] ✓ ✓ ✓ ✓ ✓ ✓
Roman et al. [146] ✓ ✓ ✓
Baktir et al. [17] ✓ ✓ ✓ ✓
Y. Mao et al. [108] ✓ ✓ ✓ ✓
W. Yu et al. [194] ✓ ✓ ✓ ✓ ✓ ✓
Elazhary [49] ✓ ✓ ✓ ✓
Y. Ai et al. [7] ✓ ✓ ✓ ✓ ✓ ✓
Alwarafy et al. [12] ✓ ✓ ✓ ✓ ✓
Jararweh et al. [78] ✓ ✓ ✓ ✓ ✓
Our Survey ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

when edge computing encounters IoT, what benefits it brings in, and what new challenges it faces.
Taking the requirements of a comprehensive survey into account, we illustrate the architecture
of ECDriven-IoT from the different levels in detail and summarize recent research advances. The
main contributions of this survey are summarized as follows:
• We reveal three new challenges in ECDriven-IoT, including the heterogeneity of IoT and
edge computing, coordination between communication and computation, and more tricky
security and privacy issues.
• We categorize existing related studies from six aspects, i.e., system architecture, operating
system, communication protocol, computing paradigm, application, and security and pri-
vacy. It gives a whole view of the current advance of ECDriven-IoT and describes possible
solutions for addressing these challenges in each aspect.
• Finally, we conclude key lessons learned after reviewing existing related work and give sev-
eral insights into future research challenges and directions in ECDriven-IoT.
The rest of this survey is organized as follows. Section 2 introduces the background of IoT and
edge computing and compares cloud computing with edge computing. The taxonomy of ECDriven-
IoT is also proposed in this section. According to the taxonomy, various hardware architectures
of ECDriven-IoT are explored in Section 3. In Section 4, we present operating systems adopted in
IoT and edge computing, which play a significant role in ECDriven-IoT. Communication protocols
and computing technologies are discussed in Sections 5 and 6, respectively. Section 7 discusses
security and privacy concerns when deploying ECDriven-IoT systems in practice. ECDriven-IoT
applications are introduced in Section 8. Section 9 provides lessons learned, open challenges along
with future research directions. Finally, we make a conclusion in Section 10.

2 BACKGROUND
In this section, we introduce the background of ECDriven-IoT, which has two main components:
IoT and edge computing. We also compare the difference between edge computing and cloud com-
puting to explain why edge computing is more suitable for IoT than cloud computing. And then,
we give a taxonomy of ECDriven-IoT.

2.1 Internet of Things

In the IoT world, all objects that exist in reality can connect to the Internet and be accessed by
users. Through a specific addressing scheme, IoT devices can cooperate to complete the designated
work [15]. The main advantage of IoT is its great impact on people’s daily life and potential user
behavior [134]. On the one hand, for individual users, the benefit is reflected in areas such as

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:5

electronic health, smart home, and life support. On the other hand, for industry, IoT also plays an
active role in automation, logistics, and intelligent transportation.
IoT has attracted immense attention from the industry and academia [6, 185]. More and more
IoT applications are focusing on achieving real-time responses, such as Virtual Reality (VR) [52],
Augmented Reality (AR) [52], and automatic driving. The very short latency is non-negotiable
for these applications. In cloud computing, due to geographical distance and network fluctuation,
the latency is too high to satisfy real-time requirements. Besides, massive data deteriorate transmis-
sion performance. So, how to effectively allocate network bandwidth and computing resources is a
challenge [189]. In the IoT community, there are different data formats and communication proto-
cols, making IoT a vertically fragmented network system [190], which poses another challenge to
accomplish the desired low-latency performance. Furthermore, most IoT devices are power-limited,
and it is necessary to balance the power consumption to extend the lifetime of IoT devices.
Before IoT further deepens its impact on the world, many challenges are still worthy of attention
and research. One of the critical issues is how to achieve good interconnection and interoperabil-
ity among IoT devices, guarantee security demands, and provide a high level of intelligence. In
addition, IoT devices usually lack of computing power and energy capacity. Therefore, a new com-
putational paradigm should target the resource efficiency in addition to scalability issues. Edge
computing as a new computing paradigm could provide such help for IoT.

2.2 Edge Computing

Edge computing essentially migrates partial computing jobs from remote cloud servers to local
edge servers. It performs data preprocessing and analysis near the data sources. Since edge servers
are closer to data-generated devices, they can have a quicker response than cloud servers. On the
contrary, the advantage of cloud computing is providing global scheduling capability and powerful
computing resources. Similar to edge computing, fog computing (FC) is a highly virtualized plat-
form that offers computing resources, storage, and control between end-users and cloud servers,
proposed by Cisco in 2012 [24]. In this survey, we refer to edge computing and FC collectively as
edge computing.

2.2.1 Cloud Computing. Cloud computing is another significant change after large computer
to client-server transformation. Users can share software and hardware resources in cloud com-
puting [70]. The complex hardware structure in cloud systems is transparent to users. So, users
do not need the expertise or direct control of cloud servers. There have been many studies on the
cloud and IoT, namely, the CloudIoT paradigm [26]. They have thoroughly investigated the main
attributes, characteristics, basic concepts, and open issues of the CloudIoT paradigm. Table 2 shows
the connection and difference between cloud computing and edge computing. Edge computing is
essentially an edge optimization of cloud computing. Both of them are designed for handling big
data. However, the main difference is that data can be distributed and processed on the closer edge
servers in edge computing. Figure 1 also shows the difference in geographic distribution between
cloud computing and edge computing. Thus, edge computing is more suitable for real-time data
processing and secure intelligent analysis.
Many studies attempt to optimize cloud computing to suit IoT scenarios [27, 44]. For example,
Zhou et al. proposed an architecture named CloudThings, which is an approach to combine cloud
computing and IoT. This architecture is a cloud-oriented IoT approach, helping IaaS, PaaS, and SaaS
in developing and managing IoT applications [199]. Pacheco et al. proposed a privacy-protected
architecture for integrating cloud computing and IoT. This architecture presents a scheme for pro-
tecting data generated by IoT devices without a secure transport layer protocol [34]. However,
the requirement of real-time response, massive data throughput, and low power still constrain the

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:6 L. Kong et al.

Table 2. Comparison of Edge Computing and Cloud Computing

Edge Computing Cloud Computing

Computing Location Edge devices, Distributed Centralized big data centers
core network
Key Feature Close to data source, Edge Centralization
and core network
Network Components Terminal device, edge All basic network
device and IoT gateways, components, Data centers
Core network hardwares
Flexibility and Scalability High Low
Size Massive small nodes or Large
according to demands
Deployment Temporary deployment or Complex deployment
deployment with minimal
planning
Bandwidth Requirements A little and well balanced Long-haul network
bandwidth requirement

Fig. 1. Cloud, edge nodes, and edge devices.

application of cloud computing in IoT [144]. Thus, another solution is to propose a new computing
paradigm to solve these problems thoroughly.

2.2.2 Edge Computing. Edge computing conforms to the computing characteristics of mobile
devices in IoT. The core of this architecture is mobile edge computing (MEC). MEC is a new
concept that integrates IT and telecommunications, which adds functions such as computation,
storage, and processing to the wireless network side. It enables more and more mobile devices
to quickly and easily access IoT, such as wearable smart devices. Yaser et al. proposed a layered
model consisting of a MEC server and a Cloudlets infrastructure [79]. This architecture aims to
increase the coverage of mobile user signals. And it allows users to complete the services they
request with minimal cost in terms of power and response latency. The main goal of the MEC
solution is to export some cloud functions to the mobile network edge, increasing available band-
width and reducing latency. Unlike the general architectural model, mobile hardware architecture
is used more in communications, using multiple software-defined network (SDN) controllers
and virtualization to solve data processing in communications [153].

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:7

Fig. 2. Taxonomy of edge-computing-driven IoT.

2.3 Taxonomy of ECDriven-IoT

The taxonomy of the ECDriven-IoT is shown in Figure 2. We categorize relevant literature into six
parts: hardware architecture, operating system, communication protocol, computing layer, secu-
rity and privacy mechanism, and application. Existing research works are reviewed and grouped
into the above six parts according to their research focuses. These parts are discussed from bottom
to up according to the layer in ECDriven-IoT systems. Although these six parts belong to different
research areas, they work together to form a complete ECDriven-IoT system.
The first layer is the lowest-level hardware architecture layer, which focuses on IoT and edge
computing hardware. Research work includes general hardware architecture that typically con-
tains terminal things and edge network devices, mobile architecture suitable for mobile IoT sce-
narios, and scalable hierarchical architecture. These three different architectures are categorized
according to different deployment scenarios. The second layer, the operating system layer, mainly
concludes several well-known IoT operating systems, such as Huawei LiteOS and mbedOS. Those
IoT operating systems are widely used in practical applications. And they are generally lightweight,
making them suitable for application to edge devices. The third layer is the study of communi-
cation protocols, including both short-range ones and long-distance ones. An important area of
research in IoT is communication interaction between various devices. Especially in edge comput-
ing environments, more communication takes place among devices. These devices often use dif-
ferent communication protocols and require cross-protocol communication. Short-range commu-
nication protocols include famous Wi-Fi, Bluetooth, ZigBee, and 6LoPWAN [162]. Long distance
communication protocols include NB-IoT, LoRaWAN, and 5G. The fourth layer is the computing
layer, including computation offloading, IoT distributed computing, caching, software-defined
network (SDN), and network function virtualization (NFV). These new computational stud-
ies are currently not given sufficient attention, but they all have large development prospects and
can optimize edge computing in the IoT. The fifth layer is the security and privacy layer. It is
necessary to consider security and privacy factors when designing secure ECDriven-IoT systems.
We analyze possible threats and existing countermeasures in ECDriven-IoT. The final layer is the
application layer. We introduce several popular ECDriven-IoT applications, including urban smart
living, industrial applications, and optimization of the entire system.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:8 L. Kong et al.

Fig. 3. Edge-computing-driven IoT model architecture.

3 ARCHITECTURE OF EDGE-COMPUTING-DRIVEN IOT

As a complement to cloud computing, the IoT system will become more complicated when meeting
edge computing. Since the number of edge nodes is large and edge nodes distribute everywhere
in geography, one research point is how to manage and maintain the ECDriven-IoT architecture.
Figure 3 shows a typical ECDriven-IoT model architecture. The architecture of the ECDriven-IoT
needs to: (1) Manage devices. There is a variety of IoT devices, and the number of IoT devices
is relatively large. Thus, they should be managed efficiently to satisfy network bandwidth and
power consumption requirements. (2) Allocate resources. Edge computing nodes lie near IoT de-
vices, which can significantly reduce communication latency. However, the processing capabilities
of edge nodes are limited. The latency will be very high if application tasks are waiting for the node.
Thus, the architecture should efficiently allocate computing resources to IoT devices. (3) Discover
services. For edge nodes, they need to discover services and allocate computing and storage re-
sources. Thus, how to efficiently discover services at a low cost is another problem that needs to
consider in architecture. (4) Schedule power. IoT nodes are often power-limited, so the architecture
should be energy efficient to reduce power consumption.

3.1 General Hardware Architecture

Unlike cloud computing, edge computing complements and extends cloud computing to edges and
endpoints. Edge computing benefits from edge devices’ proximity to sensors while leveraging the
on-demand scalability of cloud resources [42]. The distributed infrastructure of the ECDriven-IoT
contains heterogeneous resources and manages the architecture in a distributed manner. There are
various participants in this distributed architecture, including data centers, network cores, network
edges, and endpoints. Thus, the architecture should be properly designed. Figure 4 shows a typical
three-layer architecture of the ECDriven-IoT.
IoT brings not only new entry points for big data analytics but also distributed data sources at
the network edge. Bonomi et al. introduced a general hardware architecture that meets the needs
of most IoT scenarios [23]. The general hardware architecture is characterized by a low-cost con-
figuration that is easy to maintain and meets the needs of IoT architecture in general, resulting in
significant benefits. Dautov et al. introduced a distributed hierarchical data fusion architecture for
IoT networks, consisting of edge devices, a network, communications units, and cloud platforms

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:9

Fig. 4. The three-layers architecture of ECDriven-IoT.

together [43]. Different data sources are combined at each level of the IoT hierarchy to produce
timely and accurate results by utilizing the computational capabilities of intermediate nodes.
EdgeX Foundry is another typical ECDriven-IoT architecture model [47], which creates an open-
source framework for IoT edge computing. The framework is completely independent of hardware
and operating systems, supports a plug-and-play component ecosystem, unifies the market, and
accelerates the deployment of IoT solutions. Also, it addresses critical interoperability challenges
for edge nodes and data normalization in a distributed IoT edge architecture [56].
Edge computing nodes deployed in various environments are heterogeneous, while general ar-
chitecture can implement cross-platform management of heterogeneous resources. In the architec-
ture, how to efficiently manage fog/edge computing infrastructure, allocate available resources to
IoT devices, and schedule fog/edge computing resources is of significant importance [100]. Li et al.
proposed an architecture called ECIoT and studied the management of radio resources and com-
puting resources in ECIoT [98]. ECIoT focuses on resource allocation and performance control.
To improve the performance of ECIoT, they use the Lyapunov stochastic optimization method to
maximize system efficiency. Kitagami et al. proposed a multi-agent-based flexible IoT edge com-
puting architecture to balance global optimization by a cloud and local optimization by edges
for optimizing the role of cloud servers and edge servers dynamically [88]. In Reference [99], a
multi-layer resource allocation scheme was proposed, and it can adapt to the characteristics of
resource-constrained nodes at edges.

3.2 Software Defined Hardware Architecture

With the increasing demands of users and broader network access, IoT applications, network de-
velopers, service providers, and network carriers have to provide up-to-date services to users. In
the same way, communication networks are expanding exponentially, leading to the whole sys-
tem consisting of many subnets. These subnets have different communication and routing proto-
cols. Integrating these heterogeneous subnets into a unified communications platform is a critical
technical challenge, especially in a dynamic environment. For the practical envisioning of edge
computing in IoT, there is a need for a simplified architecture that hides all the complexities of
communication and provides a simple implementation.
According to the definition, SDN refers to a network architecture where the forwarding state
in the data plane is managed by a remotely controlled plane decoupled from the former [89]. On
the one hand, SDN allows a clear separation of concerns between service in the control plane and
data plane, thus making the network architecture more easily handled. On the other hand, SDN
mechanisms aim to balance the degree of centralized control/coordination through an explicit
SDN controller and decentralized operations through flow-based routing and rescheduling within
network components. It is necessary to reduce collection overhead and guarantee data effective-
ness in the ECDriven-IoT system, which makes SDN very suitable for edge computing and IoT.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:10 L. Kong et al.

Salman et al. introduced a hardware architecture that integrates new technologies such as an
SDN and virtual network functions (VNF) [153]. This architecture is used to implement and
flexibly manage distributed edge networks, improve network scalability, and reduce costs. For
example, for a typical factory, services and workloads are more IT-centric (e.g., factory data cen-
ters), and as they move down, they become OT-centric (e.g., factory machines). Software-defined
resource allocation and management is gaining momentum in the edge computing paradigm as
it can enable plant operators to better adapt to future needs. From a network perspective, this
translates into an SDN implementing VNF throughout the plant.
Yaser et al. proposed a comprehensive framework model based on a software definition to sim-
plify IoT management process [77]. It abstracts all control and management operations from un-
derlying devices and places them in the middleware layer to hide the complexity of traditional
system architectures. It is a model for forwarding, storing, and protecting generated data from IoT
objects through integrated software, and is ideal for use in edge computing and edge network en-
vironments. Qin et al. designed a software-defined architecture by extending the Multi-Network
Information Architecture (MINA) [140]. MINA is middleware with a multi-layer IoT SDN con-
troller. The IoT SDN controller they developed supports a variety of scheduling commands. At
the same time, this architecture can optimize the IoT network environment by using genetic al-
gorithms. The architecture provides differentiated service quality for different IoT tasks across
heterogeneous wireless networks.

3.3 Hybrid Hardware Architecture

In addition to general and mobile architecture, hybrid architecture has attracted the attention of
many researchers. Sun et al. introduced a more flexible IoT architecture called edgeIoT, which
uses fog computing to collect data at network edge [170]. Specifically, each fog node provides
computing power and connects to base stations (BS). The SDN-based cellular is used for packet
forwarding between fog nodes and hierarchical calculations at each fog node.
Chang et al. also proposed a hybrid cloud architecture model, called Edge Cloud, designed to
provide low-latency, high-bandwidth efficiency utilization [33]. As the name suggests, Edge Cloud
combines edge networks with cloud data centers for data processing and vulnerable storage. Cloud
data centers host regular computing and database components. This architecture takes advantage
of edge and cloud computing to reduce latency and save bandwidth resources.
Munir et al. also proposed a similar edge computing architecture and designed a reconfigurable
layered fog node architecture that can be suitable for fog computing applications [120]. Different
from edgeIoT, the bottom-up abstraction of Munir’s architecture includes the application layer,
analysis layer, virtualization layer, and hardware layer. The hierarchical architecture facilitates
the abstraction and implementation of edge computing paradigms that are distributed in nature
and involve multiple vendors. This architecture analyzes the characteristics of applications and
reconfigures the fabric resources to maximize the mobile workloads of the service for satisfying
peak workload demands.

