See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/369628636

Edge Computing Applications in Industrial IoT: A Literature Review

Chapter · March 2023

DOI: 10.1007/978-3-031-29315-3_11

CITATIONS READS

0 456

4 authors, including:

Alp Bayar Umut Sener

Middle East Technical University Middle East Technical University
2 PUBLICATIONS 0 CITATIONS 10 PUBLICATIONS 294 CITATIONS

SEE PROFILE SEE PROFILE

Kerem Kayabay
High-Performance Computing Center Stuttgart
20 PUBLICATIONS 295 CITATIONS

SEE PROFILE

All content following this page was uploaded by Alp Bayar on 07 April 2023.

The user has requested enhancement of the downloaded file.

 Edge Computing Applications in Industrial IoT:
A Literature Review

Alp Bayar1(B) , Umut Şener1 , Kerem Kayabay2 , and P. Erhan Eren1

1 Informatics Institute, Middle East Technical University, Ankara, Turkey
{alpbayar,sumut,ereren}@metu.edu.tr
2 Network Technologies Department, Tübitak Ulakbim, Ankara, Turkey

[email protected]

Abstract. The Industrial Internet of Things (IIoT) covers many computing

devices and sensors constantly generating and exchanging data in a complicated
network. Ongoing research tries to fit decentralized edge computing architec-
tures in IIoT environments with urgent computation and action requirements. This
paper investigates state-of-the-art edge computing applications based on security,
latency, resource utilization, and energy efficiency aspects as well as application
domains. Accordingly, a set of case studies for researchers and practitioners to
explore or develop applications of edge structures in industrial environments are
presented. It is identified that socio-technical issues related to organizations going
through digital transformation can point to a possible future research area.

Keywords: Edge Computing · Internet of Things · Industrial IoT · Industrial

Use Cases

1 Introduction

Internet of things (IoT) is a term coined by Kevin Ashton in 1999, enabling objects to
manage their tasks without human intervention through the data gathered from their sen-
sors and actuators. Industrial IoT (IIoT) refers to IoT applications in industrial domains
[1]. IIoT data are traditionally sent and processed in cloud-based central computing
architectures. As the volume and velocity of data increase, transmitting all the data to
a central cloud for processing becomes costly in terms of time and network resources.
Gartner predicts that around 75% of business data will be generated outside a tradi-
tional central cloud architecture by 2025. This number was around 10% in 2021 [2]. In
critical IIoT environments where a business decision requires immediate computation
and action, edge computing as a new paradigm brings faster response times. It reduces
network congestion by moving computing and storage near where data is generated.
Edge computing is a paradigm where computations occur closer to the data source.
The term edge refers to the node in the opposite direction of the cloud data center [2].
Fog computing is a similar concept introduced by Cisco, focusing more on the infras-
tructure between edge devices and central cloud servers [3]. Fog and Edge terms are used
interchangeably in this study because they try to achieve the same from the perspective

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023

J. Á. Bañares et al. (Eds.): GECON 2022, LNCS 13430, pp. 124–131, 2023.
https://doi.org/10.1007/978-3-031-29315-3_11
 Edge Computing Applications in Industrial IoT 125

of business objectives. Investigating available real-life applications and understanding

the edge architectures is significant for organizations since practitioners can understand
the technology trends and architectures for a particular IIoT edge computing use case.
While paradigms such as edge computing emerge, there is a research gap in related work
[3–7] on guiding digitally transforming organizations to utilize them strategically. These
studies review existing edge architectures, possible application domains, and objectives;
however, they lack a business strategy perspective. Accordingly, the research questions
of the study are formed:

– What kind of real-life applications of edge computing and case studies are used in the
industry regarding an IIoT environment?
– What are the main objectives of real-life edge computing applications?
– How do these objectives shape edge computing technology trends and architectures?

This paper investigates applications of edge architectures in the industry and groups
them according to their objectives and application areas. It aims to present the current
implementation status of edge technologies and trends to assist organizations in reach-
ing their business objectives. The paper is structured as follows: Sect. 2 presents the
motivation, challenges behind using edge computing in IIoT, and market trends. The
literature review methodology is explained in Sect. 3, while Sect. 4 presents the real-life
applications and case studies. Lastly, discussion and conclusion are stated.

