Edgecomputing
Edgecomputing
Abstract— Edge computing is a relatively recent phenomenon servers, storage, databases, and applications are all located
in the computing world, which takes cloud computing services in the cloud, so you can access them on any device if you
closer to the end user and is distinguished by fast processing have the Internet [2]. Besides, there is no need to prepare a
and application response time, which leads to many advantages server or use a data center, which can significantly reduce
such as faster and more efficient data processing; Safety;
costs. It is possible to store a huge amount of information in
reduced leakage on existing networks. Moreover, delay-
sensitive applications may benefit from the Edge computing the cloud, but on the other hand, there is a concern that data
paradigm's low latency, agility, and location awareness. processing will take time. Cloudlet addresses the issue of
Significant research has been conducted in the field of Edge end-to-end responsiveness between mobile devices and
computing, which is reviewed in terms of recent technologies cloud which was a challenge in the cloud. Fog computing
such as Mobile Edge Computing, Cloudlet, and Fog computing, "fog" means "fog" and was named because it is "closer to
allowing researchers to gain a better understanding of current the device than the cloud". In edge computing, processing
and potential solutions. This article summarizes the edge power is at the edge of the network, cloud computing has
computing paradigm classification, applications, and various processing power on the Internet, while fog computing has
challenges in detail.
processing power in the LAN. Therefore, scale of data
Keywords—Edge computing; Cloud computing, Mobile Edge handled is larger than that of edge computing, which is good
Computing, Fog Computing, Mobility, Security. at personal data processing, and it can perform functions
like mini-cloud computing that can optimize the amount of
I. INTRODUCTION data. Mobile Edge computing placed the edge nodes on the
network and enhanced to include the fixed network as
With the escalation of the Internet of Things (IoT), Cloud shown in fig 2.
computing, which is a collection of networks, has
tremendously changed the living and working style.
Traditional cloud computing as centralized is accosting
severe major challenges, such as high latency in real-time
applications, low spectral efficiency (SE), and a non-
adaptive machine type of communication [1]. Propelled to
explain all these difficulties, new technology is driving a
pattern that moves the capacity of incorporated distributed
computing to the edge devices of the organization or
network. Edge computing is also recognized as edge
processing. It is a type of network communication technique
to distribute the load on the system by placing a huge
number of servers near users and devices, In simple words
"putting servers at edge network closer to the device". In
other words, an alternative to storing and processing all the Fig. 1: Edge Computing Paradigm
data on the cloud, a fragment of it is handled by the The IoT, defined as a distributed network of heterogeneous
administering platform near the terminal. This makes it devices (sensors, actuators, and processors) that exchange
possible to extremely digest traffic on the Internet and messages between them without human intervention, has
eliminates communication delays. There are various types of attracted particular interest from researchers. With the
network techniques, and in addition to edge computing, increase in the number of services, other needs also arise.
there are "cloud computing" and "fog computing “and all of Some of them, and perhaps the most critical in the
these computing techniques processes the data differently, architecture intended for IoT, are the latency of data and
so let us see how it differs from edge computing [1]. Cloud efficiency in how resources are used, and data is
computing is named a "cloud" because it remotely processes manipulated.
a computer in a location away from the user via the Internet.
As shown in fig. 1, In cloud computing, services such as
254
978-1-6654-1450-0/21/$31.00 ©2021 IEEE
Authorized licensed use limited to: BOURNEMOUTH UNIVERSITY. Downloaded on July 01,2021 at 00:13:38 UTC from IEEE Xplore. Restrictions apply.
2021 2nd International Conference on Intelligent Engineering and Management (ICIEM)
Tab.1: Three Edge Computing Technologies Yuan Ai et al. (2017) presented the major three edge
Cloudlet Mobile Edge Fog computing technologies: mobile edge computing, cloudlets,
Computing Computing Computing
and fog computing. The authors explained application areas,
Architect Single layer Single layer Single or architectures, standardization efforts for mobile edge
ure multilayer computing, cloudlets, and fog computing. [4]. S.