4 OPERATING SYSTEM
The IoT nodes usually connect to the Internet through communication protocols. Due to the het-
erogeneous nature of IoT, many different communication protocols are adopted in the system.
Moreover, there are many IoT devices, including mobile phones, sensors, and other hardware
platforms, such as Aurdriono [67], Raspberry [114], Intel, and Zolertia Z1 [53]. The operating
system can bridge all the differences between these devices and provide a unified application
programming interface. Considering the limited memory and power, traditional operating sys-
tems, such as the Linux and Berkeley software distribution (BSD), are not suitable for IoT
ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:11

Fig. 5. A typical IoT OS infrastructure.

devices. The general operating system in IoT is ascendant. Many companies and research insti-
tutes invest many resources in researching IoT operating systems. Current popular IoT operating
systems include LiteOS, Contiki, Win10IoT, FreeRTOS, and mbedOS [13]. Figure 5 shows a typical
IoT operating system infrastructure. In general, IoT operating system is supported by the ker-
nel, end-to-end communication, peripheral components (e.g., the file system, Java virtual machine,
XML file parser), industry framework, and integrated development environment.
Although IoT operating systems have been developed for several years, applying edge comput-
ing in IoT brings some new demands:
• Realize scalable kernel size. The core of the operating system should be able to adapt to
various configuration environments, from low-end embedded applications with small to tens
of kilobytes of memory to complex applications with up to tens of Minionbytes of memory.
• Satisfy real-time, high-reliability, and energy-saving requirements. Kernel should
also have some features of the general embedded operating system, such as predictable exter-
nal event response time, predictable interruption response time, control, and management
mechanisms for various external hardware.
• Shield the characteristics of IoT fragmentation and provide a unified programming
interface. Fragmentation refers to various hardware device configurations, and different
application areas vary widely. The “fragmentation” feature has constrained the development
and growth of IoT.
• Reduce the cost and time of application development. The IoT operating system is a
public business development platform with rich and complete IoT basic functional compo-
nents and application development environments, which can reduce the development time
and cost of applications.
There have been many surveys focusing on IoT operating systems [58, 80, 121]. However, all of
them neglect to discuss the applicability of IoT OS in edge computing. Thus, they cannot illustrate
the new features and requirements for operating systems in the ECDriven-IoT area. In this section,
we discuss challenges for the operating system area, in terms of architecture, real-time support,
networking technologies, and energy efficiency. Also, considering many IoT operating systems
have been proposed in the community, we only illustrate some of the most used IoT operating
systems in detail. We summarize the features of five popular IoT operating systems in Table 3.

4.1 Architecture of IoT Operating System

The operating system architecture can largely influence the kernel size of the system. Current
mainstream OS architectures can be categorized as monolithic, micro-kernel, virtual machine,
ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:12 L. Kong et al.

Table 3. Comparison of Operating Systems

Feature Contiki TinyOS LiteOS RETOS RIOT
Architecture Modular Monolithic Modular Modular Modular
Real-time
No No No Yes Yes
Support
FileSystem, LoRaWAN,
LoRaWAN,
Network, 6LoWPAN, FileSystem,
Communication FileSystem, Static/Dynamic,
6LoPWAN, IPv6, Network,
Support Network, Network
Command, Multi-hop Protocol 6LoWPAN,
6LoWPAN
Line Interface GraphicalUI
Power No power Deep
Most efficient Low Timer ticks
Consumption management sleep mode

or layered ones. The monolithic architecture embeds necessary OS components and applications
within its kernel, which could increase the kernel size and the difficulty of adding new features or
deleting old ones. The micro-kernel architecture provides minimum functions in the kernel. Thus,
applications and OSs are considered decoupled modules to make them easy to be added or be re-
moved. So, the extension of such architecture will be more flexible. Also, a small kernel size makes
the micro-kernel architecture more suitable for the ECDriven-IoT. Another type of OS architecture
is the virtual architecture, in which a virtual machine mimicking hardware is exported to user pro-
grams. As an improvement to early monolithic systems, this system architecture has modules as a
layer-based architecture. Each layer has different functionalities. However, a few IoT devices adopt
the virtual and layered architecture, so we mainly focus on the first two architectures.

4.1.1 Architecture of Contiki. With a modular architecture, Contiki can efficiently reduce the
size of the system, and multiple embedded OSs choose modular architecture due to its small size.
Contiki is an event-driven OS with multi-threading supports, thus providing optional threading
facilities for every process. Due to its customization, ease of extension, and better reliability, this OS
can serve as a memory, file, and time server [80]. In the ECDriven-IoT, it can meet the requirements
of heterogeneity.

4.1.2 Architecture of LiteOS. As a UNIX-like OS, LiteOS provides an abstraction of IoT devices.
LiteOS is also a modular architecture OS. To minimize the programming learning complexities,
it provides an efficient way and operating features, thus allowing user-friendly operations. The
main feature of LiteOS is that it provides a shell and a hierarchical file system. Moreover, LiteOS
has a much smaller code footprint, thus making it suitable for other platforms. The kernel of
LiteOS is a subsystem of the whole system architecture. Dynamical loading and multi-threading
are implemented in the kernel, thus providing concurrency supports [28]. However, LiteOS slows
down the program execution under the limitation of hardware and consumption power. Multiple
approaches have been proposed to solve this problem.

4.1.3 Architecture of RETOS. RETOS can solve various problems in IoT applications. RETOS
was developed with the aims of reconfiguration, vigorous activity, and efficiency of resources, so
it can efficiently deal with difficulties faced by IoT sensor nodes. As a modular system, RETOS
ensures efficiency and reliability through a dual model, operation, and code checking. Moreover,
RETOS can prevent hardware manipulation, memory access, kernel reading, and other dangerous
operations [29]. The RETOS performs well in the field of wireless sensor networks. However, when
applied in ECDriven-IoT, RETOS cannot satisfy other requirements, such as real-time response and
a unified programming interface.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:13

4.1.4 Architecture of RIOT. The ECDriven-IoT consists of billions of IoT devices and edge nodes.
These devices usually have small memory, low power consumption, and limited communication
bandwidth. Considering the requirements of real-time systems, the ECDriven-IoT needs a broader
vision to embed intelligence to smartphones and portables, to achieve the connection of every-
thing [179]. To save memory, RIOT adopts a micro-kernel architecture, so the size of its kernel is
minimized. Adopting multi-threading aims to be effective with energy, memory, modulator, and
APIs. Also, RIOT is a highly reliable OS, which is important in the ECDriven-IoT system.

4.2 Scheduling Algorithm and Real-time Support

In the ECDriven-IoT system, computing tasks are executed locally or offloaded to edge nodes.
For real-time applications, computing tasks need to be completed in a short time. Especially in
ECDriven-IoT, execution time becomes a critical metric because of transmission latency demands.
The scheduling algorithm determines the execution orders, how tasks are executed, and when
tasks are completed. A scheduler targets high throughput, high energy efficiency, fairness, and
good resource utilization.
Scheduling is of great importance for deciding the time interval of task execution. Real-time
scheduling algorithms aim to maximize throughput and complete tasks within the given time con-
straint [9]. In edge computing scenes, computing tasks can be divided into periodic and aperiodic
tasks. So, these tasks are scheduled with periodic and aperiodic schedulers, respectively [160]. The
operating system in the ECDriven-IoT is promising to handle real-time tasks efficiently. In the
following, we will explore the scheduling schemes of the typical IoT operating systems.
4.2.1 Scheduling Algorithm of Contiki. Contiki has a hybrid programming model. It is primarily
an event-driven OS but also supports multi-threads. As Contiki is event-driven, the processes will
run to completion. Contiki provides support for multi-threading, which is implemented as a library
on the top of the kernel. As for the real-time response, Contiki is mainly event-driven and does
not implement any scheduling algorithm. Applications are handled according to their priority [32].
Thus, Contiki is not suitable for ECDriven-IoT, as it does not support real-time capabilities [58].
4.2.2 Scheduling Algorithm of TinyOS. TinyOS has multiple scheduling techniques and algo-
rithms. In the event-driven model, a hardware interruption is handled by the event handler, and
it can cause preemption. The single task queue adopts the first-in-first-out (FIFO) strategy and
has no interruption for the FIFO algorithm, which is a non-preemptive algorithm. The new version
of TinyOS has a new feature, priority scheduling. Thus, tasks with a higher priority can interrupt
low-priority ones to meet their deadline demands. Moreover, TinyOS implements cooperative al-
gorithms, such as earliest-deadline-first (EDF) and adaptive double-ring scheduling (ADRS).
The system enables preemption to ensure that tasks with a higher priority are completed before
other tasks. However, preemption involves context saving and switching, which makes the sched-
uler more complicated and consumes more power. Thus, the preemption happens only in partic-
ular conditions. Owing to its efficient scheduling algorithm, TinyOS is considered one of the best
OSs for IoT platforms. However, the real-time requirements are more strict when applied to edge
computing. TinyOS still has many shortcomings in terms of real-time applications.
4.2.3 Scheduling Algorithm of RETOS. RETOS provides high concurrency with preemption
functions. As a multi-thread OS, RETOS implements the boosting thread scheduler and introduces
event-aware thread scheduling, which boosts the priority of threads [80]. To support real-time
applications, RETOS enables developers to assign task priority explicitly and provides kernel dy-
namic priority management. Thus, RETOS can satisfy the latency requirement in the ECDriven-IoT
system.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:14 L. Kong et al.

4.2.4 Scheduling Algorithm of RIOT. Using a scheduler through fixed priority and preemption,
RIOT allows for soft real-time capabilities [71]. RIOT can handle low-priority tasks to deal with
high-priority applications. RIOT applies a simple principle to achieve real-time scheduling: When
a high-priority thread arrives, threads with low priority will be preempted, and the high-priority
task runs right now until finished. What is more, RIOT can minimize response latency and power
consumption by mimicking the parallel execution of events with the same priority. At the same
time, it brings no context switches fee.

4.3 Networking Technologies in Operating System

In the ECDriven-IoT system, IoT devices offload collected data to edge nodes or cloud data cen-
ters. Thus, the connectivity between IoT devices and edge nodes is a fundamental guarantee to
transport data. The essential elements of these devices are the device, local network, and the Inter-
net. Communication technologies in the IoT community vary from device to device, and we will
discuss them in the next section in detail. For operating systems, the network stack will hugely in-
fluence the performance of applications. Thus, operating systems should consider heterogeneous
communication protocols.

4.3.1 Networking Stack for Contiki. Contiki supports not only a full TCP/IP stack but also a
lightweight stack for low-power radio communication. Contiki implements μIP, the first stan-
dalone stack [11]. μIP supports IPv4 and IPv6 with a limited memory, which is suitable for the
ECDriven-IoT. It can communicate with both the lightweight stack and full-stack, and its peers do
not need to have a complete protocol stack. Also, Contiki applies Rime [46], so Contiki supports
low radio communication and various communication modes. The module of Rime employs simple
functions, making the stack lightweight and suitable for IoT. However, Contiki does not support
as many communication protocols as the system requires when it refers to edge computing.

4.3.2 Networking Stack for TinyOS. In the ECDriven-IoT, both IoT devices and edge nodes are of
limited energy and memory, and these nodes are connected to the Internet and communicate with
each other. Therefore, we need an operating system that can provide stable communication links
between devices. TinyOS adopts a protocol that can be used for the transport layer, networking
layer, and medium access control layer. Thus it can consume as fewer layers as possible and make
TinyOS reliable and robust [91]. Moreover, TinyOS can be suitable for various applications, so it
is a good choice in the ECDriven-IoT.

4.3.3 Networking Stack for LiteOS. In LiteOS, MAC and communication protocols are taken as
threads or files. Hence, it provides flexibility for different communication protocols. These proto-
cols can be loaded dynamically as applications. During the communication, data packets will be
sent to the port where the protocol is listening [28]. This feature makes it very suitable for IoT
applications that vary in the communication protocol, but it has difficulty satisfying the latency
requirements for treating the protocol as an application.

4.3.4 Networking Stack for RETOS. RETOS divides the kernel into static and dynamic parts for
adapting to resource-constrained hardware environments. This design enables an easy program-
ming interface for application developers. The static kernel part is optimized at the device driver
level and guarantees the kernel performance in transmitting data packets and maintaining network
connectivity. The dynamic kernel part is similar to LiteOS and implemented as loaded modules. So,
different routing and communication protocols can be managed as the dynamic part and applied
to different applications [31].

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:15

4.4 Power Consumption

The energy efficiency of the operating system in ECDriven-IoT is an essential requirement, as
most IoT nodes in the system are power-limited. However, all nodes should communicate with
other nodes, which is a process that consumes energy. Hence, a device should consume as little
energy as possible [41, 175]. The heavy research area is based on energy-efficient protocols [30].
4.4.1 Power Consumption of Contiki. Since Contiki’s kernel does not embed any power man-
agement algorithm, the power management strategies are customized by application developers.
Contiki provides an interface to applications and allows them to manage the power system. In the
ECDriven-IoT, the IoT nodes need power management schemes to meet the lower-power limits. As
an event-driven OS, Contiki wakes up to respond to an interruption, and the poll handlers handle
these events. In this manner, power management schemes must be designed to reduce the overall
power consumption. When using Contiki, programmers must pay attention to power management
to achieve the high energy efficiency of applications.
4.4.2 Power Consumption of TinyOS. Different techniques have been incorporated into TinyOS
to achieve minimum power utilization. TinyOS with software thread integration is a method in
which energy is conserved in TinyOS. By integrating software threads, TinyOS makes full use of
idle time during transmission, processing, and sensing of data [139]. In TinyOS, which supports
high-power listening (HPL), TinyOS estimates the overall load of the sensing nodes and then
dynamically allocates the required energy to the sensing nodes [93]. This method can only be pos-
sible with an accurate estimation of energy consumption in sensing nodes. Sensing nodes consume
energy in a variety of ways [3]. TinyOS, in this case, is the most efficient OS, because it estimates
the energy consumption by the sensing nodes, TinyOS itself, and its components. TinyOS supports
various methods for estimating the energy consumption of different applications.
4.4.3 Power Consumption of LiteOS. LiteOS is a multi-threaded operating system, and while it
does not introduce any overhead, it consumes more energy than TinyOS. However, LiteOS has a
small memory footprint, so it can reduce energy consumption to a minimum, making it suitable
for the ECDriven-IoT to some degree.
4.4.4 Power Consumption of RETOS. RETOS is a multi-threading operating system, so it adopts
many scheduling operations. Threads scheduling involves significant context switching. All these
operations are energy exhaustive. To address these issues, RETOS adopts a variable time tick, and
timer requests are scheduled according to the remaining tasks, which can effectively minimize
energy consumption.

5 COMMUNICATION PROTOCOL
The IoT sensing layer collects sensing data (e.g., sound, light, electricity) through sensors. Based
on the terminal module of the network layer, base stations are connected to the network layer to
realize data transmission after data acquisition. The network layer is responsible for transmitting
data collected by the sensing layer. It should use different communication technologies based on
specific scenario characteristics. The application layer can be viewed as the data and business
platform of the IoT. As the collection point of all IoT terminal data, the data platform is responsible
for unified data storage and analysis.
The communication protocol at the network layer is a group of competitors. It is also the focus
of this section. In the ECDriven-IoT system, each terminal device and edge network device can
be regarded as an independent individual. The communication between such independent com-
ponents has the characteristics of hardware heterogeneity, low power, and short communication

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:16 L. Kong et al.

Fig. 6. Various wireless network communication protocols.

time. These features pose a great challenge in selecting and designing communication protocols.
IoT network layer communication protocols can be divided into short-range and long-distance
communication protocols. Short-range communication protocols include Wi-Fi [72], Bluetooth
technology [66], ZigBee [50], and UWB [8]. Long-distance communication protocols include NB-
IoT [5], LoRaWAN [165], and 5G [14]. Figure 6 shows the rate and coverage of various wireless
network communication protocols.