2 Edge Computing in IIoT

This paper groups edge computing studies in IIoT according to four main objectives:
latency, security, energy efficiency, and resource utilization. Latency is a common
concern between edge applications. In an experiment [8] on a face recognition task, it
has been shown that moving applications from the cloud to the edge reduces 900 to
160 ms. in response time. With less latency, edge nodes can help monitor and control
processes [9] or machine status [10], make forecasts under uncertainty [11]. Edge com-
puting platforms are heterogeneous considering data communication, protocols, policies,
platforms, and energy consumption, and these differences result in interoperability chal-
lenges [5]. Using computationally less complex heterogeneous devices at the edge brings
vulnerabilities regarding privacy and security in the network [12], which is the second
objective the literature focuses on. For efficiently managing these platforms, there are
solutions to increase flexibility, scalability, and availability of edge devices stated as
resource utilization objectives. Finally, energy efficiency is a vital aspect to consider
in edge architecture s while utilizing edge nodes, referring to the energy consumption of
edge devices [13]. Report in [14] identifies deployment strategies as full edge provider,
partner edge provider, aggregator edge provider, and limited edge provider [14]. Know-
ing these roles and strategies in the market is essential to achieve business goals and
forming an edge computing adoption strategy.
126 A. Bayar et al.

3 Research Method
This study has been structured based on Kitchenham’s guidelines [15] in order to find
case studies and real-life examples of edge computing applications regarding IIoT envi-
ronments. Snowballing approach is combined with a database search. The following
keywords are determined as a starting point of the search: “Edge,” “Fog,” “Comput-
ing,” “Industrial,” and “IoT.” Web of Science and IEEE Explore are selected for search
databases. Search results are stored in MS Excel. 67 initial results were found after
searching the Web of Science Core Collection. Only four of them were released before
2017, and the number of published papers had increased by the year of 2021. 15 studies
were selected after the initial elimination. After fully reading these articles, and includ-
ing references of the references of those studies, 28 papers were selected as Primary
Studies (PS). The studies focusing on edge computing use cases, applications, and archi-
tectures in IIoT, covering on algorithms, tools, benefits, and implementation challenges
were regarded as PS, and 16 of them are journal articles, while 11 studies are conference
papers, and one is a book chapter. Authors conducted iterative meetings for identify-
ing main objectives and trends for grouping papers. Domain-specific IoT articles with
no edge computing implementation were excluded from the search. Studies targeting
domains other than manufacturing such as smart cities, autonomous driving, and health-
care, were excluded. The papers that include technical solutions were included if they
explain a particular case study application in IIoT domain.

4 Case Studies and Applications

This section reviews applications of edge architectures in the industry and groups them
according to objective and application areas. Studies are explained regarding how the
architectures and trends are shaped to fulfill the objectives, allowing practitioners to have
an idea while developing implementations for their cases. Studies focusing on security,
latency, resource utilization, and energy efficiency are presented in Table 1. Objectives
are shown in bold, while trends are shown in italic.
There are many studies related to blockchain and edge computing convergence in
the literature concerning the security of IIoT systems, such as [12, 24, 31]. An approach
that uses blockchain and context-aware security for IIoT environments was proposed
in [17], and implementation in an additive manufacturing site was presented. When
data is provided with blockchain instead of cloud, it reduces communication costs and
increases bandwidth efficiency. [16] integrates blockchain for increasing security in an
agroindustry platform to monitor and support decisions in a dairy farm. ZigBee was used
to connect IoT gateways and sensors and a Raspberry Pi as edge node for preprocessing
IoT data and forwarding it to the cloud. A data processing framework is proposed in
[25], enabling secure data storage using edge in IIoT systems. Data management and
encryption challenges are summarized, and solutions proposed were evaluated using
simulations in a prototype to monitor the temperature in a factory. [18] implemented
security application in a simulated smart factory to observe the effectiveness under
cyber security scenarios. An edge computing smart grid architecture has been proposed
in [27], where data load balancing can be achieved with low latency and increased
security.
 Edge Computing Applications in Industrial IoT 127