Latency Medium Medium Low Bhattacharyya et al. (2017) the authors described the Edge
computing that processes the gathered data from end devices
Mobility Medium Medium High
at the edge of the network. By covering a large range of
Context Low High Medium technologies, edge computing addresses the various concern
Awarene as battery life constraint, bandwidth usage, latency, data
ss security, and data privacy. The need for edge computing
Manage Centralized Hierarchical Hierarchical
ment (Push from Cloud Services and Pull from the Internet of
Usage Resource 5G requirements IoT Things) is discussed [6]. Salman Taherizadeh et al. (2017)
Intensive in the authors explained the auto-scaling applications in edge
applications telecommunicatio computing which maintained the online services at a
n network decentralized location. They broadly explained two aspects
of this paper. In the first section, they major focused on the
II. RELATED WORK different types of edge computing applications (IoT
Applications, Micro-service applications, Time-critical
In this research article a review has been provided on the applications). For these applications, auto-scaling challenges
overview of edge computing based on various issues and when the workload dynamically changes are discussed. In
challenges of edge computing for diversified applications. the second section of this paper, container-based
This main work aims to summarize edge computing and to visualization auto-scaling technologies are discussed. Self-
make aware of its challenges with modern technologies. adaptive application at runtime enhances the performance.
Weisong Shi et al. (2016) the authors explained the In the classification of auto-scaling applications in edge
definition of edge computing which addresses the concern computing, the authors explained the cloud framework,
of response time where concerns are latency, resource like Virtualization Technology, Monitoring approach,
battery-life constraint, bandwidth cost-saving, as well as Operational behavior, Adjustment ability, Architectural
data safety and privacy [2]. Fang Liu et al. (2019) support, Image Delivery, and scalability techniques [7].
summarized the existing edge computing systems and JunjieXie et al. (2020) addressed the issue of low latency
related tools. The authors divided the paper into two parts: requirements in mobile edge computing. The author
System View and Application View. In system View, Open proposed a fast data-sharing framework HDS (Hybrid data
Source Edge computing projects and edge computing sharing) to meet the requirements of low latency by dividing
systems & tools are discussed wherein application view, the gathered data location service into two regions: Intra-
deep learning optimization at the edge are discussed. [3]. region and inter-region. With the Hybrid information
255
Authorized licensed use limited to: BOURNEMOUTH UNIVERSITY. Downloaded on July 01,2021 at 00:13:38 UTC from IEEE Xplore. Restrictions apply.
2021 2nd International Conference on Intelligent Engineering and Management (ICIEM)
sharing system which comprises 100 areas, the creator switches, gateway, access points which store and compute
accomplished low latency, low usage overhead, and 50.21% all the data provided by the end devices. High-performance
more limited query ways, and 92.75% fewer false positives. servers are placed in the third layer, the cloud layer which is
In this whole network, the total edge server used was 1000 widely used in global applicable situations and
to 10000 [14]. Batyr Charyyev et al. (2020) the authors organizations that require large-scale centralized processing.
performed the cloud edge latency comparison. They Edge computing industry alliance (ECC) released
performed an extensive measurement to assess the latency framework 3.0. Multi-layer functionality as reference
characteristics of end-users to the edge servers and cloud architecture is shown in fig 3.
data centers. It estimated latency from 8,456 end-clients to
6,341 Akamai edge workers and 69 cloud areas. At last, the IV. APPLICATION AREAS OF EDGE COMPUTING
paper's outcome is that while 58% of end-clients can arrive
at a close by edge worker in under 10 ms, just 29% of end- It is proposed to bring some of the cloud services closer to
clients get a comparable dormancy from a close by cloud the level at which the data are found as well as using
area. [18]. N.Saranya et al. (2020) the authors reviewed the multiple levels, according to the needs of the different
various data latency techniques in Mobile Edge Computing. applications. Due to multiple benefits of edge computing
As in the centralized cloud, one cannot achieve low latency. like speed, security, scalability, versatility, and reliability
Mobile Edge Computing makes efficient use of the edge computing is used in many real time applications.
resources and decreased the movement of large data Basically, edge computing is used to push applications and
generated by the edge devices. Edge computing is important data to logical extremes by centralized hubs in a network.
for solving fatal situations such as Conflicts in Autonomous We use edge computing in real time applications because
vehicles, Fire, Environmental Hazards. [20]. better responsiveness and robustness make this system more
effective. Edge computing is distributed kind of system
Tab.2: Comparison of Techniques on various Parameters where large data volume is processed and compiled.
Ref. Scalability Resource Security and Distribution properties gives features like supply chain
Latency
Constraint Privacy
tracking, point of sale system and distributed artificial
[8] intelligence processing.
256
Authorized licensed use limited to: BOURNEMOUTH UNIVERSITY. Downloaded on July 01,2021 at 00:13:38 UTC from IEEE Xplore. Restrictions apply.