5.1 Short-range Communication Protocols

We consider placing all computing processes in edge devices and networks as much as possible in
an IoT environment. Therefore, the inter-communication process between edge devices and edge
networks in computing is particularly critical. The short-range communication protocols have sev-
eral advantages and disadvantages, each of which applies to different IoT environments. Therefore,
according to specific communication environments and requirements, it is worthwhile to study
and improve the communication protocols so that the calculation process can be more efficient.
The most popular Wi-Fi technology has a fast transmission speed. However, with the speed
increase, the power consumption also increases sharply, and then the transmission distance be-
comes a bottleneck. Long-distance transmission requires an access point (AP) to bridge the data
link as a middleman, which will largely increase the cost. Therefore, Wi-Fi technology is more suit-
able for indoor wireless Internet access scenarios and terminal applications such as PCs and PDAs.
Both Bluetooth and Wi-Fi work on the 2.4 GHz band, so there are some interference problems in
the same frequency band. Bluetooth consumes slightly less power than Wi-Fi, and the transmis-
sion speed is far lower than Wi-Fi. It is widely used in asset tracking, location tags, and medical
sensors, such as smartwatches and Bluetooth positioning. ZigBee technology has relatively low
power consumption and short communication distance. It is mainly used in wireless sensors and
medical scenes. UWB technology has a relatively clean frequency band and no interference from
other communication technologies. It is currently used in high-precision positioning scenarios.
These popular short-range communication protocols can be well applied in the ECDriven-IoT as
the communication basis between edge devices.
5.1.1 Wi-Fi. Vivek et al. adopted Wi-Fi and ZigBee to implement a home automation sys-
tem [178]. With the help of light, temperature, and safety feedback loops, the system pro-
vides comfortable brightness, temperature regulation, and basic safety by utilizing popular Wi-Fi
signals provided by IoT devices, such as smart air conditioners, smart lights, and thermostats. Shi
et al. used Wi-Fi signals to capture the behavioral characteristics of daily human activities in their
paper [161]. This method does not require hardware devices and only needs to recognize users’
unique physiological and behavioral characteristics through the Wi-Fi signal, thereby realizing

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:17

some functions such as user authentication. Meanwhile, Acer et al. utilized Wi-Fi aware network
search to analyze IoT data [4].
Wi-Fi technology’s data rate is fast enough for ECDriven-IoT applications. However, as trans-
mission speed increases, the power consumption of devices also increases dramatically. In edge
devices, energy saving is a critical factor. So, Wi-Fi may not adapt to some scenarios well in
ECDriven-IoT systems.

5.1.2 Bluetooth. In recent years, much of the work on wireless sensor networks targets to be
efficient, low cost, scalable, and easy to deploy. Optimizing battery usage and power consump-
tion reduces costs and extends sensor life. Bluetooth is an ideal communication protocol for the
ECDriven-IoT. Its low power consumption makes edge devices run for a long time and reduces
maintenance. Generally speaking, edge computing does not require high data transmission speed.
Nair et al. introduced an architecture that uses the Bluetooth low energy (BLE) communi-
cation standard and hybrid topologies to reduce the power consumption of communication sys-
tems [122]. The BLE is considered a low-power version of traditional Bluetooth. However, the
extensive use of BLE in deployments can lead to high collision rates, especially in device-intensive
IoT environments. To alleviate this contradiction, Harris et al. proposed opportunistic listening,
an extension of the BLE active mode with tags and scanning devices [69].
For smart cars that use Bluetooth technology, users can connect their smartphones with their
cars. In this case, they can replace the phone’s speaker and microphone with the car’s ones, and
use the car’s devices to make a call or message. At the same time, you can also use your mobile
phone to read diagnostic data about your body everywhere [125].

5.1.3 ZigBee. The smart home is an IoT application closely related to human life. ZigBee is a
widely-used communication protocol in the smart home. At the same time, ZigBee is also the ideal
communication protocol for the ECDriven-IoT. Because of its low power consumption, ZigBee can
be suitable for IoT environments that include massive wireless sensors.
Moravcevic et al. proposed a way to integrate the ZigBee protocol into smart homes [116]. This
approach firstly defines a home device as a service that can add ZigBee devices from different
manufacturers to the system. Various home devices on the market today can communicate using
the ZigBee protocol. So, energy-efficient devices that support ZigBee can be added to the smart
home system.
In addition to smart home applications, ZigBee can combine with other new technologies to
control and communicate between IoT devices. Ferreira et al. proposed a model combining event
capture and device control [55]. This model is implemented using basic general techniques such as
RESTful API or UPnP. With ZigBee communication technology, this easy-to-capture body interac-
tion allows developers to make fun and useful applications. Another important aspect of this tech-
nology is the data exchange between various types of endpoints by using standardized communica-
tion protocols. It allows a wide variety of programs to utilize data exchange to achieve specific user
needs, even though these programs are independently developed by developers around the world.
Meanwhile, the privacy and security of IoT in the communication protocol layer have also re-
ceived the attention of researchers. Ronen et al. discovered a new type of worm threat [147]. When
there are too many IoT devices, and the density exceeds a certain amount, the worm will spread
rapidly in a large area of the communication layer, and adjacent IoT devices will infect each other.
They verified the infection with the help of the Philips Hue Smart Light platform. They use Zig-
Bee only as their wireless communication technology and found that this worm threat can be
transmitted directly between various types of adjacent light bulbs, which are light bulbs that can
communicate with each other on the same platform. The contagious nature of this attack can cause
city lights compromised on a large scale in a matter of minutes.
ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:18 L. Kong et al.

5.1.4 UWB. Ultra-wideband (UWB) is a communication technology that uses a non-

sinusoidal narrow pulse of nanoseconds to microseconds to transmit data. UWB was used in early
applications of short-distance high-speed data transmission. In recent years, many researchers
have begun to use their sub-nanosecond ultra-narrow pulses for short-range accurate indoor po-
sitioning. The UWB architecture finds its place in surveillance systems, medical applications, and
IoT applications. Antennas using UWB communication technology have compact hardware and
low power consumption. This feature is critical for portable IoT devices. Bekasiewicz et al. de-
scribed this UWB antenna structure for IoT [21]. This well-designed structure enables small-sized
physical areas while maintaining electrical performance. What sets UWB apart from other short-
range communication protocols is that its frequency band is relatively clean. It is usually used for
precise positioning, but the application scenario is not rich enough.
5.1.5 6LoPWAN. When we first considered the sensor communication network, the first thing
that came to mind was the use of Internet Protocol (IP). IP is unsuitable for sensor or personal
area networks, because it is too heavy for these applications. Recently, more and more research
has started to work on low data rates, low power consumption, and small-size IP protocols. The
IPv6 low-power wireless personal area network (6LoWPAN) is a low-speed wireless network
standard. It supports the use of IP in IEEE 802.15.4 wireless networks [76]. The key point in the
breakthrough of 6LoWPAN-related work is to achieve a very compact and efficient IP, eliminating
the communication difficulties brought by the unique protocol standard. The 6LoWPAN protocol
has the following features:
• Popularity. IP networks have been adopted widely. IPv6, the core technology of the next-
generation Internet, is also accelerating its popularity. It is more acceptable to use IPv6 in
low-speed wireless personal area networks.
• Applicability. The IP network protocol stack architecture is widely recognized, and the
low-speed wireless personal area network can be developed simply and efficiently based on
this architecture.
• Adequate address space. When IPv6 is applied to low-speed wireless personal area net-
works, the biggest highlight is the large address space, which is precisely needed for deploy-
ing large-scale, low-speed wireless personal area network equipment.
• Stateless automatic address configuration. When a node boots up in IPv6, it can auto-
matically obtain a MAC address and configure an IPv6 address. This feature is attractive for
sensor networks, because it is not feasible to configure the user interface for sensor nodes
in most cases, and nodes must have automatic configuration capabilities.
• Easy access. Low-speed wireless personal area networks use IPv6 technology to make them
easier to access other IP-based networks and next-generation Internet, enabling them to take
full advantage of IP network technologies.
These features are ideal for ECDriven-IoT communication environments, especially for IoT ap-
plications that require large-scale deployment of low-power communication devices. Mulligan et al.
introduced a simple 6LoWPAN protocol architecture and compared it with ZigBee [119]. Ma et al.
introduced the advantages of 6LoWPAN and details of some key technologies [103].
The short-distance communication protocols in IoT mainly include Wi-Fi, Bluetooth, ZigBee,
UWB, and 6LoPWAN. Table 4 shows the main features and differences between them.
5.1.6 Cross Technology Communication. Because of the hardware complexity and network het-
erogeneity of the ECDriven-IoT, cross-technology communication (CTC) is becoming a popu-
lar research direction. Considering the IoT environment of densely deployed devices, mainstream
wireless technologies typically share radio spectrums. Wireless technologies shared spectrums

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:19

Table 4. Comparison of Wi-Fi, Bluetooth, ZigBee, UWB, and 6LoPWAN

Wi-Fi Bluetooth ZigBee UWB 6LoPWAN
Data Rate 1 Gbps or more 1 Mbps 100 Kbps 53–480 Mbps 250 Kbps/40 Kbps
Communication
20–200 m 20–200 m 2–20 m 0.2–40 m 10–100 m
Distance
Frequency
2.4 GHz/5.8 GHz 2.4 GHz 2.4 GHz 3.1 GHz/10.6 GHz 2.4 GHz/915 MHz
Band
Security Low High Medium High High
Power
High Medium Low High Medium
Consumption
Cost High Low Medium High Low
Mobile phone
PC, PDA wireless Wireless Accurate
Application transmission, Smart home
Internet access sensing locating
Medical health

Table 5. Comparison of NB-IoT, LoRaWAN, and 5G

Spectrum Module Coverage Battery Data Rate Flexibility
Cost Cost Performance
NB-IoT Authorized, ≤20 dollars 18–21 km Fast power 200 Kbps Limited by
high cost consumption operator
LoRaWAN Unauthorized, ≤10 dollars 12–15 km Long lasting 0.2–37.5 Kbps Self-built
low cost electricity network
5G Authorized, ∼200 dollars 10–100 m Long Life Time 10 Gbit/s Base-
high cost station
provide

will inevitably interfere with each other. But every coin has two sides. This drawback also makes
cross-protocol communication possible.
For example, Zhou et al. introduced a cross-technology communication protocol, ZiFi [200]. The
system uses ZigBee radios to identify the presence of Wi-Fi networks through the unique interfer-
ence characteristics generated by Wi-Fi beacons, which can significantly improve the standby en-
ergy efficiency of Wi-Fi devices. Kim et al. introduced FreeBee [87], which supports three popular
wireless technologies (Wi-Fi, ZigBee, and Bluetooth) across technology broadcasts. FreeBee’s core
idea is to modulate symbolic messages by changing the timing of three standard beacon frames
without additional frames and traffic.

5.2 Long-distance Communication Protocols

In a long-distance scenario, if terminal devices cannot solve the power supply problem, a technol-
ogy with lower power consumption and broader coverage is needed to meet the requirements of
IoT communication. Thus, driven by business and technology, some researchers and enterprises
have developed a new type of communication technology, LPWAN, a low-power WAN technol-
ogy [165]. Long-distance and low-power communications have a broader application prospect
in future IoT environments, because not all IoT and edge devices are in close proximity. Thus,
LPWAN is more suitable for machine-to-machine (M2M) communication in edge computing.
Table 5 shows the main features and differences between these three long-distance communica-
tion technologies.
LPWAN is a long-range wireless network communication technology that has been widely
used to optimize M2M communications in IoT applications. The main technical advantages of

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:20 L. Kong et al.

LPWAN are ultra-low power consumption, long-distance, low throughput, and broad coverage.
Some typical applications include urban coverage, remote meter reading, manhole cover testing,
and offshore fishing vessel testing. Long-distance communication protocols are also often used for
IoT communication for specific remote edge devices.
5.2.1 NB-IoT. NB-IoT is called narrowband IoT and can be deployed directly on LTE networks.
Good compatibility reduces the cost of deployment. It has lower power consumption. Theoretically,
the terminal module carrying NB-IoT uses a battery and has a standby time of up to 10 years. The
reduction in module costs has also led more companies in the market to use this technology. In
3GPP, an LTE-based narrowband system has been introduced to support the IoT [22].
Mangalvedhe et al. [107] introduced the NB-IoT system design, some potential problems, and
solutions for the actual deployment system. Adhikary et al. focus on the coverage of NB-IoT in
IoT environments [5]. They believe that NB-IoT provides broader coverage than traditional LTE
systems. Petrov et al. proved the possibility of applying NB-IoT to IoT cars [136]. They conducted a
comprehensive system-level assessment revealing the impact of in-vehicle NB-IoT communication
on critical metrics, such as reliability, transmission delay, and energy efficiency. The results show
that the development potential of NB-IoT may meet the future performance requirements of IoT
vehicles.
The NB-IoT has four features: (1) Wide coverage area. NB-IoT provides better indoor coverage.
(2) Strong connectivity ability. NB-IoT supports more than 100,000 connections in a single
workspace. (3) Low power consumption. Usually, the standby time of the NB-IoT terminal device
can last for several years. (4) Low cost. The NB-IoT license band can be deployed in-band, guard
band or independent carrier mode to coexist with existing networks. Therefore, NB-IoT can be
widely used in various related industries, such as intelligent remote meter reading, asset tracking,
intelligent parking, and intelligent mechanized agriculture.
5.2.2 LoRaWAN. LoRaWAN is a long-distance communication protocol different from NB-IoT.
It is an ultra-long-range wireless transmission technology based on chirp spread spectrum technol-
ogy promoted and adopted by Semtech. At the most basic level, wireless protocols like LoRaWAN
are relatively simple. LoRaWAN is a star topology [132]. This type of structure is generally bet-
ter than a mesh network, because it has advantages in maintaining battery power and increasing
communication range.
Many researchers have studied the performance and metrics of systems using this protocol. Pet-
ric et al. used the LoRa FABIAN protocol stack to generate and then observe the traffic between
IoT nodes and LoRa stations to perform the test [135]. In addition to long working life and low
production costs, coverage is a key performance indicator for long-distance communication proto-
cols. Petajajarvi et al. studied the coverage of LoRaWAN technology through actual measurement
work [137]. Bor et al. developed a platform for LoRa performance evaluation and described a pro-
tocol that leverages LoRa’s unique features on top of LoRa’s physical layer [25]. This protocol
enables energy-efficient wide-area multi-hop data collection.
Because of the similar name, many people confuse LoRaWAN with LoRa. However, LoRaWAN
refers to the networking protocol of the MAC layer, and LoRa is just a protocol for the physical
layer. From the perspective of network layering, LoRaWAN can use any physical layer protocol,
and LoRa can also be used as the physical layer of other networking technologies. Several technolo-
gies that compete with LoRaWAN also use LoRa at the physical layer. LoRa is one of the LPWAN
communication technologies and a long-distance communication solution based on chirp spread
spectrum technology. This solution changes the previous trade-offs between transmission distance
and power consumption to provide users with a simple system that can achieve long-distance, long
battery life, and large capacity, thereby expanding the network.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:21

5.2.3 5G. With many emerging scenes, such as autonomous driving, and smart cities, requir-
ing higher data rates, the fifth-generation (5G) cellular network has arisen with a high data
rate and broad communication areas. In ECDriven-IoT, the low latency and high energy efficiency
requirements lead to smaller transmission time intervals. Moreover, small cells can achieve high
area capacity in densification. All of these have led to new radio access technologies and a new
core network [158]. With the help of higher frequencies, large-scale antennae can be deployed at
base stations. Thus array gains can overcome the shortage of higher path loss and can gain spatial
multiplexing [94].
With a resilient cloud-native core network and end-to-end support for network slicing, 5G
is distinguished by high flexibility and scalable network technology. Based on three major user
case domains, 5G can support deterministic and isochronous communication with high reliability
and availability. 5G can be applied in the ECDriven-IoT with hard guarantees for latency bounds,
packet loss, and reliability, as well as synchronization down to the nanosecond level [59]. More-
over, the seamless change of the application server can be supported by 5G with low latency. 5G
application enablers will be studied for interactions between users, application servers, and the
network in a complementary manner [82].

6 COMPUTING
We are not only concerned with the underlying hardware and communication protocols but also
computational processes in IoT and edge networks. These research areas include algorithmic ac-
celeration for different scenarios, distributed computing in IoT [48], green computing [92], and
caching [183], as well as SDN and NFV. These new computational studies lack sufficient focus, but
they all have huge development prospects and can optimize edge computing in IoT.

6.1 Computation Offloading

The ECDriven-IoT has many hardware and protocol problems, so the computing capability of
edge nodes is limited and how to compute efficiently is still a tricky challenge in the ECDriven-
IoT [90, 104]. Some innovative algorithms have been proposed to overcome the limitations of the
ECDriven-IoT.
Data can be processed and pruned in edge nodes before being transmitted to the cloud through
intelligent gateways. However, considering edge nodes have limited computing capabilities and
energy power, only a part of the data can be processed locally. In computation offloading, it refers
to the offloading decision [39], server selection, Wireless resource allocation, transmission power
setting, computation resource allocation, and the slot partition. Aazam et al. expanded the integra-
tion of IoT and cloud computing. They analyzed the network architecture and performance of this
concept [1].
Chen et al. proposed a game-theoretic approach to achieve efficient computation offloading
for edge computing and formulate the distributed computation offloading decision-making prob-
lem among end devices as a multi-user computation offloading game [37, 38]. Mao et al. in-
vestigated a green edge computing system with energy harvesting devices and developed a
computation offloading strategy that jointly decides the offloading decision, the CPU-cycle fre-
quencies for mobile execution, and the transmit power for computation offloading [109]. More-
over, research has been done on resource allocation for a multi-user system based on time-
division multiple access (TDMA) and orthogonal frequency-division multiple access
(OFDMA). The optimal resource allocation is formulated as a convex optimization problem for
minimizing the weighted-sum mobile energy consumption under the constraint of computation
latency [193].