Table 1. Papers are grouped by their objective and applications

Papers Main Objective Application Domain

[16] Security Agricultural Monitoring
[17] Security Additive Manufacturing
[18] Security Simulated Factory
[19] Resource Utilization Real-Time Gas Pressure Control
[20] Resource Utilization Predictive Maintenance
[7] Latency + Energy Efficiency Active Maintenance
[13] Latency + Energy Efficiency Software Defined Network
[21] Energy Efficiency + Resource Utilization Air Quality Monitoring
[22] Energy Efficiency + Resource Utilization Smart Manufacturing System
[6] Security + Latency Active Maintenance
[23] Security + Latency Smart Grid
[24] Security + Latency Simulation
[25] Security + Resource Utilization Factory Temperature Monitoring
[26] Latency + Security + Resource Conveyor Routing, Distributed Predictive
Utilization Maintenance
[27] Latency + Resource Utilization Simulation with real IIoT Data
[28] Security + Resource Utilization Real-Time Machine Data Analytics
[29] Latency + Resource Utilization Image Classification Simulation
[30] Latency + Resource Utilization Real-Time Anomaly Detection
[31] Latency + Resource Utilization Numerical Experiments
[32] Latency + Resource Utilization Vertical Plant Wall System
[33] Latency + Resource Utilization Industrial Robotics
[34] Latency + Resource Utilization + Energy Smart Manufacturing
Efficiency

Energy efficiency and resource utilization in terms of computational workload

are vital aspects to consider while utilizing edge nodes. Software Defined Networking
is widely used for managing a network of middleware devices, as reviewed in [13],
where the trade-off between energy efficiency and latency is evaluated. [35] proposes a
technical architecture that leverages SDN with Network Function Virtualization (NFV)
and serverless architectures to reduce the high energy consumption of edge architec-
tures. An adaptive data transmission algorithm using SDN in edge computing for IIoT
is proposed in [34] to find an optimal route for traffic load, task deadlines, and energy
consumption. To achieve an industrial internet, [22] proposed a framework consisting of
high-level embedded microcontrollers and gateway systems. With the help of distributed
computing, the gateway efficiently performs network management, data collection, and
communication, considering power consumption and providing better scalability than
128 A. Bayar et al.

traditional IIoT solutions. A similar problem is addressed in [27], which obtains opti-
mized scheduling of IIoT data according to priorities. [7] proposes an architecture to
resolve latency and boost energy efficiency in manufacturing utilizes SDN in the net-
work domain. In the proposed architecture, application domain provides monitoring
and control services; data domain provides data cleaning and feature extraction using
Hadoop; and the network domain utilizes SDN and Time Sensitive Networking (TSN) to
manage devices such as Raspberry Pi connected with OPC UA. This edge architecture is
compared to the existing private cloud on a candy packaging production line in terms of
productivity [6]. Although the network’s speed decreased from 16MB/s to 6 MB/s after
switching to edge, the results show that edge provides more productivity in high volume
mass production. To monitor the real-time status of machinery and conduct predictive
maintenance, a database is created in [20], small enough to fit in memories of edge
devices using Python SQLite.
Containerized edge architectures are evaluated in [32] regarding industrial require-
ments, measuring round trip time, bandwidth, processing capabilities, and latency while
doing machine learning tasks for predictive maintenance. Microsoft Azure IoT Edge is
utilized for running container applications on Raspberry Pi. Results show that container-
ization does not decrease performance while increasing flexibility and scalability. A
manufacturing process control system has been proposed in [36] to monitor production
lines and collect and analyze data to increase efficiency. Stream data were collected from
sensors through communication adapters working on OPC UA and MTConnect proto-
cols. Edge node, streaming data in real-time, provides control signals. [21] implemented
an air quality monitoring system using data from Arduino sensors spread across a univer-
sity campus. Low Power Wide Area Network (LoRA) gets data from sensors, transmits
data to the LoRa gateway, then to an edge gateway of Kubernetes minion installed on a
Raspberry Pi for final unified delivery to the data center. MQTT protocol enables send-
ing alerts to devices if there are anomalies in the data. An open-source architecture for
industrial networks called IFog4.0 is proposed in [19] with case studies in an emulated
gas regulation station environment. A Fog-Management module has been developed to
manage Docker containers. Fog Computing Platform reference architecture is proposed
for IIoT applications in [26], using open standards, OPC UA and TSN. A machine is
provided with packages containing tags, and the system delivers them to the destination
by accessing a database through reading the tag. Network configurations and the benefits
of using an edge architecture in different use cases of the same architecture are explained
in detail in [37]. Similar fog-based industrial robotic systems are proposed in [33] and
[28].
Utilizing deep learning requires high computation power and bandwidth. In an edge
architecture tailored for deep learning [29], complexity is optimized in line with the
computational capacity of edge devices. To evaluate the solution, authors formed a
convolutional neural network using real-world IIoT data. They applied experiments that
reduce network traffic while maintaining the model’s classification accuracy in an object
identification task of 30 different components. One way of processing deep learning in the
edge is by inferencing, executing pre-trained models with newly generated visual content
from mobile edge devices. [38] formulated the inference offloading problem to minimize
energy consumption and evaluated the performance using simulations. In order to use
 Edge Computing Applications in Industrial IoT 129