2021 2nd International Conference on Intelligent Engineering and Management (ICIEM)
257
Authorized licensed use limited to: BOURNEMOUTH UNIVERSITY. Downloaded on July 01,2021 at 00:13:38 UTC from IEEE Xplore. Restrictions apply.
2021 2nd International Conference on Intelligent Engineering and Management (ICIEM)
like glasses or a wristwatch, it is used in fields such as problem is that the mobile edge computing system
health management, medical fields, and industry. needs to consider. In the initial phase, systems
Besides, smart watches can be used in conjunction with considered the functionalities rather than
smartphones act as wearable devices and are becoming implementation issues like the price model. With the
popular among consumers as edge computing that upgrade of edge processing, the genuine formation
utilizes mobile devices. concern ought to be illuminated before we appreciate
all the advantages.
C. Human Behaviour Analysis Service
As the face recognition system, the wearable device is a C. Security and Privacy
mechanism that performs image processing of face Security plays important role in every field as every
recognition at the edge part and analyses it in user is concerned about privacy for their confidential
cooperation with the data on the cloud. Wearable information. While working in a network, various
devices can predict and track the movement of a person attacks can be performed on the edge network. The
extracted from camera images; the analysis data of the following is different types of security threats:
tracking is processed in the temporary memory of the • Phishing - An attempt to extract sensitive or
computer. The created coordinate data is aggregated on confidential information from the user of cloud-
the cloud for each camera, and besides this becomes a connected IoT equipment. Cybercriminals use this
flow line of people, and it is used for the analysis of method to distribute electronic content to a wide
purchasing behavior at retail stores. range of victims, prompting them to take specific
actions, such as clicking a link or sending a reply to
D. Edge Computing as the savior of the IoT Era an email. If they do so, they will provide personal
In modern times, vast amounts of data are stored in information such as username, password, bank or
cloud and processed on the cloud, such as the spread of financial details, or credit card details without their
IoT, digitization, diversification of data, and quality knowledge. Phishing distribution methods include
improvement. However, since the amount of data is email, online advertising, and SMS. A
increasing year by year, the processing on the cloud phishing type is spear phishing - an attack aimed at
cannot be completed in time, and scalability problems specific people or groups of people, e.g., corporate
(delay in data processing) will occur. Meanwhile, edge management boards, celebrities, or high-ranking
computing, which distributes communication traffic, government officials.
has come to be used, and needs are increasing • Malware - Malware that is specially designed to
significantly in scenes where real-time data processing harm an edge device, system, or data. It includes
is required. Therefore, edge computers can be said to be several types of tools, including adware, ransom
the savior of the IoT era and will increase the value of ware, Trojans, viruses, and worms.
big data [5]. • Rootkit - Malware that enables secretly accessing
parts of an edge device, software, or system. It
VI. CHALLENGES IN EDGE COMPUTING often modifies the hardware operating system in
such a way that it remains hidden from the user.
Edge computing is still in its infancy. The major challenges • Trojan - A form of non-replicating malware that
in Edge Computing have been classified as: contains hidden functionality. The Trojan usually
A. Bandwidth does not try to propagate or inject itself into other
The more data at the network bandwidth-edge, the more files.
the bandwidth will change. It should balance the greater • Backdoor - Malware targeting remote entry into a
bandwidth across the network. From the latency’s point system, device, edge device, or software, or
of view, the high bandwidth can reduce the network bypassing traditional security measures installed in
transmission time, especially for the huge amount of them.
data. In cloud higher bandwidth provided at data centers • Spyware - Malware that spies on an edge device
and lower at edge but this will be challenge at edge user, intercept keystrokes, emails, documents, and
network to provide higher bandwidth to end devices. If even turns on a video camera without their
the workload can be handled at the edge, the delay can knowledge.
be improved, and bandwidth will also be saved [6].
D. Uncompromising Quality of Service (Latency)
B. Multi-domain Management and Cooperation Locating the compute at the edge is one of the parameters
Multiple types of resources are involved in Edge that provide QoS. In edge computing, the focus is
computing, where all resources belong to various performing the computing anywhere close to the data
owners. How to access all the required resources for collected by the computer. To check the network
application and allocate them according to the need of performance, usually Latency is vital QoS metrics which is
the user’s application, services. In an emergency, a not only determined by computation time. The computation
258
Authorized licensed use limited to: BOURNEMOUTH UNIVERSITY. Downloaded on July 01,2021 at 00:13:38 UTC from IEEE Xplore. Restrictions apply.