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:22 L. Kong et al.

6.2 Distributed Computing

As IoT and edge computing both have the feature of being distributed everywhere, how to organize
system resources and combine them is another difficulty in computing in the ECDriven-IoT. In this
context, distributed computing provides an opportunity to solve this problem efficiently.
Distributed computing is the process of aggregating the power of several computing entities
that are logically distributed and may even be distributed in geography, to collaboratively run a
single computational task transparently and coherently, so that they appear as a single, centralized
system.
Chien et al. introduced the idea of distributed computing in IoT [40]. They proposed a distributed
smart camera architecture used in video sensor networks to accelerate computer vision algorithms
for smart cameras in the IoT. Similarly, edge computing is also a distributed computing method
in IoT. Hesham et al. realized this vision. They used edge computing in a distributed computing
environment to move workloads from a centralized cloud to the network edge while verifying
edge efficiency and resourcefulness [48]. Distributed computing is a cheap and efficient alternative
computing method. It can compute in any location, so it can efficiently make use of the computing
capability of edge nodes and reduce the transmission bandwidth requirements, which can push
the development of the ECDriven-IoT.

6.3 Caching
In computation offloading and distributed computing, massive data are generated and transmitted
among edge nodes, which influences communication latency and IoT devices’ power. Caching is an
effective method to increase computing speed and save computing bandwidth. In the ECDriven-
IoT scenes, some applications require few computing resources and storage in IoT devices and
edge nodes. Thus, these remaining caches can be utilized more efficiently, reducing latency and
improving system efficiency. The idea of caching is transplanted to edge computing to reduce the
data transmission cost and system delay. At the same time, it will make design and development
more difficult.
Combining caching and edge computing with IoT is a promising means of alleviating traffic in a
backhaul. With network stability taken into account, Du et al. formulated a stochastic optimization
problem to jointly optimize the offloading decision and cache decision making [45]. Xia et al. [186]
investigated a cache-aided mobile edge computing network, where the source offloads the compu-
tation task to multiple destinations having computation capacity with the help of a cache-aided
relay. However, their work does not explore the cache-aid with IoT. So when applied to IoT, how
to efficiently solve the complexity of IoT and edge computing is still a challenge.
Distributed caching is widely used in the caching deployment of base stations. However, the
caching capacity of a single BS is generally particularly limited, which will degrade the perfor-
mance of the wireless mobile network [154]. Li et al. designed a collaborative cache scheme in the
heterogeneous mobile edge computing network, in which the edge caching of macro base stations
and small base stations are utilized to bring storage resources closer to users [96]. For most IoT de-
vices, the smaller the cache size, the more complex the topology. Therefore, more research should
explore the combination of edge computing and IoT in caching to make it a more efficient way to
reduce latency and energy consumption.
Another important research direction for cache computing is cached content placement [128].
In general, cached content placement system information is updated continuously to improve the
cache hit ratio. In an edge computing environment, caching typically occurs on user devices. The
cache at the user device may allow the user to download the requested content in a more effi-
cient manner using device-to-device (D2D) communication [184]. As the density of edge devices
increases, the cache advantages of user devices will be reflected in their low deployment costs.
ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:23

6.4 Software Defined Network and Network Function Virtualization

With the development of the ECDriven-IoT, the number of users has grown exponentially. Net-
work developers, service providers, and network carriers provide up-to-date services to the users.
Thus, the network is expanding exponentially in the same way and how to efficiently operate these
networks is complex. Thus, SDN has emerged as efficient network deployment and management
solution [16]. SDN provides a separation between the control plane and the data plane, which
equips network developers with the ability to efficiently expand the network and the convenience
to manage network resources. Furthermore, network operators can configure, upgrade, and main-
tain network resources dynamically. Since SDN is logically defined, the controller can access these
network resources more efficiently [195]. In edge computing, SDN can provide flexibility and man-
ageability. For the ECDriven-IoT, the data generated and collected in IoT devices need to be routed
to the edge or cloud. SDN can alleviate these complex communication requirements with service
discovery, provisioning, and orchestration at the edge nodes.
Network function virtualization (NFV) deals with the hardware-oriented function transfor-
mation, such as firewalls or DNS for the software applications [111]. It can provide dynamic service
orchestration. Thus, efficient services deployment can be realized without hardware support and
achieve the service function chain (SFC) [86]. Due to the heterogeneous nature, ECDriven-IoT,
SDN, and NFV can be integrated into a whole system and interact with each other. NFV can operate
as a service orchestrator, and SDN can automate the service chaining by installing customized flow
rules at the forwarding stations. This system can improve performance in real-time applications
and reduce transmission delays, which are significant metrics in these application scenes [81].

7 SECURITY AND PRIVACY ISSUES IN ECDRIVEN-IOT

When applying edge computing to IoT, new and unforeseen security and privacy problems will
arise. Due to the high mobility and heterogeneous features of ECDriven-IoT, the system is more
vulnerable to potentially malicious activities. In addition, many advanced security mechanisms
cannot be transplanted to IoT devices and edge nodes owing to the limited computing capabilities
and power. In the ECDriven-IoT system, the communication between IoT devices and edge nodes
is relatively frequent, thus making the network more unstable. In terms of privacy, many users’
privacy-sensitive information will be stored in IoT devices and edge nodes, or transported to the
cloud server. In such a distributed architecture, security and privacy become crucial challenges.
This architecture is more vulnerable to attacks and threats. In the communication, computation,
and storage process, malicious attacks will be encountered [127]. Figure 7 shows possible security
and privacy attacks and their solutions in ECDriven-IoT.

7.1 Security and Privacy Threats

In this section, we will illustrate potential attacks in the ECDriven-IoT system. Different kinds of
threats of ECDriven-IoT networks will be introduced, as well as their sources at different levels.
Owing to the features of IoT and edge computing, as well as the application scenes, the ECDriven-
IoT faces many security and privacy threats, such as distributed denial of service (DDoS) at-
tacks, physical attacks, eavesdropping or sniffing, and privacy leakage [101, 146].

7.1.1 DDoS Attacks. DDoS attacks toward edge-computing nodes consist of outage attacks,
sleep deprivation, and battery draining. In outage attacks, edge nodes will be exposed to unautho-
rized users and unable to perform in the designed way [117]. As a much harder-to-detect attack,
sleep deprivation adversaries overwhelm edge-computing nodes with an undesired set of legiti-
mate requests. Battery-draining attackers will deplete the battery of the edge-computing nodes or

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:24 L. Kong et al.

Fig. 7. Security and privacy threats and solutions in ECDriven-IoT.

IoT sensors/devices. DDoS attacks can also occur at the communication layer with continuous or
intermittent jamming [12].
7.1.2 Eavesdropping or Sniffing. In eavesdropping, adversaries can listen over communication
links to acquire private information, thus leading to privacy concerns. Through this attack, at-
tackers can take much important information about the system, such as user names, personal
information, or some commercial secrets [146].
7.1.3 Jamming Attacks. Jamming attacks are a kind of energy-consumption denial-of-service
attack. They can be launched in the link or physical layer. These attacks often utilize the weakness
of IoT systems and edge computing architectures. In jamming attacks, adversaries intentionally
flood the network with forged messages to exhaust the systems’ communication bandwidth, com-
puting sources, and storage volumes, making the whole system unable to carry out tasks [169].
7.1.4 Malicious Hardware/Software Injection. Attackers can inject malicious inputs into the
edge-computing node servers and perform hacking by adding unauthorized software or hardware
components to the communication between IoT devices and edge-computing nodes. This attack
can also make adversaries acquire many unauthorized data, thus raising privacy concerns [12].
7.1.5 Unauthorized Control Access. In the ECDriven-IoT paradigm, edge computing and IoT
nodes communicate with each other to access or share their data. However, these devices and
nodes can not use complicated methods to authorize permission access. Attackers can access one
of the unsecured edge nodes and possibly control the whole system, which is of great danger.
7.1.6 Privacy Leakage. As for privacy, the ECDriven-IoT can be applied to many personal
scenes, such as healthcare and smart homes. Thus, such personal information will be collected
by IoT devices and then transmitted to edge nodes to be processed and stored. However, consid-
ering the limited self-protecting ability of edge computing nodes and IoT devices, privacy leakage
potential exists in the ECDriven-IoT system. Furthermore, the ECDriven-IoT system will acquire

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:25

position information to serve the users, so attackers can obtain users’ physical position or other
sensitive information if they compromise the devices. Thus, how to guarantee data privacy in the
process from collection to storage is a crucial problem [12].
7.1.7 Other Attacks. The ECDriven-IoT is an emerging paradigm that combines heterogeneous
resources and devices. Thus the system is vulnerable to many attacks from different levels. Be-
yond the attacks mentioned above, there still are non-network side-channel attacks [117], routing
information attacks [182], forgery attacks [163], replay/freshness attacks [54], and inessential log
attacks [105]. Considering the importance of system security, the community has proposed many
attack countermeasures to protect this paradigm.

7.2 Solutions and Countermeasures for Security and Privacy

In this subsection, mainstream solutions against security and privacy attacks are discussed in detail.
And we analyze the advantages and disadvantages of existing countermeasures when applied to
the ECDriven-IoT.
7.2.1 Cryptographic Schemes. Cryptographic schemes are widely used and serve as efficient
strategies to protect communication protocols against various attacks [83]. Encryption/decryption
solutions are inapplicable for wired networks owing to the limited resources in IoT nodes. The
standard encryption/decryption methods are memory- and computing-exhaustive. However, edge-
computing nodes are typically tiny sensors with limited resources, such as battery power, comput-
ing capabilities, and storage memory [100]. These techniques have been investigated and improved
to suit the ECDriven-IoT paradigm. Chen et al. proposed a new security access method without
cryptographic schemes for the ECDriven-IoT paradigm. This solution benefits from the differ-
ence in the hardware of heterogeneous wireless accesses instead of password authentication [36].
Alababy et al. constructed a valid network security model to protect data and suggested a solution
to protect the system from several attacks[10]. Mollah et al. proposed a secure data-sharing scheme
and a secure searching strategy. This sharing scheme uses public and private key encryption to
ensure its security. Thus, applications can perform secure data search and sharing [113].
7.2.2 Secure Data Aggregation, Deduplication, and Analysis. When considering how to
strengthen the system security and privacy paradigm, we naturally take data security and privacy
as the most crucial component. For data security, the system must trace data from aggregation,
transmission to analysis, throughout the data lifetime. Secure data aggregation (SDA) is a
highly secure, privacy-preserving, and efficient data compression strategy [181]. Individual
devices send their data to edge computing nodes. And then, edge nodes aggregate these data by
computing the multiplication of individual data in SDA. To provide users fair incentives, Okay
et al. employed signature techniques, the Boneh-Goh-Nissim cryptosystem, and secret sharing. By
adding noises into the data for differential privacy, oblivious data security and fault tolerance can
be achieved [112]. In Reference [151], to protect against false data injection attacks, they proposed
to filter out the inserted data. To achieve better privacy, the Paillier cryptosystem was modified
to achieve better privacy protection and is used to encrypt consumption data from users [129].
In the ECDriven-IoT paradigm, there is an increasing demand for systems that can provide
cost-efficient secure data storage. For example, Storer et al. proposed secure data deduplication
(SDD) to achieve secure data storage. And many deduplication methods have been proposed
since then [97, 141]. As an effective way to achieve data security and space efficiency, SDD
can be applied to single-server and distributed storage systems [168]. Furthermore, artificial
intelligence (AI) functionalities have shifted from cloud servers to edge devices, which can
potentially improve security and privacy in the ECDriven-IoT [12].

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:26 L. Kong et al.

7.2.3 Combining the ECDriven-IoT With Blockchain Technologies. Blockchain is viewed as a

distributed tamper-resistant database that can be maintained, shared, replicated, and synchronized
by multiple participants in the peer-to-peer (P2P) network [191]. Considering the security and
privacy problems in edge computing, blockchain can be a potential technology to establish a
secure, trusted, and decentralized intelligent system in ECDriven-IoT [75, 84, 187]. When applying
blockchain in edge computing and IoT, it can ensure a reliable tracking of ECDriven-IoT data
transmission and eliminate the requirement for a central trusted intermediary between the com-
municating IoT edge devices [126]. Aiming to improve authentication efficiency, Guo et al. [63]
combined edge computing with blockchain to build a distributed and trusted authentication sys-
tem. This system can guarantee trusted authentication and reliable traceability in edge-computing
nodes. It consists of both a physical network layer, a blockchain network layer, and a blockchain
edge layer to support edge computing. As for the use of blockchain, the system can prevent
network connections from being attacked. Zhao et al. proposed a flexible and configurable
blockchain architecture that provides a mutual authentication protocol and secure consensus,
making it suitable for the ECDriven-IoT. In this architecture, user-defined sensitive data will be
encrypted before storage. Besides, the smart contract is adopted to achieve conditional access,
which can protect blockchain data and transactions [198]. In the ECDriven-IoT, amounts of data is
shared among edge nodes. But owing to the lack of trust, data sharing is hard to complete, and it
is difficult to overcome the computation limitations at the edge. Xu et al. developed a blockchain-
based big-data-sharing framework to support various applications across resource-limited edge
nodes with a low-computation-complexity consensus mechanism. This framework can be applied
to edge devices with low computation and provide security and privacy protection [187]. Further-
more, Kang et al. proposed to utilize consortium blockchain to establish a secure and distributed
vehicular blockchain system for data management and storage by deploying smart contracts [84].
7.2.4 Trusted Execution Environment. With the emergence of the ECDriven-IoT, edge devices
can process large data streams. However, this process exposes the data to a sophisticated vulner-
able attack environment at the edge. The trusted execution environment (TEE) can isolate
data and their computations to shield them from edge attacks. Guan et al. [62] proposed a system
shielding legacy applications from untrusted operating systems by constructing a trusted execu-
tion environment for security-critical applications. Thus, edge applications can execute in these
environments to prevent data from attacks. To optimize data plane performance when achieving
the TEE, Heejin et al. advocated a stream analytics engine called StreamBox-TZ to offer strong
data security, verifiable results, and good performance, thus making efficient data analytics in the
edge possible [131].
7.2.5 Intrusion Detection System. The intrusion detection system (IDS) mainly focuses on
detecting attacks [164, 166]. However, except for monitoring the network’s operations and links,
the IDS can mitigate security threats and report suspicious activities to make the system more
stable and secure. Furthermore, the IDS can detect routing attacks and Black Hole attacks [173].
Hosse [74] presented a new distributed and lightweight IDS based on an artificial immune sys-
tem (AIS). This system consists of the cloud, fog, and edge layers, making it suitable for edge
computing. Wang et al. proposed an IDS architecture for the ECDriven-IoT, which integrates a
trust evaluation mechanism and service template with balanced dynamics [180]. This trust evalu-
ation mechanism can strengthen the system’s security.
7.2.6 Other Solutions. Table 6 shows the advantages and disadvantages of these solutions.
As there are many kinds of attacks on security and privacy, these solutions are various to make
ECDriven-IoT as secure as possible. Beside aforementioned solutions, policy-based mechanisms

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:27

Table 6. Solutions and Countermeasures of Security and Privacy Threats

Solutions Layer Advantages Disadvantages
Battery power, computing capability
Cryptographic Schemes Communication Layer Highly secure
storage memory
Secure Data Aggregation, Protect data security Consume power, render sensitive
Data Layer
Deduplication, Analysis and privacy data to intruders network bandwidth
Trusted, reliable, More complicated system
Combine with Blockchain Architecture Layer
and secure more computing capability
Intrusion Detection System Communication Layer Mitigate security threats Resource consumption

Fig. 8. Application scenarios of edge-computing-driven IoT.

[117], secure firmware update [117], and reliable routing protocols [102] also play a very
important role in making the system secure. As for the security of data, many solutions have been
proposed in academia, such as de-patterning data transmissions [197], decentralization [197], and
authorization [101]. In many applications, these solutions can be combined with other solutions to
work together. Although there have been many solutions, security and privacy in ECDriven-IoT
remain a big challenge in the present.

8 APPLICATIONS
The ECDriven-IoT is suitable for many applications. This section will explain how it works and
how to satisfy the requirements in these application scenes. The ECDriven-IoT plays a major role
in responsive and latency-sensitive IoT applications.