deep learning for anomaly detection, performance of different architectures are tested
in [30]. Trade-off of choosing the architecture considering scalability, bandwidth, and
delay have been presented. The author concludes that scaling cloud computation power
results in full cloud outperforming the edge.

5 Discussion and Conclusion

In this study, a comprehensive literature review was conducted for edge computing archi-
tecture and applications to business cases in IIoT. The main contributions of this paper
are stated as follows: (I) Explanations of case studies retrieved from the literature are
provided as guidance for organizations and (II) Different application domains are inves-
tigated regarding security, latency, resource utilization, and energy efficiency objectives.
Additionally, trends, motivation, and implementation challenges of edge computing are
explained for both researchers and professionals. Furthermore, this study presents tai-
lored approaches for effective resource utilization in terms of efficient computation needs
of the businesses such as complex deep learning applications or simpler monitoring tasks.
Although there are plenty of studies focusing on when and why to implement edge
architectures, all example applications are experimental. The main research highlights
are explained as below:

• While integrating different approaches, there is a trade-off between increasing security,

reducing latency, increasing energy consumption, and making the network more com-
plex with potential interoperability problems [24]. For example, deploying blockchain
to increase security results in more energy consumption.
• Further trends of edge computing are identified: Blockchain has been incorporated into
IIoT networks for security. Approaches such as SDN, containerization, and offloading
algorithms are used to increase resource utilization and energy efficiency.

There are some limitations of this study. First, snowballing may have reduced the
reproducibility of the search process, and it is likely to obtain further examples. Secondly,
a limited number of non-academic resources, such as reports of institutes or consulting
agencies, have been reviewed to present trends in edge computing. Also, grouping papers
by their objectives and identified trends may contain subjectivity.
Findings show that there is no framework or guidance for selecting and implement-
ing an appropriate edge computing architecture. In future work, we plan to investigate
technology management approaches to bridge the socio-technical knowledge gaps in the
literature.

References
1. What is the Internet of Things (IoT)? https://www.oracle.com/tr/internet-of-things/what-is-
iot/. Accessed 20 Dec 2021
2. What edge computing means for infrastructure and operations leaders. https://www.gartner.
com/smarterwithgartner/what-edge-computing-means-for-infrastructure-and-operations-lea
ders. Accessed 20 Dec 2021
130 A. Bayar et al.