2021 2nd International Conference on Intelligent Engineering and Management (ICIEM)
of the workload should be performed at closer layer. Also, with Transmission Power Control and Virtual Machine Migration,”
IEEE Trans. Comput., vol. 67, no. 9, pp. 1287–1300, 2018.
this layer must have enough computation capability to do
[10] R. Bruschi, F. Davoli, P. Lago, and J. F. Pajo, “A Multi-Clustering
the calculations at the edge network. Approach to Scale Distributed Tenant Networks for Mobile Edge
Computing,” IEEE J. Sel. Areas Commun., vol. 37, no. 3, pp. 499–
E. Partition and Offloading tasks (resource allocation) 514, 2019.
[11] S. Maheshwari, D. Raychaudhuri, I. Seskar, and F. Bronzino,
Workload allocation is not an easy task. Distributed
“Scalability and performance evaluation of edge cloud systems for
computing should consider location as an additional aspect latency constrained applications,” Proc. - 2018 3rd ACM/IEEE
of computation. Resource mobility defines how to Symp.Edge Comput.SEC 2018, pp. 286–299, 2018.
dynamically discover the resources in the network and [12] S. H. Mortazavi, M. Salehe, C. S. Gomes, C. Phillips, and E. De
Lara, “CloudPath: A multi-Tier cloud computing framework,” 2017
manage all the required available resources, including long-
2nd ACM/IEEE Symp. Edge Comput.SEC 2017, 2017.
term resources and short-term resources. Also, in some [13] K. Kaur, S. Garg, G. S. Aujla, N. Kumar, J. J. P. C. Rodrigues, and
situations, few edge devices or end devices are damaged, M. Guizani, “Edge Computing in the Industrial Internet of Things
then at that situation challenge is how the system can be Environment: Software-Defined-Networks-Based Edge-Cloud
Interplay,” IEEE Commun. Mag., vol. 56, no. 2, pp. 44–51, 2018.
resumed as soon as possible with the best replacements [6].
[14] J. Xie, D. Guo, X. Shi, H. Cai, C. Qian, and H. Chen, “A Fast
At the edge, the resources are resource-constrained as Hybrid Data Sharing Framework for Hierarchical Mobile Edge
having smaller processors and a limited power budget, also Computing,” Proc. - IEEE INFOCOM, vol. 2020-July, pp. 2609–
resource heterogeneous where the processors are with 2618, 2020.
[15] W. Xia, J. Zhang, T. Q. S. Quek, S. Jin, and H. Zhu, “Mobile Edge
different architectures (Data Flow, Control and Tenancy)
Cloud-Based Industrial Internet of Things: Improving Edge
and workload need to change dynamically [29]. Intelligence with Hierarchical SDN Controllers,” IEEE Veh.
Technol. Mag., vol. 15, no. 1, pp. 36–45, 2020.
VII. CONCLUSION [16] M. Uddin, S. Mukherjee, H. Chang, and T. V. Lakshman, “SDN-
Based multi-protocol edge switching for IoT service automation,”
In this paper, we came up with the introduction of edge
IEEE J. Sel. Areas Commun., vol. 36, no. 12, pp. 2775–2786, 2018.
computing, Applications, Architecture, and benefits of edge [17] T. P. Raptis, A. Passarella, and M. Conti, “Industrial Edge
computing over centralized computing. Most of the Networks,” vol. 38, no. 5, pp. 915–927, 2020.
applications require prompt response, so edge computing [18] B. Charyyev and M. Gunes, “Latency Characteristics of Edge and
Cloud,” no. June 2020.
process the data at the edge of the network that eliminates
[19] A. Singh, G. S. Aujla, and R. S. Bali, “Intent-Based Network for
the roundtrip travel time provides a real-time response and Data Dissemination in Software-Defined Vehicular Edge
native authority. At last, we put forward the edge computing Computing,” IEEE Trans. Intell. Transp. Syst., pp. 1–9, 2020.
challenges. In the future, with the persistent advancement of [20] Z. Ning et al., “Intelligent Edge Computing in Internet of Vehicles:
A Joint Computation Offloading and Caching Solution,” IEEE
the Internet and human culture, edge processing will assume
Trans. Intell. Transp. Syst., pp. 1–14, 2020.
a more significant job in various applications like smart [21] X. Ren, G. S. Aujla, A. Jindal, R. S. Batth and P. Zhang, "Adaptive
homes, smart widgets, intelligent transport, quick audio, and Recovery Mechanism for SDN Controllers in Edge-Cloud supported
video streaming. FinTech Applications," in IEEE Internet of Things Journal (2021).
https://doi.org/10.1109/JIOT.2021.3064468
[22] T. G. Rodrigues, K. Suto, H. Nishiyama, N. Kato, and K. Temma,
REFERENCES “Cloudlets Activation Scheme for Scalable Mobile Edge Computing
with Transmission Power Control and Virtual Machine Migration,”
[1] P. Singh, A. Kaur, G. S. Aujla, R. S. Batth and S. Kanhere, 2020 IEEE Trans. Comput., vol. 67, no. 9, pp. 1287–1300, 2018.