8.1 Smart Homes and Smart Cities

One of the pioneering applications of IoT technology is in home automation and consumer elec-
tronics [145]. Shi et al. introduced some of the challenges and application prospects in smart
life [162]. Figure 8 shows some typical application scenarios of the ECDriven-IoT. More and more
applications are benefiting from the advantages of edge computing, such as smart homes, smart
vehicles, smart medical systems, and intelligent monitoring.
The smart home is a popular IoT application scenario, and some established market products
are widely acclaimed. These products range from simple thermostat sensors to more sophisticated
automation systems, like smart metering, smart heating and lighting, smart cleaning services, and
smart home entertainment systems. However, the smart home is not simply adding IoT commu-
nication modules to traditional home products. In addition to smart devices communicating with
each other, IoT data such as room environment data is also essential for smart homes. Therefore,

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:28 L. Kong et al.

the deployment of a large number of inexpensive sensors and controllers is needed as part of a
collaborative effort in-house. The large amount of data generated by these sensors will be trans-
mitted and used by other IoT devices. Considering data transmission bandwidth pressure and
privacy data protection requirements, edge computing can be an ideal choice for building smart
homes [174]. Furthermore, edge computing will bring other features such as easy installation, re-
location, privacy preservation, and flexibility [148].
Smart homes can be extended to smart communities and even smart cities, and are expected
to become an indispensable part of human life. ECDriven-IoT systems can also serve as the ideal
architecture for smart cities. According to the data growth trend of a city today, the data will
grow exponentially in the future. These data are generated by public safety, health, utilities, and
transportation. Processing the data at the network edge is more efficient than building a cloud
data center. Next, considering the sudden events and public safety in the city, edge computing
can save data transmission time and reduce response latency. This benefit is critical for appli-
cations that require predictability and low latency. In addition, edge computing can make deci-
sions and diagnoses from the network edge, where events occur faster than in the cloud center.
Finally, the natural advantage of edge computing is location awareness. Some geo-based applica-
tions like transportation can collect and analyze data to avoid the dilemma of transmission to the
cloud [171].

8.2 Smart Healthcare

The ECDriven-IoT can make a big difference in smart healthcare, where IoT is widely adopted.
Wearable low-power IoT medical sensors for monitoring health-related data and tracking records
are now popular in public healthcare facilities [73]. Embedding sensors and actuators on patients
are to help doctors monitor patients’ health status and provide feedback to healthcare providers.
However, performance without edge computing is not good enough in terms of latency and ac-
curacy [172]. Remote patient monitoring is a typical use case in smart healthcare. It provides
convenience for doctors and patients far away from medical facilities. The records of patients
have to be processed immediately and securely. Thus, transmission latency can be the bottleneck
that prevents smart healthcare from being applied widely. With the potential benefits, the role
of edge computing in the health and social assistance industries becomes more evident. Much
research on the employment of the ECDriven-IoT in health care has been investigated. In Ref-
erence [176], a remote patient health monitoring scheme was proposed in smart homes via the
concept of edge computing at the gateways. This monitoring system adopts advanced techniques
at the edge of the network. These techniques involve data mining, distributed storage, and notifi-
cation services. Rahmani et al. introduced the smart gateway concept and explored its application
in remote health monitoring [143]. The medical data generated in edge nodes will be collected
and processed to update the monitoring system’s parameters. Due to the geo-distributed nature
of the network, the system can provide real-time notification for patients and privacy in data
gathering.
Gia et al. introduced a medical application of the ECDriven-IoT [60]. The system utilizes edge
computing on the intelligent gateway to enhance the health monitoring system. The specific mea-
sures are data mining at the network edge, and distributed storage to enhance notification services.
Stanciu et al. used blockchain technology as a starting point and integrated blockchain technology
into an edge computing platform to implement a distributed control system [167].

8.3 Mobile VR and AR

The progress of smartphones and smart glasses has increased the popularity of augmented reality
applications [192]. With the development of virtual reality (VR) and augmented reality (AR),

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:29

humans can interact more naturally with the virtual world through the data that are collected
by IoT devices [156]. With IoT sensors, AR technologies can extend the real world to the virtual
world [18]. In the initial step, cloud computing provides the demands of computational power,
which can satisfy these requirements in latency and quality. However, VR and AR can be applied to
more scenes, such as tourism, smart transportation networks, and robotic-assisted surgeries. Cloud
computing is no longer used to satisfy the requirements in latency and throughout the network, as
these scenes are strict in latency, which may decrease user satisfaction. For low-latency offloading
services in VR and AR, edge computing can effectively reduce the latency in combining these
processed data with physical reality.
Edge computing can also migrate computing tasks from mobile devices to edge nodes to in-
crease the computational capacity of VR devices, save battery life, and reduce latency at the same
time [35]. Additionally, edge computing can be connected with the cloud for stronger computing
capabilities when needed [19].
Zao et al. proposed an architecture that combines edge nodes and cloud data centers to leverage
the augmented brain-computer interface [196]. The main benefit of this architecture is the low
latency and real-time interaction, which can provide a more comfortable playing experience in VR
and AR application scenes.

8.4 Industry Application

The Industry Internet of Things (IIoT) is known as Industry 4.0, which means the new era
in the industry area [95]. IIoT incorporates numerous advanced communication and automation
technologies, AI, and big data analysis to improve intelligence and connectivity in industry [188].
Today, new intelligent technologies are applied to accelerate the innovation and transformation
of the factory workforce. IoT can collect data in extreme scenes to protect workers from danger.
Furthermore, these collected data can be stored and analyzed to make better decisions.
IIoT provides many benefits, such as improving operational efficiency, connectivity, and scal-
ability, and saving the time cost for manufacturing processes [133]. Combined with many smart
machines, IIoT aims for higher accuracy, greater efficiency, and more constant working capabilities
than humans. As a complement to IoT, edge computing can play a very important role in IIoT. For
instance, real-time edge analysis and enhanced edge security are the two main ECDriven-IoT appli-
cation scenes. Additionally, edge computing can provide an opportunity to address shortcomings
in the IIoT domain [123].
Edge computing can optimize the performance of traditional IIoT. Instead of transmitting the
sensor data to the cloud directly, edge computing can process those data in edge nodes to reduce
the data volume and bandwidth. Processing sensor data in the edge nodes can also reduce the
latency and preserve the storage in the cloud, improving the service quality of many applications,
including video streaming [85]. Harper et al. proposed a fog-computing-based communication
architecture that will substantially minimize the energy consumption of the IoT nodes [68]. Edge
computational capabilities are further used to predict future data measurements and reduce the
throughput from IoT devices to control units.

8.5 System Evolution

Sarkar et al. showed the parameters and features of edge computing by mathematical calcula-
tions [155]. Their research analyzed and compared the power consumption, service delay, carbon
dioxide emissions, and cost of edge computing and cloud computing. Villari et al. introduced a new
concept, osmotic computing [177], an emerging calculation paradigm similar to edge computing.
It also supports data processing at the network edge while providing IoT services. The author also

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:30 L. Kong et al.

discussed some of its characteristics and future directions. Morabito et al. showed how to enhance
edge computing with lightweight virtualization in the IoT [115].
Dastjerdi et al. presented an introduction to the concepts and characteristics of fog computing.
They also analyzed what a complete edge computing software system looks like, including the
system design patterns, API, and service management [42]. In their research, Gupta et al. started
from a software perspective and first proposed several challenges to be solved in implementing
the edge and IoT paradigms [65]. The most critical challenge is resource management technology.
In other words, how to determine which application modules are deployed in the edge device to
minimize latency and ensure adequate throughput. At the same time, network congestion and
energy costs also need to be considered in the future.

9 LESSONS LEARNED, OPEN CHALLENGES, AND FUTURE DIRECTIONS

9.1 Key Lessons Learned
We have illustrated many challenges ECDriven-IoT has met when applied to reality. In contrast,
we also can discover many opportunities and advantages that ECDriven-IoT will eventually bring.
Edge computing and IoT, when they are deployed independently, both have many shortcomings,
which have prevented them from being widely used and developed. But when combined, they
can help each other in bandwidth, power consumption, latency, security, and so on. We draw
some lessons from our extensive survey of related work, including ECDriven-IoT architecture and
standards, efficient communication, application, and security.
9.1.1 Unified Architecture and Standard. Since edge and IoT devices are heterogeneous from
bottom hardware to top system design, current related research is fragmented and lacks a unified
measurement for proposed ECDriven-IoT solutions. For example, multiple embedded operating
systems are designed to abstract heterogeneous IoT and edge devices, but ECDriven-IoT applica-
tions cannot migrate directly between these operating systems, because they have different ex-
posed interfaces. Establishing standards is very important for the development of a field, and the
same is true for ECDriven-IoT. ECDriven-IoT should have a unified architecture or interface stan-
dard to facilitate its deployment and usage. Otherwise, the benefits of ECDriven-IoT will be largely
hindered by the heterogeneous nature of IoT and edge computing. Though some existing studies
have provided architecture design schemes of ECDriven-IoT, few of them are committed to push-
ing forward standards establishment in this area.
9.1.2 Efficient Communication and Computation Coordination. The communication latency of
ECDriven-IoT applications can be further reduced owing to the decentralized fabric of edge com-
puting. However, when it comes to different IoT applications, different communication protocols,
and network conditions, how to ensure that the communication between IoT devices and edge
nodes is efficient is a critical concern. Specifically, how to coordinate edge nodes to complete the
computing work of IoT devices is a key optimization problem. Besides, diverse communication
protocols in the IoT area make the encoding-decoding in edge nodes more complex. One solution
for edge nodes to handle various protocols is supporting several communication protocols, but
this solution is relatively costly. Another solution is the newly emerged cross-technology com-
munication (CTC) technologies, which enable two or more different communication protocols
to communicate with each other. Although many CTC technologies have been explored to make
communication more efficient, they are only limited to two or three technologies [64]. Thus, how
to achieve efficient communication still need more exploration.
9.1.3 Practical Security and Privacy Solutions. Security and privacy issues of ECDriven-IoT are
more complicated due to the heterogeneous and distributed architecture. Current solutions cover

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:31

multiple layers of ECDriven-IoT, i.e., architecture layer, communication layer, and data layer. But
these solutions are far from satisfactory because of high computation complexity or specific hard-
ware demands. For example, blockchain-based solutions for architecture security bring extra com-
munication overheads and storage costs; TEE-based solutions for computation security rely on
trusted hardware. ECDriven-IoT should focus on exploring lightweight and the common security
and privacy solutions, which are practical and efficient even in resource-constrained devices.

9.1.4 Different Designs for Different Application Scenes. The design of ECDriven-IoT cannot be
illustrated simply by a single model, and the system requirements vary for different application
scenarios. The ECDriven-IoT system design should be flexibly adapted to make it more efficient
and suitable for various ECDriven-IoT applications. For example, we may be more concerned with
the computational and communication complexity of real-time applications (e.g., smart health,
autonomous driving, VR) but more concerned with the power consumption of long-life protection
applications (e.g., field environmental monitoring). It is a trade-off in the design of ECDriven-
IoT systems. Therefore, when designing an ECDriven-IoT system, we should fully consider what
metrics the application really cares about and give a suitable application-specific solution.

9.2 Challenges and Future Directions

There are many challenges to be solved in edge computing, especially related to IoT. In this
section, we will discuss some of the open research challenges and potential future work in the
ECDriven-IoT.

9.2.1 Heterogeneous Platforms in Edge Computing and IoT. In a traditional cloud computing
data center, users do not need to know how the program works or care about the underlying
hardware architecture. However, in the ECDriven-IoT, edge devices and networks need to take on
computing tasks while considering heterogeneous hardware platforms. The heterogeneous nature
of ECDriven-IoT leads to a significant increase in programming workload for developers. The
future development of IoT relies on edge computing, and the application scenarios are rich and
colorful. Effectively solving the difficulties brought by heterogeneous platforms will make more
developers invest in such work.
How to discover resources and services in a distributed computing environment is an area to
be explored. To make full use of the edge devices of the network, it is necessary to establish a
discovery mechanism to find the appropriate nodes that can be deployed in a distributed manner.
Because of the sheer number of devices available, these mechanisms cannot rely on manuals. In
addition, various heterogeneous devices are needed to meet the latest computing needs, such as
large-scale machine learning tasks. These mechanisms must seamlessly integrate computational
workflows at different levels without increasing latency or compromising the user experience. The
original cloud-based methods are no longer applicable in edge computing.

9.2.2 Task Allocation in the ECDriven-IoT. The biggest challenge for the ECDriven-IoT is how
to deploy large-scale computing and storage capabilities dynamically [20]. The appropriate de-
ployment will make device sides work together efficiently and seamlessly. Evolving distributed
computing has spawned many technologies that are used to facilitate the task of partitioning in
multiple geographies. However, on the edge side, partitioned computing not only poses the chal-
lenge of efficient partitioning but also encounters bottlenecks in automatic allocation without the
capacity or location of edge nodes. Therefore, a new scheduling strategy is needed to assign tasks
to edge nodes. It is a prominent issue that must be addressed for large-scale deployments of IoT
edge devices and networks and will affect the scale of the development of the ECDriven-IoT.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:32 L. Kong et al.

9.2.3 Data Abstraction in Edge Computing and IoT. Although data-generation devices in the IoT
do not need to send generated data to the data center frequently, the edge node needs some required
data to perform analysis work. Data abstraction refers to these data pre-processing algorithms and
solutions, including noise cancellation, data classification, and computing. For example, gateways
need to process some events such as noise cancellation. However, IoT devices are rich and varied,
and different devices use different data formats. It brings the first challenge to data abstraction.
The second challenge is how to effectively determine the level of data abstraction. Considering
data security issues, the application does not get all the raw data, but only abstracts the parts it is
interested in. If too little of the raw data is filtered, then the application will not get the information
it needs. However, keeping too much raw data can cause storage problems. In addition, the data
generated by edge devices are often unreliable due to external interference from sensors. Therefore,
extracting accurate information from unreliable raw data is another challenge.
The application needs to control objects to provide a specific service, such as reading and writing
data. The data abstraction layer combines the presentation of data and corresponding operations
and provides a unified interface. In addition, finding a universal way of data abstraction is not easy
because of the diversity of devices, different ways of presenting data, and different corresponding
operations.
9.2.4 Edge Nodes Security. The security of the ECDriven-IoT requires end-to-end protection.
As the device is closer to IoT, the difficulties in network edge-side access control and threat pro-
tection will increase dramatically. Edge-side security mainly includes device security, network
security, data security, and application security. In addition, the confidentiality of critical data and
the protection of personal privacy data are vital areas of IoT security [149].
Several issues must be addressed before end devices (e.g., switches, base stations) are used as
edge nodes for shared access. First, risks associated with users and owners of edge devices need to
be defined. Second, when the device is used as an edge computing node, the original functionality
of the device cannot be compromised. Third, multiple users on edge nodes need security as their
primary concern. Fourth, the minimum service level needs to be guaranteed to the users of edge
nodes. Finally, workloads, computing power, data locations and migration, maintenance costs, and
energy consumption need to establish an appropriate pricing model.
9.2.5 Development Tools for Edge-computing-driven IoT. As the number of edge nodes support-
ing general-purpose computing continues to increase, the demand for development frameworks
and toolkits will continue to grow. Edge analysis is different from existing work. Since edge anal-
ysis is implemented in user-driven applications, existing tools may not be suitable for expressing
edge analysis workflows. The programming model needs to use edge nodes to support the paral-
lelism of tasks and perform calculations on multiple levels of hardware. At the same time, program-
ming languages need to consider the hardware heterogeneity in the workflow and the computing
power of various resources. So, ECDriven-IoT is more complicated than existing cloud computing
models.