3. Qiu, T., Chi, J., Zhou, X., Ning, Z., Atiquzzaman, M., Wu, D.O.: Edge computing in industrial
internet of things: architecture, advances and challenges. IEEE Commun. Surv. Tutorials 22,
2462–2488 (2020)
4. Yu, W., et al.: A survey on the edge computing for the internet of things. IEEE Access. 6,
6900–6919 (2018)
5. Khan, W., Ahmed, E., Hakak, S., Yaqoob, I., Ahmed, A.: Edge computing: a survey. Future
Gener. Comput. Syst. 97, 219–235 (2019)
6. Chen, B., Wan, J., Celesti, A., Li, D., Abbas, H., Zhang, Q.: Edge computing in IoT-based
manufacturing. IEEE Commun. Mag. 56, 103–109 (2018)
7. Chalapathi, G.S.S., Chamola, V., Vaish, A., Buyya, R.: Fog/Edge Computing For Security,
Privacy, and Applications, pp. 293–325. Springer, Cham (2021)
8. Yi, S., Hao, Z., Qin, Z., Li, Q.: Fog computing: platform and applications. In: 3rd IEEE
Workshop on Hot Topics in Web Systems and Technologies (HotWeb), Washington DC,
USA, pp. 73–78 (2015)
9. Li, L., Ota, K., Dong, M.: Deep learning for smart industry: efficient manufacture inspection
system with fog computing. IEEE Trans. Industr. Inf. 14, 4665–4673 (2018)
10. Bose, S.K., Kar, B., Roy, M., Gopalakrishnan, P.K., Basu, A.: ADEPOS: anomaly detection
based power saving for predictive maintenance using edge computing. In: Proceedings of
the 24th Asia and South Pacific Design Automation Conference, pp. 597–602. ACM, Tokyo
(2019)
11. Taïk, A., Cherkaoui, S.: Electrical load forecasting using edge computing and federated learn-
ing. In: ICC 2020 - 2020 IEEE International Conference on Communications (ICC), Dublin,
Ireland, pp. 1–6 (2020)
12. Wu, Y., Dai, H.-N., Wang, H.: Convergence of blockchain and edge computing for secure
and scalable IIot critical infrastructures in industry 4.0. IEEE Internet Things J. 8, 2300–2317
(2021)
13. Kaur, K., Garg, S., Aujla, G.S., Kumar, N., Rodrigues, J.J.P.C., Guizani, M.: Edge computing
in the industrial internet of things environment: software-defined-networks-based edge-cloud
interplay. IEEE Commun. Mag. 56, 44–51 (2018)
14. Brava, C., Backström, H.: Whitepaper on edge computing deployment strategies. https://
www.ericsson.com/en/reports-and-papers/white-papers/edge-computing-and-deployment-
strategies-for-communication-service-providers. Accessed 20 Dec 2021
15. Kitchenham, B.: Procedures for performing systematic reviews. Keele, UK, 33 (2004)
16. Sittón-Candanedo, I., Alonso, R.S., Corchado, J.M., Rodríguez-González, S., Casado-Vara,
R.: A review of edge computing reference architectures and a new global edge proposal.
Futur. Gener. Comput. Syst. 99, 278–294 (2019)
17. Portal, G., de Matos, E., Hessel, F.: An edge decentralized security architecture for industrial
IoT applications. In: IEEE 6th World Forum on Internet of Things (WF-IoT), New Orleans,
USA, pp. 1–6 (2020)
18. Güven, E.Y., Çamurcu, A.Y.: Edge computing security application: Kılıç. In: 3rd Interna-
tional Conference on Computer Science and Engineering (UBMK), Sarajevo, Bosnia and
Herzegovina, pp. 248–253 (2018)
19. Ghazi Vakili, M., Demartini, C., Guerrera, M., Montrucchio, B.: Open source fog architecture
for industrial IoT automation based on industrial protocols. In: IEEE 43rd Annual Computer
Software and Applications Conference (COMPSAC), Milwaukee, WI, USA, pp. 570–578
(2019)
20. Oyekanlu, E.: Predictive edge computing for time series of industrial IoT and large scale crit-
ical infrastructure based on open-source software analytic of big data. In: IEEE International
Conference on Big Data (Big Data), Boston, MA, USA, pp. 1663–1669 (2017)
 Edge Computing Applications in Industrial IoT 131