"DaaS: Dew Computing as a Service for Intelligent Intrusion [23] S. Kumar and R. H. Goudar, “Cloud Computing – Research Issues,
Detection in Edge-of-Things Ecosystem," in IEEE Internet of Things Challenges, Architecture, Platforms, and Applications: A Survey,”
Journal, https://doi.org/10.1109/JIOT.2020.3029248 Int. J. Futur. Comput. Commun. vol. 1, no. 4, pp. 356–360, 2012.
[2] W. Shi, J. Cao, Q. Zhang, Y. Li, and L. Xu, “Edge Computing: [24] Kumar VDA, Kumar A, Batth RS, “Efficient data transfer in edge
Vision and Challenges,” IEEE Internet Things J., vol. 3, no. 5, pp. envisioned environment using artificial intelligence based edge node
637–646, 2016. algorithm”, Transactions on Emerging Telecommunications and
[3] F. Liu, G. Tang, Y. Li, Z. Cai, X. Zhang, and T. Zhou, “A Survey on Technologies.,2020;e4110. https://doi.org/10.1002/ett.4110
Edge Computing Systems and Tools,” Proc. IEEE, pp. 1–24, 2019. [25] Q. Zheng and Z. Ping, “Simulation study on latency-aware network
[4] Y. Ai, M. Peng, and K. Zhang, “Edge computing technologies for in edge computing,” 2019 6th Int. Conf. Control. Decis. Inf.
Internet of Things: a primer,” Digit. Commun.Networks, vol. 4, no. Technol. CoDIT 2019, pp. 150–155, 2019.
2, pp. 77–86, 2018. [26] Y. C. Hu, M. Patel, D. Sabella, N. Sprecher, and V. Young, “Mobile
[5] B. Varghese, N. Wang, S. Barbhuiya, P. Kilpatrick, and D. S. Edge Computing A key technology towards 5G,” no. 11, 2015.
Nikolopoulos, “Challenges and Opportunities in Edge Computing,” [27] H. Liu, F. Eldarrat, H. Alqahtani, A. Reznik, X. De Foy, and Y.
Proc. - 2016 IEEE Int. Conf. Smart Cloud, Smart Cloud 2016, pp. Zhang, “Mobile edge cloud system: Architectures, challenges, and
20–26, 2016. approaches,” IEEE Syst. J., vol. 12, no. 3, pp. 2495–2508, 2018.
[6] S. Bhattacharyya, “Research on Edge Computing : A Detailed [28] Wang, C., Batth, R.S., Zhang, P., Aujla, G.S., “VNE solution for
Study,” Academia.Edu, vol. 2, no. 6, pp. 9–13, 2016. network differentiated QoS and security requirements: from the
[7] S. Taherizadeh and V. Stankovski, “Auto-scaling applications in perspective of deep reinforcement learning”. Computing (2021).
edge computing: Taxonomy and challenges,” ACM Int. Conf. https://doi.org/10.1007/s00607-020-00883-w
Proceeding Ser., pp. 158–163, 2017. [29] M. C. Computing, X. Chen, L. Jiao, and W. Li, “Efficient Multi-
[8] C. Feature and E. Deep, “Private and Scalable Personal Data User Computation Offloading for,” IEEE/ACM Trans. Netw., vol.
Analytics Cloud Deep Learning,” 2018. 24, no. 5, pp. 2795–2808, 2016.
[9] T. G. Rodrigues, K. Suto, H. Nishiyama, N. Kato, and K. Temma, [30] C.-H. Hong and B. Varghese, “Resource Management in Fog/Edge
“Cloudlets Activation Scheme for Scalable Mobile Edge Computing Computing,” ACM Comput. Surv., vol. 52, no. 5, pp. 1–37, 2019
259
Authorized licensed use limited to: BOURNEMOUTH UNIVERSITY. Downloaded on July 01,2021 at 00:13:38 UTC from IEEE Xplore. Restrictions apply.