10 CONCLUSION
In this article, we present a comprehensive survey of the ECDriven-IoT, including supporting
technologies and research challenges in this field. We first categorize existing studies to help re-
searchers find innovative research topics. We then propose some open issues worthy of study and
contribute to the development of the industry. Currently, the research on the ECDriven-IoT topic
is still highly fragmented, which is not conducive to the research and development of the field.
Therefore, this survey helps review and summarize existing research work and promote cross-
cooperation in related areas.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:33

REFERENCES
[1] Mohammad Aazam and Eui-Nam Huh. 2014. Fog computing and smart gateway-based communication for cloud of
things. In Proceedings of the International Conference on Future Internet of Things and Cloud. IEEE, 464–470.
[2] Nasir Abbas, Yan Zhang, Amir Taherkordi, and Tor Skeie. 2018. Mobile edge computing: A survey. IEEE Internet
Things J. 5, 1 (2018), 450–465.
[3] Stefano Abbate, Marco Avvenuti, Daniel Cesarini, and Alessio Vecchio. 2012. Estimation of energy consumption for
TinyOS 2.x-based applications. Procedia Comput. Sci. 10 (2012), 1166–1171.
[4] Utku Günay Acer, Aidan Boran, Claudio Forlivesi, Werner Liekens, Fernando Pérez-cruz, and Fahim Kawsar. 2015.
Sensing WiFi network for personal IoT analytics. In Proceedings of the International Conference on the Internet of
Things (IOT’15). IEEE, 104–111.
[5] Ansuman Adhikary, Xingqin Lin, and Y.-P. Eric Wang. 2016. Performance evaluation of NB-IoT coverage. In Proceed-
ings of the IEEE Vehicular Technology Society (VTC’16). IEEE, 1–5.
[6] Bilal Afzal, Muhammad Umair, Ghalib Asadullah Shah, and Ejaz Ahmed. 2019. Enabling IoT platforms for social IoT
applications: Vision, feature mapping, and challenges. Future Gen. Comput. Syst. 92 (2019), 718–731.
[7] Yuan Ai, Mugen Peng, and Kecheng Zhang. 2018. Edge computing technologies for internet of things: A primer. Dig.
Commun. Netw. 4, 2 (2018), 77–86.
[8] G. R. Aiello and G. D. Rogerson. 2003. Ultra-wideband wireless systems. IEEE Microwave Mag. 4, 2 (2003), 36–47.
[9] K. Akkaya and M. Younis. 2003. An energy-aware QoS routing protocol for wireless sensor networks. In Proceedings
of the International Conference on Distributed Computing Systems Workshops. IEEE, 710–715.
[10] Fadele Ayotunde Alaba, Mazliza Othman, Ibrahim Abaker Targio Hashem, and Faiz Alotaibi. 2017. Internet of things
security: A survey. J. Netw. Comput. Appl. 88 (2017), 10–28.
[11] Thamer A. Alghamdi, Aboubaker Lasebae, and Mahdi Aiash. 2013. Security analysis of the constrained application
protocol in the Internet of Things. In Proceedings of the International Conference on Future Generation Communication
Technologies. IEEE, 163–168.
[12] Abdulmalik Alwarafy, Khaled A. Al-Thelaya, Mohamed Abdallah, Jens Schneider, and Mounir Hamdi. 2021. A survey
on security and privacy issues in edge-computing-assisted internet of things. IEEE Internet Things J. 8, 6 (2021),
4004–4022.
[13] Mahdi Amiri-Kordestani and Hadj Bourdoucen. 2017. A survey on embedded open source system software for the
internet of things. In Proceedings of the Free and Open Source Software Conference. 6.
[14] Jeffrey G. Andrews, Stefano Buzzi, Wan Choi, Stephen V. Hanly, Angel Lozano, Anthony C. K. Soong, and
Jianzhong Charlie Zhang. 2014. What will 5G be? IEEE J. Select. Areas Commun. 32, 6 (2014), 1065–1082.
[15] Luigi Atzori, Antonio Iera, and Giacomo Morabito. 2010. The internet of things: A survey. Comput. Netw. 54, 15
(2010), 2787–2805.
[16] Israr Iqbal Awan, Nadir Shah, Muhammad Imran, Muhammad Shoaib, and Nasir Saeed. 2019. An improved mecha-
nism for flow rule installation in-band SDN. J. Syst. Architect. 96 (2019), 1–19.
[17] Ahmet Cihat Baktir, Atay Ozgovde, and Cem Ersoy. 2017. How can edge computing benefit from software-defined
networking: A survey, use cases, and future directions. IEEE Commun. Surveys Tutor. 19, 4 (2017), 2359–2391.
[18] Luciano Baresi, Danilo Filgueira Mendonça, and Martin Garriga. 2017. Empowering low-latency applications through
a serverless edge computing architecture. In Service-Oriented and Cloud Computing. Springer International Publish-
ing, Cham, 196–210.
[19] Ejder Bastug, Mehdi Bennis, Muriel Medard, and Merouane Debbah. 2017. Toward interconnected virtual reality:
Opportunities, challenges, and enablers. IEEE Commun. Mag. 55, 6 (2017), 110–117.
[20] Michael Till Beck, Martin Werner, Sebastian Feld, and S. Schimper. 2014. Mobile edge computing: A taxonomy. In
Proceedings of the International Conference on Advances in Future Internet. Citeseer, 48–55.
[21] A. Bekasiewicz and S. Koziel. 2016. Compact UWB monopole antenna for internet of things applications. Electron.
Lett. 52, 7 (2016), 492–494.
[22] Yihenew Dagne Beyene, Riku Jantti, Olav Tirkkonen, Kalle Ruttik, Sassan Iraji, Anna Larmo, Tuomas Tirronen, and
Johan Torsner. 2017. NB-IoT technology overview and experience from cloud-RAN implementation. IEEE Wireless
Commun. 24, 3 (2017), 26–32.
[23] Flavio Bonomi, Rodolfo Milito, Preethi Natarajan, and Jiang Zhu. 2014. Fog computing: A platform for internet of
things and analytics. In Big Data and Internet of Things: A Roadmap for Smart Environments. Springer International
Publishing, Cham, 169–186.
[24] Flavio Bonomi, Rodolfo Milito, Jiang Zhu, and Sateesh Addepalli. 2012. Fog computing and its role in the internet of
things. In Proceedings of the 1st MCC Workshop on Mobile Cloud Computing. ACM, New York, NY, 13–16.
[25] Martin C. Bor, Utz Roedig, Thiemo Voigt, and Juan M. Alonso. 2016. Do LoRa low-power wide-area networks scale? In
Proceedings of the ACM International Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems.
ACM, New York, NY, 59–67.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:34 L. Kong et al.

[26] Alessio Botta, Walter de Donato, Valerio Persico, and Antonio Pescapé. 2016. Integration of cloud computing and
internet of things: A survey. Future Gen. Comput. Syst. 56 (2016), 684–700.
[27] Fangbo Cai, Nafei Zhu, Jingsha He, Pengyu Mu, Wenxin Li, and Yi Yu. 2019. Survey of access control models and
technologies for cloud computing. Cluster Comput. 22, 3 (2019), 6111–6122.
[28] Qing Cao, Tarek Abdelzaher, John Stankovic, and Tian He. 2008. The LiteOS operating system: Towards unix-like
abstractions for wireless sensor networks. In Proceedings of the International Conference on Information Processing in
Sensor Networks (IPSN’08). IEEE, 233–244.
[29] Juan José Martínez Castillo and Karina Aviles Rodriguez. 2012. Security architecture for Ad hoc NOMOHi networks:
Development of a project based on emergency rural telecommunications. In Proceedings of the World Congress on
Internet Security (WorldCIS’12). IEEE, 183–187.
[30] Korhan Cengiz and Tamer Dag. 2015. A review on the recent energy-efficient approaches for the internet protocol
stack. EURASIP J. Wireless Commun. Netw. 2015, 1 (2015), 1–22.
[31] Hojung Cha, Sukwon Choi, Inuk Jung, Hyoseung Kim, Hyojeong Shin, Jaehyun Yoo, and Chanmin Yoon. 2007. RE-
TOS: Resilient, expandable, and threaded operating system for wireless sensor networks. In Proceedings of the Inter-
national Symposium on Information Processing in Sensor Networks. IEEE, 148–157.
[32] Tej Bahadur Chandra, Pushpak Verma, and A. K. Dwivedi. 2016. Operating systems for internet of things: A compar-
ative study. In Proceedings of the International Conference on Information and Communication Technology for Compet-
itive Strategies (ICTCS’16). ACM, New York, NY, Article 47, 6 pages.
[33] Hyunseok Chang, Adiseshu Hari, Sarit Mukherjee, and T. V. Lakshman. 2014. Bringing the cloud to the edge. In
Proceedings of the IEEE Conference on Computer Communications Workshops (INFOCOM’14). IEEE, 346–351.
[34] Kai Chih Chang, Razieh Nokhbeh Zaeem, and K. Suzanne Barber. 2018. Enhancing and evaluating identity privacy
and authentication strength by utilizing the identity ecosystem. In Proceedings of the Workshop on Privacy in the
Electronic Society. ACM, New York, NY, 114–120.
[35] Mingzhe Chen, Walid Saad, and Changchuan Yin. 2018. Virtual reality over wireless networks: Quality-of-service
model and learning-based resource management. IEEE Trans. Commun. 66, 11 (2018), 5621–5635.
[36] Songlin Chen, Yixin Jiang, Hong Wen, Wenjie Liu, Jie Chen, Wenxin Lei, and Aidong Xu. 2018. A novel terminal
security access method based on edge computing for IoT. In Proceedings of the International Conference on Networking
and Network Applications (NaNA’18). IEEE, 394–398.
[37] Xu Chen. 2014. Decentralized computation offloading game for mobile cloud computing. IEEE Trans. Parallel Distrib.
Syst. 26, 4 (2014), 974–983.
[38] Xu Chen, Lei Jiao, Wenzhong Li, and Xiaoming Fu. 2015. Efficient multi-user computation offloading for mobile-edge
cloud computing. IEEE/ACM Trans. Netw. 24, 5 (2015), 2795–2808.
[39] Zhipeng Cheng, Minghui Min, Zhibin Gao, and Lianfen Huang. 2020. Joint task offloading and resource allocation
for mobile edge computing in ultra-dense network. In Proceedings of the IEEE Global Communications Conference.
IEEE, 1–6.
[40] Shao-Yi Chien, Wei-Kai Chan, Yu-Hsiang Tseng, Chia-Han Lee, V. Srinivasa Somayazulu, and Yen-Kuang Chen. 2015.
Distributed computing in IoT: System-on-a-chip for smart cameras as an example. In Proceedings of the 20th Asia and
South Pacific Design Automation Conference. IEEE, 130–135.
[41] Supratim Das, Amarjeet Singh, Surinder Pal Singh, and Amit Kumar. 2015. A low overhead dynamic memory manage-
ment system for constrained memory embedded systems. In Proceedings of the International Conference on Computing
for Sustainable Global Development (INDIACom’15). IEEE, 809–815.
[42] A. Dastjerdi and R. Buyya. 2016. Fog computing: Helping the internet of things realize its potential. Computer 49, 8
(2016), 112–116.
[43] Rustem Dautov and Salvatore Distefano. 2017. Three-level hierarchical data fusion through the IoT, edge, and cloud
computing. In Proceedings of the International Conference on Internet of Things and Machine Learning. ACM, New
York, NY, Article 1, 5 pages.
[44] Josep Domingo-Ferrer, Oriol Farrás, Jordi Ribes-González, and David Sánchez. 2019. Privacy-preserving cloud com-
puting on sensitive data: A survey of methods, products and challenges. Comput. Commun. 140–141 (2019), 38–60.
[45] Jianbo Du, Liqiang Zhao, Jie Feng, Xiaoli Chu, and F. Richard Yu. 2018. Economical revenue maximization in cache
enhanced mobile edge computing. In Proceedings of the IEEE International Conference on Communications (ICC’18).
IEEE, 1–6.
[46] Adam Dunkels. 2007. Rime-a lightweight layered communication stack for sensor networks. In Proceedings of the
European Conference on Wireless Sensor Networks, Vol. 44. Citeseer, 2.
[47] EdgeX. 2018. EdgeX Foundry. Retrieved from https://www.edgexfoundry.org/get-started/.
[48] Hesham El-Sayed, Sharmi Sankar, Mukesh Prasad, Deepak Puthal, Akshansh Gupta, Manoranjan Mohanty, and Chin-
Teng Lin. 2018. Edge of things: The big picture on the integration of edge, IoT and the cloud in a distributed computing
environment. IEEE Access 6 (2018), 1706–1717.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:35

[49] Hanan Elazhary. 2019. Internet of things (IoT), mobile cloud, cloudlet, mobile IoT, IoT cloud, fog, mobile edge, and
edge emerging computing paradigms: Disambiguation and research directions. J. Netw. Comput. Applications 128
(2019), 105–140.
[50] Sinem Coleri Ergen. 2004. ZigBee/IEEE 802.15. 4 summary. UC Berkeley (Sept. 10, 2004).
[51] Ericsson. 2010. CEO to shareholders: 50 billion connections 2020. Retrieved from https://www.ericsson.com/en/press-
releases/2010/4/ceo-to-shareholders-50-billion-connections-2020.
[52] Mohamed Fahim, Brahim Ouchao, Abdeslam Jakimi, and Lahcen El Bermi. 2019. Application of a non-immersive
VR, IoT based approach to help moroccan students carry out practical activities in a personal learning style. Future
Internet 11, 1 (2019), 15.
[53] S. Farah, A. Benachenhou, G. Neveux, D. Barataud, G. Andrieu, and T. Fredon. 2015. Real-time microwave remote
laboratory architecture. In Proceedings of the European Microwave Conference (EuMC). IEEE, 1315–1318.
[54] Yuxiang Feng, Wenhao Wang, Yukai Weng, and Huanming Zhang. 2017. A replay-attack resistant authentication
scheme for the internet of things. In Proceedings of the IEEE International Conference on Computational Science and
Engineering (CSE’17) and IEEE International Conference on Embedded and Ubiquitous Computing (EUC’17), Vol. 1. IEEE,
541–547.
[55] Hiro Gabriel Cerqueira Ferreira, Edna Dias Canedo, and Rafael Timóteo de Sousa. 2013. IoT architecture to enable
intercommunication through REST API and UPnP using IP, ZigBee and arduino. In Proceedings of the IEEE 9th Inter-
national Conference on Wireless and Mobile Computing, Networking and Communications (WiMob’13). IEEE, 53–60.
[56] EdgeX Foundry. 2021. Why EdgeX. Retrieved from https://www.edgexfoundry.org/why_edgex/why-edgex/.
[57] Khusanbek Gafurov and Tai-Myoung Chung. 2019. Comprehensive survey on internet of things, architecture, se-
curity aspects, applications, related technologies, economic perspective, and future directions. J. Info. Process. Syst.
15 (2019), 797–819.
[58] Padmini Gaur and Mohit P. Tahiliani. 2015. Operating systems for IoT devices: A critical survey. In Proceedings of
the IEEE Region 10 Symposium. IEEE, 33–36.
[59] Amitabha Ghosh, Andreas Maeder, Matthew Baker, and Devaki Chandramouli. 2019. 5G evolution: A view on 5G
cellular technology beyond 3GPP release 15. IEEE Access 7 (2019), 127639–127651.
[60] Tuan Nguyen Gia, Mingzhe Jiang, Amir-Mohammad Rahmani, Tomi Westerlund, Pasi Liljeberg, and Hannu Ten-
hunen. 2015. Fog computing in healthcare internet of things: A case study on ECG feature extraction. In IEEE Interna-
tional Conference on Computer and Information Technology; Ubiquitous Computing and Communications; Dependable,
Autonomic and Secure Computing; Pervasive Intelligence and Computing. IEEE, 356–363.
[61] Dimitrios Glaroudis, Athanasios Iossifides, and Periklis Chatzimisios. 2020. Survey, comparison and research chal-
lenges of IoT application protocols for smart farming. Comput. Netw. 168, C (2020), 14.
[62] Le Guan, Peng Liu, Xinyu Xing, Xinyang Ge, Shengzhi Zhang, Meng Yu, and Trent Jaeger. 2017. Trustshadow: Secure
execution of unmodified applications with arm trustzone. In Proceedings of the 15th Annual International Conference
on Mobile Systems, Applications, and Services. ACM, New York, NY, 488–501.
[63] Shaoyong Guo, Xing Hu, Song Guo, Xuesong Qiu, and Feng Qi. 2020. Blockchain meets edge computing: A distributed
and trusted authentication system. IEEE Trans. Industr. Inform. 16, 3 (2020), 1972–1983.
[64] Xiuzhen Guo, Yuan He, Jia Zhang, and Haotian Jiang. 2019. WIDE: Physical-level CTC via digital emulation. In
Proceedings of the International Conference on Information Processing in Sensor Networks (IPSN’19). IEEE, 49–60.
[65] Harshit Gupta, Amir Vahid Dastjerdi, Soumya K. Ghosh, and Rajkumar Buyya. 2017. iFogSim: A toolkit for modeling
and simulation of resource management techniques in the Internet of Things, Edge and Fog computing environments.
Software: Pract. Exper. 47, 9 (2017), 1275–1296.
[66] J. C. Haartsen. 2000. The bluetooth radio system. IEEE Person. Commun. 7, 1 (2000), 28–36.
[67] Oliver Hahm, Emmanuel Baccelli, Hauke Petersen, and Nicolas Tsiftes. 2016. Operating systems for low-end devices
in the internet of things: A survey. IEEE Internet Things J. 3, 5 (2016), 720–734.
[68] K. Eric Harper, Thijmen de Gooijer, Johannes O. Schmitt, and David Cox. 2016. Microdatabases for the industrial
Internet. Retrieved from https://arxiv.org/abs/1601.04036.
[69] Albert F. Harris III, Vansh Khanna, Guliz Tuncay, Roy Want, and Robin Kravets. 2016. Bluetooth low energy in dense
IoT environments. IEEE Commun. Mag. 54, 12 (2016), 30–36.
[70] Brian Hayes. 2008. Cloud computing. Commun. ACM 51, 7 (2008), 9–11.
[71] John L. Hennessy and David A. Patterson. 2011. Computer Architecture: A Quantitative Approach. Elsevier.
[72] P. S. Henry and Hui Luo. 2002. WiFi: What’s next? IEEE Commun. Mag. 40, 12 (2002), 66–72.
[73] M. Shamim Hossain and Ghulam Muhammad. 2016. Cloud-assisted industrial internet of things (IIoT)—Enabled
framework for health monitoring. Comput. Netw. 101 (2016), 192–202.
[74] Farhoud Hosseinpour, Payam Amoli, Juha Plosila, Timo Hämäläinen, and Hannu Tenhunen. 2016. An intrusion
detection system for Fog computing and IoT based logistic systems using a smart data approach. Int. J. Dig. Content
Technol. Appl. 10 (12 2016), 34–46.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:36 L. Kong et al.