21. Kristiani, E., Yang, C.-T., Huang, C.-Y., Wang, Y.-T., Ko, P.-C.: The implementation of a
cloud-edge computing architecture using OpenStack and Kubernetes for air quality monitor-
ing application. Mobile Netw. Appl. 26(3), 1070–1092 (2020). https://doi.org/10.1007/s11
036-020-01620-5
22. Chen, C.-H., Lin, M.-Y., Liu, C.-C.: Edge Computing gateway of the industrial internet of
things using multiple collaborative microcontrollers. IEEE Network 32, 24–32 (2018)
23. Okay, F.Y., Ozdemir, S.: A fog computing based smart grid model. In: International Sympo-
sium on Networks, Computers and Communications (ISNCC), Yasmine Hammamet, Tunisia,
pp. 1–6 (2016)
24. Kumar, T., et al.: BlockEdge: blockchain-edge framework for industrial IoT networks. IEEE
Access. 8, 154166–154185 (2020)
25. Fu, J.-S., Liu, Y., Chao, H.-C., Bhargava, B.K., Zhang, Z.-J.: Secure data storage and searching
for industrial IoT by integrating fog computing and cloud computing. IEEE Trans. Industr.
Inf. 14, 4519–4528 (2018)
26. Pop, P., et al.: The FORA fog computing platform for industrial IoT. Inf. Syst. 98, 101727
(2021)
27. Chekired, D.A., Khoukhi, L., Mouftah, H.T.: Industrial IoT data scheduling based on hierar-
chical fog computing: a key for enabling smart factory. IEEE Trans. Industr. Inf. 14, 4590–4602
(2018)
28. Denzler, P., Ruh, J., Kadar, M., Avasalcai, C., Kastner, W.: Towards consolidating industrial
use cases on a common fog computing platform. In: 25th IEEE International Conference
on Emerging Technologies and Factory Automation (ETFA), Vienna, Austria, pp. 172–179
(2020)
29. Liang, F., Yu, W., Liu, X., Griffith, D., Golmie, N.: Toward edge-based deep learning in
industrial internet of things. IEEE Internet Things J. 7, 4329–4341 (2020)
30. Ferrari, P., et al.: Performance evaluation of full-cloud and edge-cloud architectures for Indus-
trial IoT anomaly detection based on deep learning. In: II Workshop on Metrology for Industry
4.0 and IoT (MetroInd4.0 IoT), Naples, Italy, pp. 420–425 (2019)
31. Lee, C.K.M., Huo, Y.Z., Zhang, S.Z., Ng, K.K.H.: Design of a smart manufacturing system
with the application of multi-access edge computing and blockchain technology. IEEE Access.
8, 28659–28667 (2020). https://doi.org/10.1109/ACCESS.2020.2972284
32. Liu, Y., Lan, D., Pang, Z., Karlsson, M., Gong, S.: Performance evaluation of containerization
in edge-cloud computing stacks for industrial applications: a client perspective. IEEE Open
J. Ind. Electron. Soc. 2, 153–168 (2021)
33. Shaik, M.S., et al.: Enabling fog-based industrial robotics systems. In: 25th IEEE International
Conference on Emerging Technologies and Factory Automation (ETFA), Vienna, Austria,
pp. 61–68 (2020)
34. Li, X., Li, D., Wan, J., Liu, C., Imran, M.: Adaptive transmission optimization in SDN-based
industrial internet of things with edge computing. IEEE Internet Things J. 5, 1351–1360
(2018)
35. Djemame, K.: Energy efficiency in edge environments: a serverless computing approach. In:
Tserpes, K., et al. (eds.) Economics of Grids, Clouds, Systems, and Services, pp. 181–184.
Springer International Publishing, Cham (2021). https://doi.org/10.1007/978-3-030-92916-
9_15
36. Wu, D., et al.: A fog computing-based framework for process monitoring and prognosis in
cyber-manufacturing. J. Manuf. Syst. 43, 25–34 (2017)
37. Barzegaran, M., et al.: Fogification of electric drives: an industrial use case. In: 25th
IEEE International Conference on Emerging Technologies and Factory Automation (ETFA),
Vienna, Austria, pp. 77–84 (2020)
38. Xu, Z., et al.: Energy-aware inference offloading for DNN-driven applications in mobile edge
clouds. IEEE Trans. Parallel Distrib. Syst. 32, 799–814 (2021)

View publication stats