[75] Junqin Huang, Linghe Kong, Guihai Chen, Min-You Wu, Xue Liu, and Peng Zeng. 2019. Towards secure indus-
trial IoT: Blockchain system with credit-based consensus mechanism. IEEE Trans. Industr. Inform. 15, 6 (2019),
3680–3689.
[76] Jonathan Hui and Pascal Thubert. 2011. Compression format for IPv6 datagrams over IEEE 802.15. 4-based networks.
Technical Report 6282. Internet Engineering Task Force (IETF), 1–23. Retrieved from https://www.rfc-editor.org/rfc/
rfc6282.txt.
[77] Yaser Jararweh, Mahmoud Al-Ayyoub, Elhadj Benkhelifa, Mladen Vouk, Andy Rindos, et al. 2015. SDIoT: A software
defined based internet of things framework. J. Ambient Intell. Human. Comput. 6, 4 (2015), 453–461.
[78] Yaser Jararweh, Mohammad Alsmirat, Mahmoud Al-Ayyoub, Elhadj Benkhelifa, Ala’ Darabseh, Brij Gupta, and
Ahmad Doulat. 2017. Software-defined system support for enabling ubiquitous mobile edge computing. Comput.
J. 60, 10 (2017), 1443–1457.
[79] Yaser Jararweh, Ahmad Doulat, Omar AlQudah, Ejaz Ahmed, Mahmoud Al-Ayyoub, and Elhadj Benkhelifa. 2016.
The future of mobile cloud computing: Integrating cloudlets and mobile edge computing. In Proceedings of the 23rd
International Conference on Telecommunications (ICT’16). IEEE, 1–5.
[80] Farhana Javed, Muhamamd Khalil Afzal, Muhammad Sharif, and Byung-Seo Kim. 2018. Internet of things (IoT) oper-
ating systems support, networking technologies, applications, and challenges: A comparative review. IEEE Commun.
Surveys Tutor. 20, 3 (2018), 2062–2100.
[81] Fatma Ben Jemaa, Guy Pujolle, and Michel Pariente. 2016. Cloudlet-and NFV-based carrier Wi-Fi architecture for a
wider range of services. Ann. Telecommun. 71, 11 (2016), 617–624.
[82] Sunmi Jun, Yoohwa Kang, Jaeho Kim, and Changki Kim. 2020. Ultra-low-latency services in 5G systems: A perspec-
tive from 3GPP standards. ETRI J. 42, 5 (2020), 721–733.
[83] Ajay Kakkar. 2020. A survey on secure communication techniques for 5G wireless heterogeneous networks. Info.
Fusion 62 (2020), 89–109.
[84] Jiawen Kang, Rong Yu, Xumin Huang, Maoqiang Wu, Sabita Maharjan, Shengli Xie, and Yan Zhang. 2019. Blockchain
for secure and efficient data sharing in vehicular edge computing and networks. IEEE Internet Things J. 6, 3 (2019),
4660–4670.
[85] Hajime Kanzaki, Kevin Schubert, and Nicholas Bambos. 2017. Video streaming schemes for industrial IoT. In Pro-
ceedings of the International Conference on Computer Communication and Networks (ICCCN’17). IEEE, 1–7.
[86] Wazir Zada Khan, Ejaz Ahmed, Saqib Hakak, Ibrar Yaqoob, and Arif Ahmed. 2019. Edge computing: A survey. Future
Gen. Comput. Syst. 97 (2019), 219–235.
[87] Song Min Kim and Tian He. 2015. FreeBee: Cross-technology communication via free side-channel. In Proceedings
of the Annual International Conference on Mobile Computing and Networking (MobiCom’15). ACM, New York, NY,
317–330.
[88] Shinji Kitagami, Tadashi Ogino, Takuo Suganuma, and Norio Shiratori. 2017. Proposal of a multi-agent based flexible
IoT edge computing architecture harmonizing its control with cloud computing. In Proceedings of the International
Symposium on Computing and Networking (CANDAR’17). IEEE, 223–229.
[89] Diego Kreutz, Fernando M. V. Ramos, Paulo Esteves Veríssimo, Christian Esteve Rothenberg, Siamak Azodolmolky,
and Steve Uhlig. 2015. Software-defined networking: A comprehensive survey. Proc. IEEE 103, 1 (2015), 14–76.
[90] Karthik Kumar, Jibang Liu, Yung-Hsiang Lu, and Bharat Bhargava. 2013. A survey of computation offloading for
mobile systems. Mobile Netw. Appl. 18, 1 (2013), 129–140.
[91] Vijay Kumar, George Oikonomou, Theo Tryfonas, Dan Page, and Iain Phillips. 2014. Digital investigations for IPv6-
based wireless sensor networks. Dig. Invest. 11, S2 (2014), S66–S75.
[92] Patrick Kurp. 2008. Green computing. Commun. ACM 51, 10 (2008), 11–13.
[93] O. Landsiedel, K. Wehrle, and S. Gotz. 2005. Accurate prediction of power consumption in sensor networks. In Pro-
ceedings of the IEEE Workshop on Embedded Networked Sensors. IEEE, 37–44.
[94] Erik G. Larsson, Ove Edfors, Fredrik Tufvesson, and Thomas L. Marzetta. 2014. Massive MIMO for next generation
wireless systems. IEEE Commun. Mag. 52, 2 (2014), 186–195.
[95] Heiner Lasi, Peter Fettke, Hans-Georg Kemper, Thomas Feld, and Michael Hoffmann. 2014. Industry 4.0. Bus. Info.
Syst. Eng. 6, 4 (2014), 239–242.
[96] Bo Li, Qiang He, Feifei Chen, Hai Jin, Yang Xiang, and Yun Yang. 2020. Auditing cache data integrity in the edge
computing environment. IEEE Trans. Parallel Distrib. Syst. 32, 5 (2020), 1210–1223.
[97] Jiliang Li, Zhou Su, Deke Guo, Kim-Kwang Raymond Choo, Yusheng Ji, and Huayan Pu. 2021. Secure data
deduplication protocol for edge-assisted mobile CrowdSensing services. IEEE Trans. Vehic. Technol. 70, 1 (2021),
742–753.
[98] Shichao Li, Ning Zhang, Siyu Lin, Linghe Kong, Ajay Katangur, Muhammad Khurram Khan, Minming Ni, and Gang
Zhu. 2018. Joint admission control and resource allocation in edge computing for internet of things. IEEE Netw. 32,
1 (2018), 72–79.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:37

[99] Fuhong Lin, Yutong Zhou, Xingsuo An, Ilsun You, and Kim-Kwang Raymond Choo. 2018. Fair resource allocation in
an intrusion-detection system for edge computing: Ensuring the security of internet of things devices. IEEE Consum.
Electron. Mag. 7, 6 (2018), 45–50.
[100] Jie Lin, Wei Yu, Nan Zhang, Xinyu Yang, Hanlin Zhang, and Wei Zhao. 2017. A survey on internet of things:
Architecture, enabling technologies, security and privacy, and applications. IEEE Internet Things J. 4, 5 (2017),
1125–1142.
[101] Dan Liu, Zheng Yan, Wenxiu Ding, and Mohammed Atiquzzaman. 2019. A survey on secure data analytics in edge
computing. IEEE Internet Things J. 6, 3 (2019), 4946–4967.
[102] Yang Lu and Li Da Xu. 2018. Internet of things (IoT) cybersecurity research: A review of current research topics. IEEE
Internet Things J. 6, 2 (2018), 2103–2115.
[103] Xin Ma and Wei Luo. 2008. The analysis of 6LowPAN technology. In Proceedings of the IEEE Pacific-Asia Workshop
on Computational Intelligence and Industrial Application, Vol. 1. IEEE, 963–966.
[104] Pavel Mach and Zdenek Becvar. 2017. Mobile edge computing: A survey on architecture and computation offloading.
IEEE Commun. Surveys Tutor. 19, 3 (2017), 1628–1656.
[105] Shahid Mahmood, Amin Ullah, and Anas Khalid Kayani. 2019. Fog computing trust based architecture for internet
of things devices. Int. J. Comput. Commun. Netw. 1, 1 (2019), 18–25.
[106] Luca Mainetti, Luigi Patrono, and Antonio Vilei. 2011. Evolution of wireless sensor networks towards the internet
of things: A survey. In Proceedings of the International Conference on Software, Telecommunications and Computer
Networks. IEEE, 1–6.
[107] Nitin Mangalvedhe, Rapeepat Ratasuk, and Amitava Ghosh. 2016. NB-IoT deployment study for low power wide area
cellular IoT. In Proceedings of the IEEE 27th Annual International Symposium on Personal, Indoor, and Mobile Radio
Communications (PIMRC’16). IEEE, 1–6.
[108] Yuyi Mao, Changsheng You, Jun Zhang, Kaibin Huang, and Khaled B. Letaief. 2017. A survey on mobile edge com-
puting: The communication perspective. IEEE Commun. Surveys Tutor. 19, 4 (2017), 2322–2358.
[109] Yuyi Mao, Jun Zhang, and Khaled B. Letaief. 2016. Dynamic computation offloading for mobile-edge computing with
energy harvesting devices. IEEE J. Select. Areas Commun. 34, 12 (2016), 3590–3605.
[110] Diego Mendez Mena, Ioannis Papapanagiotou, and Baijian Yang. 2018. Internet of things: Survey on security. Info.
Secur. J.: Global Perspect. 27, 3 (2018), 162–182.
[111] Rashid Mijumbi, Joan Serrat, Juan-Luis Gorricho, Niels Bouten, Filip De Turck, and Raouf Boutaba. 2016. Net-
work function virtualization: State-of-the-art and research challenges. IEEE Commun. Surveys Tutor. 18, 1 (2016),
236–262.
[112] Takuho Mitsunaga, Yoshifumi Manabe, and Tatsuaki Okamoto. 2010. Efficient secure auction protocols based on the
Boneh-Goh-Nissim encryption. In Advances in Information and Computer Security. Springer, Berlin, 149–163.
[113] Muhammad Baqer Mollah, Md. Abul Kalam Azad, and Athanasios Vasilakos. 2017. Secure data sharing and searching
at the edge of cloud-assisted internet of things. IEEE Cloud Comput. 4, 1 (2017), 34–42.
[114] N. Montavont and T. Noel. 2002. Handover management for mobile nodes in IPv6 networks. IEEE Commun. Mag. 40,
8 (2002), 38–43.
[115] Roberto Morabito, Vittorio Cozzolino, Aaron Yi Ding, Nicklas Beijar, and Jorg Ott. 2018. Consolidate IoT edge com-
puting with lightweight virtualization. IEEE Netw. 32, 1 (2018), 102–111.
[116] Vladimir Moravcevic, Milan Tucic, Roman Pavlovic, and Aleksandar Majdak. 2015. An approach for uniform repre-
sentation and control of ZigBee devices in home automation software. In Proceedings of the IEEE 5th International
Conference on Consumer Electronics (ICCE’15). IEEE, 237–239.
[117] Arsalan Mosenia and Niraj K. Jha. 2017. A comprehensive study of security of internet-of-things. IEEE Trans. Emerg.
Top. Comput. 5, 4 (2017), 586–602.
[118] Carla Mouradian, Diala Naboulsi, Sami Yangui, Roch H. Glitho, Monique J. Morrow, and Paul A. Polakos. 2018. A
comprehensive survey on Fog computing: State-of-the-art and research challenges. IEEE Commun. Surveys Tutor. 20,
1 (2018), 416–464.
[119] Geoff Mulligan. 2007. The 6LoWPAN architecture. In Proceedings of the 4th Workshop on Embedded Networked Sensors.
ACM, New York, NY, 78–82.
[120] Arslan Munir, Prasanna Kansakar, and Samee U. Khan. 2017. IFCIoT: Integrated Fog cloud IoT: A novel architectural
paradigm for the future Internet of Things. IEEE Consum. Electron. Mag. 6, 3 (2017), 74–82.
[121] Arslan Musaddiq, Yousaf Bin Zikria, Oliver Hahm, Heejung Yu, Ali Kashif Bashir, and Sung Won Kim. 2018. A survey
on resource management in IoT operating systems. IEEE Access 6 (2018), 8459–8482.
[122] Karan Nair, Janhavi Kulkarni, Mansi Warde, Zalak Dave, Vedashree Rawalgaonkar, Ganesh Gore, and Jonathan Joshi.
2015. Optimizing power consumption in IoT based wireless sensor networks using bluetooth low energy. In Proceed-
ings of the International Conference on Green Computing and Internet of Things (ICGCIoT’15). IEEE, 589–593.
[123] Rick Nelson. 2017. Smart factories leverage cloud, edge computing. Eval. Eng. 56, 6 (2017), 14.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:38 L. Kong et al.

[124] Kim Thuat Nguyen, Maryline Laurent, and Nouha Oualha. 2015. Survey on secure communication protocols for the
internet of things. Ad Hoc Netw. 32 (2015), 17–31.
[125] Dennis Kengo Oka, Takahiro Furue, Lennart Langenhop, and Tomohiro Nishimura. 2014. Survey of vehicle IoT blue-
tooth devices. In Proceedings of the IEEE 7th International Conference on Service-Oriented Computing and Applications.
IEEE, 260–264.
[126] Babatunji Omoniwa, Riaz Hussain, Muhammad Awais Javed, Safdar Hussain Bouk, and Shahzad A. Malik. 2019.
Fog/Edge computing-based IoT (FECIoT): Architecture, applications, and research issues. IEEE Internet Things J. 6, 3
(2019), 4118–4149.
[127] Jose A. Onieva, Ruben Rios, Rodrigo Roman, and Javier Lopez. 2019. Edge-assisted vehicular networks security. IEEE
Internet Things J. 6, 5 (2019), 8038–8045.
[128] Pouya Ostovari, Abdallah Khreishah, and Jie Wu. 2013. Cache content placement using triangular network coding.
In Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC’13). IEEE, 1375–1380.
[129] Michael O’Keeffe. 2008. The paillier cryptosystem. Math. Dept. 18 (Apr. 2008), 1–16.
[130] Jianli Pan and James McElhannon. 2018. Future edge cloud and edge computing for internet of things applications.
IEEE Internet Things J. 5, 1 (2018), 439–449.
[131] Heejin Park, Shuang Zhai, Long Lu, and Felix Xiaozhu Lin. 2019. StreamBox-TZ: Secure stream analytics at the edge
with TrustZone. In Proceedings of the USENIX Annual Technical Conference (ATC’19). 537–554.
[132] Yao Peng, Longfei Shangguan, Yue Hu, Yujie Qian, Xianshang Lin, Xiaojiang Chen, Dingyi Fang, and Kyle Jamieson.
2018. PLoRa: A passive long-range data network from ambient LoRa transmissions. In Proceedings of the ACM Special
Interest Group on Data Communications (SIGCOMM’18). ACM, New York, NY, 147–160.
[133] Charith Perera, Chi Harold Liu, Srimal Jayawardena, and Min Chen. 2014. A survey on internet of things from
industrial market perspective. IEEE Access 2 (2014), 1660–1679.
[134] Charith Perera, Arkady Zaslavsky, Peter Christen, and Dimitrios Georgakopoulos. 2014. Context aware computing
for the internet of things: A survey. IEEE Commun. Surveys Tutor. 16, 1 (2014), 414–454.
[135] Tara Petrić, Mathieu Goessens, Loutfi Nuaymi, Laurent Toutain, and Alexander Pelov. 2016. Measurements, perfor-
mance and analysis of LoRa FABIAN, a real-world implementation of LPWAN. In Proceedings of the IEEE 27th Annual
International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC’16). IEEE, 1–7.
[136] Vitaly Petrov, Andrey Samuylov, Vyacheslav Begishev, Dmitri Moltchanov, Sergey Andreev, Konstantin Samouylov,
and Yevgeni Koucheryavy. 2018. Vehicle-based relay assistance for opportunistic crowdsensing over narrowband IoT
(NB-IoT). IEEE Internet Things J. 5, 5 (2018), 3710–3723.
[137] Juha Petäjäjärvi, Konstantin Mikhaylov, Marko Pettissalo, Janne Janhunen, and Jari Iinatti. 2017. Performance of
a low-power wide-area network based on LoRa technology: Doppler robustness, scalability, and coverage. Int. J.
Distrib. Sensor Netw. 13, 3 (2017), 1550147717699412.
[138] Pawani Porambage, Jude Okwuibe, Madhusanka Liyanage, Mika Ylianttila, and Tarik Taleb. 2018. Survey on multi-
access edge computing for internet of things realization. IEEE Commun. Surveys Tutor. 20, 4 (2018), 2961–2991.
[139] Zane D. Purvis and Alexander G. Dean. 2008. TOSSTI: Saving time and energy in TinyOS with software thread
integration. In Proceedings of the IEEE Real-Time and Embedded Technology and Applications Symposium. IEEE,
354–363.
[140] Zhijing Qin, Grit Denker, Carlo Giannelli, Paolo Bellavista, and Nalini Venkatasubramanian. 2014. A software defined
networking architecture for the internet-of-things. In Proceedings of the IEEE Network Operations and Management
Symposium (NOMS’14). IEEE, 1–9.
[141] Meikang Qiu, Sun-Yuan Kung, and Keke Gai. 2020. Intelligent security and optimization in Edge/Fog Computing.
Future Generation Computer Systems 107 (2020), 1140–1142. DOI:https://doi.org/10.1016/j.future.2019.06.002
[142] Wajid Rafique, Lianyong Qi, Ibrar Yaqoob, Muhammad Imran, Raihan Ur Rasool, and Wanchun Dou. 2020. Com-
plementing IoT services through software defined networking and edge computing: A comprehensive survey. IEEE
Commun. Surveys Tutor. 22, 3 (2020), 1761–1804.
[143] Amir M. Rahmani, Tuan Nguyen Gia, Behailu Negash, Arman Anzanpour, Iman Azimi, Mingzhe Jiang, and Pasi
Liljeberg. 2018. Exploiting smart e-health gateways at the edge of healthcare internet-of-things: A Fog computing
approach. Future Gen. Comput. Syst. 78 (2018), 641–658.
[144] Muhammad Raheel Raza, Asaf Varol, and Nurhayat Varol. 2020. Cloud and Fog computing: A survey to the con-
cept and challenges. In Proceedings of the International Symposium on Digital Forensics and Security (ISDFS’20).
IEEE, 1–6.
[145] Biljana L. Risteska Stojkoska and Kire V. Trivodaliev. 2017. A review of internet of things for smart home: Challenges
and solutions. J. Cleaner Prod. 140 (2017), 1454–1464.
[146] Rodrigo Roman, Javier Lopez, and Masahiro Mambo. 2018. Mobile edge computing, Fog et al.: A survey and analysis
of security threats and challenges. Future Gen. Comput. Syst. 78 (2018), 680–698.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:39

[147] Eyal Ronen, Adi Shamir, Achi-Or Weingarten, and Colin O’Flynn. 2017. IoT goes nuclear: Creating a ZigBee chain
reaction. In Proceedings of the IEEE Symposium on Security and Privacy (S&P’17). IEEE, 195–212.
[148] Dario Sabella, Alessandro Vaillant, Pekka Kuure, Uwe Rauschenbach, and Fabio Giust. 2016. Mobile-edge computing
architecture: The role of MEC in the internet of things. IEEE Consum. Electron. Mag. 5, 4 (2016), 84–91.
[149] Ahmad-Reza Sadeghi, Christian Wachsmann, and Michael Waidner. 2015. Security and privacy challenges in indus-
trial internet of things. In Proceedings of the ACM/EDAC/IEEE Design Automation Conference (DAC’15). IEEE, 1–6.
[150] Yuvraj Sahni, Jiannong Cao, and Lei Yang. 2019. Data-aware task allocation for achieving low latency in collaborative
edge computing. IEEE Internet Things J. 6, 2 (2019), 3512–3524.
[151] Ahsan Saleem, Abid Khan, Saif Ur Rehman Malik, Haris Pervaiz, Hassan Malik, Masoom Alam, and Anish Jindal.
2020. FESDA: Fog-enabled secure data aggregation in smart grid IoT network. IEEE Internet Things J. 7, 7 (2020),
6132–6142.
[152] Ola Salman, Imad Elhajj, Ali Chehab, and Ayman Kayssi. 2018. IoT survey: An SDN and Fog computing perspective.
Comput. Netw. 143 (2018), 221–246.
[153] Ola Salman, Imad Elhajj, Ayman Kayssi, and Ali Chehab. 2015. Edge computing enabling the internet of things. In
Proceedings of the IEEE 2nd World Forum on Internet of Things (WF-IoT’15). IEEE, 603–608.
[154] Zihao Sang, Songtao Guo, Quyuan Wang, and Ying Wang. 2021. GCS: Collaborative video cache management strat-
egy in multi-access edge computing. Ad Hoc Netw. 117 (2021), 102516.
[155] Subhadeep Sarkar, Subarna Chatterjee, and Sudip Misra. 2018. Assessment of the suitability of Fog computing in the
context of internet of things. IEEE Trans. Cloud Comput. 6, 1 (2018), 46–59.
[156] Mahadev Satyanarayanan, Pieter Simoens, Yu Xiao, Padmanabhan Pillai, Zhuo Chen, Kiryong Ha, Wenlu Hu, and
Brandon Amos. 2015. Edge analytics in the internet of things. IEEE Pervas. Comput. 14, 2 (2015), 24–31.
[157] Pallavi Sethi and Smruti R. Sarangi. 2017. Internet of things: Architectures, protocols, and applications. J. Electric.
Comput. Eng. 2017 (2017), 9324035.
[158] Mansoor Shafi, Andreas F. Molisch, Peter J. Smith, Thomas Haustein, Peiying Zhu, Prasan De Silva, Fredrik Tufvesson,
Anass Benjebbour, and Gerhard Wunder. 2017. 5G: A tutorial overview of standards, trials, challenges, deployment,
and practice. IEEE J. Select. Areas Commun. 35, 6 (2017), 1201–1221.
[159] Sajjad Hussain Shah and Ilyas Yaqoob. 2016. A survey: Internet of things (IOT) technologies, applications and chal-
lenges. In Proceedings of the IEEE Smart Energy Grid Engineering (SEGE’16). IEEE, 381–385.
[160] Majlesi Shahrbanoo, Mehrpour Ali, and Mehran Mohsenzadeh. 2012. An approach for agile SOA development using
agile principals. Retrieved from https://arxiv.org/abs/1204.0368.
[161] Cong Shi, Jian Liu, Hongbo Liu, and Yingying Chen. 2017. Smart user authentication through actuation of daily
activities leveraging WiFi-enabled IoT. In Proceedings of the ACM International Symposium on Theory, Algorithmic
Foundations, and Protocol Design for Mobile Networks and Mobile Computing (Mobihoc’17). ACM, New York, NY,
Article 5, 10 pages.
[162] Weisong Shi, Jie Cao, Quan Zhang, Youhuizi Li, and Lanyu Xu. 2016. Edge computing: Vision and challenges. IEEE
Internet Things J. 3, 5 (2016), 637–646.
[163] Kyung-Ah Shim. 2019. Universal forgery attacks on remote authentication schemes for wireless body area networks
based on internet of things. IEEE Internet Things J. 6, 5 (2019), 9211–9212.
[164] Jose Costa Sapalo Sicato, Sushil Kumar Singh, Shailendra Rathore, and Jong Hyuk Park. 2020. A comprehensive
analyses of intrusion detection system for IoT environment. J. Info. Process. Syst. 16, 4 (2020), 975–990.
[165] Rashmi Sharan Sinha, Yiqiao Wei, and Seung-Hoon Hwang. 2017. A survey on LPWA technology: LoRa and NB-IoT.
ICT Express 3, 1 (2017), 14–21.
[166] S. Smys, B. Abul, and W. Haoxiang. 2020. Hybrid intrusion detection system for internet of things (IoT). J. ISMAC 2,
4 (2020), 190–199.
[167] Alexandru Stanciu. 2017. Blockchain based distributed control system for edge computing. In Proceedings of the
International Conference on Control Systems and Computer Science (CSCS’17). IEEE, 667–671.
[168] Mark W. Storer, Kevin Greenan, Darrell D. E. Long, and Ethan L. Miller. 2008. Secure data deduplication. In Proceedings
of the ACM International Workshop on Storage Security and Survivability. ACM, New York, NY, 1–10.
[169] Hung-Min Sun, Shih-Pu Hsu, and Chien-Ming Chen. 2007. Mobile jamming attack and its countermeasure in wireless
sensor networks. In Proceedings of the International Conference on Advanced Information Networking and Applications
Workshops (AINAW’07), Vol. 1. IEEE, 457–462.
[170] Xiang Sun and Nirwan Ansari. 2016. EdgeIoT: Mobile edge computing for the internet of things. IEEE Commun. Mag.
54, 12 (2016), 22–29.
[171] Tarik Taleb, Sunny Dutta, Adlen Ksentini, Muddesar Iqbal, and Hannu Flinck. 2017. Mobile edge computing potential
in making cities smarter. IEEE Commun. Mag. 55, 3 (2017), 38–43.
[172] Tuyen X. Tran, Abolfazl Hajisami, Parul Pandey, and Dario Pompili. 2017. Collaborative mobile edge computing in
5G networks: New paradigms, scenarios, and challenges. IEEE Commun. Mag. 55, 4 (2017), 54–61.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
174:40 L. Kong et al.

[173] Fan-Hsun Tseng, Li-Der Chou, and Han-Chieh Chao. 2011. A survey of black hole attacks in wireless mobile ad hoc
networks. Hum.-centric Comput. Info. Sci. 1, 1 (2011), 4.
[174] Carlo Vallati, Antonio Virdis, Enzo Mingozzi, and Giovanni Stea. 2016. Mobile-edge computing come home connect-
ing things in future smart homes using LTE device-to-device communications. IEEE Consum. Electr. Mag. 5, 4 (2016),
77–83.
[175] M. Vellanki, S. P. R. Kandukuri, and A. Razaque. 2016. Node level energy efficiency protocol for internet of things. J.
Theoret. Comput. Sci. 3 (2016), 5.
[176] Prabal Verma and Sandeep K. Sood. 2018. Fog assisted-IoT enabled patient health monitoring in smart homes. IEEE
Internet Things J. 5, 3 (2018), 1789–1796.
[177] Massimo Villari, Maria Fazio, Schahram Dustdar, Omer Rana, and Rajiv Ranjan. 2016. Osmotic computing: A new
paradigm for edge/cloud integration. IEEE Cloud Comput. 3, 6 (2016), 76–83.
[178] G. V. Vivek and M. P. Sunil. 2015. Enabling IOT services using WIFI-ZigBee gateway for a home automation system.
In Proceedings of the IEEE International Conference on Research in Computational Intelligence and Communication
Networks (ICRCICN’15). IEEE, 77–80.
[179] Haoqin Wang, Zhen Chen, Guanping Xiao, and Zheng Zheng. 2016. Network of networks in Linux operating system.
Physica A: Stat. Mech. Appl. 447 (2016), 520–526.
[180] Tian Wang, Guangxue Zhang, Anfeng Liu, Md Zakirul Alam Bhuiyan, and Qun Jin. 2019. A secure IoT service
architecture with an efficient balance dynamics based on cloud and edge computing. IEEE Internet Things J. 6, 3
(2019), 4831–4843.
[181] Xiaoding Wang, Sahil Garg, Hui Lin, Georges Kaddoum, Jia Hu, and M. Shamim Hossain. 2020. A secure data ag-
gregation strategy in edge computing and blockchain empowered internet of things. IEEE Internet Things J. (2020),
1–1.
[182] Mohammad Wazid, Poonam Reshma Dsouza, Ashok Kumar Das, Vivekananda Bhat K. Neeraj Kumar, and Joel J. P. C.
Rodrigues. 2019. RAD-EI: A routing attack detection scheme for edge-based internet of things environment. Int. J.
Commun. Syst. 32, 15 (2019), e4024.
[183] Hua Wei, Hong Luo, Yan Sun, and Mohammad S. Obaidat. 2020. Cache-aware computation offloading in IoT systems.
IEEE Syst. J. 14, 1 (2020), 61–72.
[184] Juan Wen, Kaibin Huang, Sheng Yang, and Victor O. K. Li. 2017. Cache-enabled heterogeneous cellular networks:
Optimal tier-level content placement. IEEE Trans. Wireless Commun. 16, 9 (2017), 5939–5952.
[185] Bernd W. Wirtz, Jan C. Weyerer, and Franziska T. Schichtel. 2019. An integrative public IoT framework for smart
government. Govern. Info. Quart. 36, 2 (2019), 333–345.
[186] Junjuan Xia, Chao Li, Xiazhi Lai, Shiwei Lai, Fusheng Zhu, Dan Deng, and Liseng Fan. 2020. Cache-aided mobile edge
computing for B5G wireless communication networks. EURASIP J. Wireless Commun. Netw. 2020, 1 (2020), 1–10.
[187] Chenhan Xu, Kun Wang, Peng Li, Song Guo, Jiangtao Luo, Baoliu Ye, and Minyi Guo. 2018. Making big data open in
edges: A resource-efficient blockchain-based approach. IEEE Trans. Parallel Distrib. Syst. 30, 4 (2018), 870–882.
[188] Li Da Xu, Wu He, and Shancang Li. 2014. Internet of things in industries: A survey. IEEE Trans. Industr. Inform. 10, 4
(2014), 2233–2243.
[189] Xiaolong Xu, Qingxiang Liu, Yun Luo, Kai Peng, Xuyun Zhang, Shunmei Meng, and Lianyong Qi. 2019. A compu-
tation offloading method over big data for IoT-enabled cloud-edge computing. Future Gen. Comput. Syst. 95 (2019),
522–533.
[190] Zhiwei Xu, Lu Chao, and Xiaohui Peng. 2019. T-REST: An open-enabled architectural style for the internet of things.
IEEE Internet Things J. 6, 3 (2019), 4019–4034.
[191] Dylan Yaga, Peter Mell, Nik Roby, and Karen Scarfone. 2018. Blockchain Technology Overview. NIST Intera-
gency/Internal Report (NISTIR), National Institute of Standards and Technology, Gaithersburg, MD. DOI:https:
//doi.org/10.6028/NIST.IR.8202
[192] Shanhe Yi, Cheng Li, and Qun Li. 2015. A survey of Fog computing: Concepts, applications and issues. In Proceedings
of the Workshop on Mobile Big Data. ACM, New York, NY, 37–42.
[193] Changsheng You, Kaibin Huang, Hyukjin Chae, and Byoung-Hoon Kim. 2016. Energy-efficient resource allocation
for mobile-edge computation offloading. IEEE Trans. Wireless Commun. 16, 3 (2016), 1397–1411.
[194] Wei Yu, Fan Liang, Xiaofei He, William Grant Hatcher, Chao Lu, Jie Lin, and Xinyu Yang. 2018. A survey on the edge
computing for the internet of things. IEEE Access 6 (2018), 6900–6919.
[195] Faisal A. Zaman, Abdallah Jarray, and Ahmed Karmouch. 2019. Software defined network-based edge cloud resource
allocation framework. IEEE Access 7 (2019), 10672–10690.
[196] John K. Zao, Tchin Tze Gan, Chun Kai You, Sergio José Rodríguez Méndez, Cheng En Chung, Yu Te Wang, Tim
Mullen, and Tzyy Ping Jung. 2014. Augmented brain computer interaction based on Fog computing and linked data.
In Proceedings of the International Conference on Intelligent Environments. IEEE, 374–377.

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.
Edge-computing-driven Internet of Things: A Survey 174:41

[197] Jiale Zhang, Bing Chen, Yanchao Zhao, Xiang Cheng, and Feng Hu. 2018. Data security and privacy-preserving in
edge computing paradigm: Survey and open issues. IEEE Access 6 (2018), 18209–18237.
[198] Ma Zhaofeng, Wang Xiaochang, Deepak Kumar Jain, Haneef Khan, Gao Hongmin, and Wang Zhen. 2019. A
blockchain-based trusted data management scheme in edge computing. IEEE Trans. Industr. Inform. 16, 3 (2019),
2013–2021.
[199] Jiehan Zhou, Teemu Leppanen, Erkki Harjula, Mika Ylianttila, Timo Ojala, Chen Yu, Hai Jin, and Laurence Tianruo
Yang. 2013. CloudThings: A common architecture for integrating the internet of things with cloud computing. In
Proceedings of the IEEE International Conference on Computer Supported Cooperative Work in Design (CSCWD’13).
IEEE, 651–657.
[200] Ruogu Zhou, Yongping Xiong, Guoliang Xing, Limin Sun, and Jian Ma. 2010. ZiFi: Wireless LAN discovery via
ZigBee interference signatures. In Proceedings of the ACM Annual International Conference on Mobile Computing and
Networking (MobiCom’10). ACM, New York, NY, 49–60.

Received 2 May 2021; revised 22 July 2022; accepted 2 August 2022

ACM Computing Surveys, Vol. 55, No. 8, Article 174. Publication date: December 2022.