Task offloading in Edge and Cloud Computing: A survey on mathematical, artificial intelligence and control theory solutions
Computer Networks 195 (2021) 108177
A
1
Contents lists available at ScienceDirect
Computer Networks
journal homepage: www.elsevier.com/locate/comnet
Survey paper
Task offloading in Edge and Cloud Computing: A survey on mathematical,
artificial intelligence and control theory solutions
Firdose Saeik a, Marios Avgeris b, Dimitrios Spatharakis b, Nina Santi c, Dimitrios Dechouniotis b,
John Violos a, Aris Leivadeas a,∗, Nikolaos Athanasopoulos d, Nathalie Mitton c,
Symeon Papavassiliou b
a Department of Software and Information Technology Engineering, École de Technologie Supérieure, Montréal, Canada
b Department of Electrical and Computer Engineering, National Technical University of Athens, Greece
c INRIA, France
d School of Electronics, Electrical Engineering and Computer Science, Queen’s University, Belfast, UK
A R T I C L E I N F O
Keywords:
Edge Computing
Task offloading
Resource allocation
Control theory
Mathematical optimization
Artificial intelligence
A B S T R A C T
Next generation communication networks are expected to accommodate a high number of new and resource-
voracious applications that can be offered to a large range of end users. Even though end devices are becoming
more powerful, the available local resources cannot cope with the requirements of these applications. This
has created a new challenge called task offloading, where computation intensive tasks need to be offloaded
to more resource powerful remote devices. Naturally, the Cloud Computing is a well-tested infrastructure
that can facilitate the task offloading. However, Cloud Computing as a centralized and distant infrastructure
creates significant communication delays that cannot satisfy the requirements of the emerging delay-sensitive
applications. To this end, the concept of Edge Computing has been proposed, where the Cloud Computing
capabilities are repositioned closer to the end devices at the edge of the network. This paper provides a detailed
survey of how the Edge and/or Cloud can be combined together to facilitate the task offloading problem.
Particular emphasis is given on the mathematical, artificial intelligence and control theory optimization
approaches that can be used to satisfy the various objectives, constraints and dynamic conditions of this
end-to-end application execution approach. The survey concludes with identifying open challenges and future
directions of the problem at hand.
1. Introduction
Wireless communications have come a long way over the last
40 years allowing a plethora of new applications and services to
proliferate. This wireless growth has revolutionized the way humans
and machines interact with each other and between them. Specifically,
as wireless technologies evolve, the data rate, mobility, coverage and
spectral efficiency rapidly increase [1], permitting radical changes
on the grounds of our society and our personal communication. At
the same time, with the advent of the Internet of Things (IoT) and
emergent applications such as Virtual Reality (VR) and driverless cars,
the demand for wireless communications with even higher-speeds
and ubiquitous connectivity becomes a necessity that requires more
efficient wireless communication systems.
∗ Corresponding author.
E-mail addresses: firdose.saeik. .ca (F. Saeik), .gr (M. Avgeris), .gr (D. Spatharakis),
nina. (N. Santi), .gr (D. Dechouniotis), .gr (J. Violos), aris. (A. Leivadeas),
n. .uk (N. Athanasopoulos), nathalie. (N. Mitton), .gr (S. Papavassiliou).
5G is an exemplary wireless communication system that tries to
minimize the gap between the new emergent applications and their
high-performance requirements. Specifically, 5G promises the support
of increased bandwidth and connection density, as well as low-latency
communication, with the induction of the enhanced mobile broadband
(eMMB), the massive machine-type communication (mMTC) and the
ultra-reliable low latency communication (uRLLC) services [2]. How-
ever, even though the performance of the wireless access networks
continues to increase, allowing the support of new and more intelli-
gent applications, the end devices cannot always cope with the strict
computational requirements of these resource voracious applications.
Inevitably, the answer to where we can find an increased avail-
ability of computational resources, accompanied with the necessary
vailable online 18 May 2021
389-1286/© 2021 Elsevier B.V. All rights reserved.
https://doi.org/10.1016/j.comnet.2021.108177
Received 19 October 2020; Received in revised form 24 January 2021; Accepted 1
4 May 2021
http://www.elsevier.com/locate/comnet
http://www.elsevier.com/locate/comnet
mailto:firdose.saeik. .ca
mailto: .gr
mailto: .gr
mailto:nina.
mailto: .gr
mailto: .gr
mailto:aris.
mailto:n. .uk
mailto:nathalie.
mailto: .gr
https://doi.org/10.1016/j.comnet.2021.108177
https://doi.org/10.1016/j.comnet.2021.108177
http://crossmark.crossref.org/dialog/?doi=10.1016/j.comnet.2021.108177&domain=pdf
Computer Networks 195 (2021) 108177F. Saeik et al.
reliability to offer a seamless communication for the wireless applica-
tions, always lies around the Cloud. The Cloud is a well tested and
used solution that can extend the resource capabilities of the end
devices with powerful data center topologies. Besides, Cloud is well
equipped with the appropriate automation tools and platforms in order
to offer the necessary transparency to the end devices, while hiding the
complexity and the logistic details of this resource extension.
Hence, the practice of offloading computation intensive tasks of
resource-intensive applications from the end devices to a centralized
Cloud infrastructure, is a well explored solution [3][4]. Nonetheless,
as the focus of new applications turned towards high throughput and
low latency communications, the Cloud started to expose its inherent
limitations. The long distance between the end devices and the Cloud
infrastructure, the use of a best-effort and unreliable intermittent trans-
port network, the cost of traversing the backhaul network and the
increased security surface throughout this long communication path,
created the need for alternative solutions.
There is no question, that these substitute solutions should introduce
a more distributed infrastructure that will enhance the local efficiency
by bringing Cloud-like capabilities closer to the end devices, at the
edge of the network. This is exactly how the term Edge Computing
was coined. Even though, multiple flavors of the Edge Computing exist
(e.g., Fog Computing, Mobile Cloud Computing, Cloudlet, Mobile Edge
Computing), they all agree that additional and existing computational
and networking resources at the edge of the network should be inserted
and regrouped.
This new infrastructural component that creates an additional re-
source layer between the end devices and the Cloud, is able to re-
duce the increased bandwidth consumption at the backhaul, transport
and Cloud networks and also reduce the communication delay and
support applications with real time requirements. In particular, end
devices are now capable of offloading their resource-intensive tasks
to a nearby Edge device, thus minimizing the overall execution time
without adding excessive communication paths towards a distant Cloud
infrastructure. This solution, called task offloading, allows enhancing
the user’s experience by providing lower latency, better reliability and
improved energy efficiency for battery-powered devices.
Even though the notion of Edge Computing exists for almost a
decade, the problem of task offloading has only recently started to
be investigated. Nonetheless, it has gained a lot of attention from the
industry and the academic community, leading to the publication of
many scientific and research papers over the last couple of years. A
great effort has also been made to classify and categorize the different
types of task offloading by a number of surveys and tutorials.
These surveys focused on multiple aspects such as architecture [5–
7], resource allocation [8,9], communication [8,10,11], caching [10],
mobility management [6,10,12], integration with wireless, IoT and 5G
technologies [5,8,13,14], decision on task offloading [6,11], applica-
tion partitioning [12,15], application models [8,12,16,17], application
scenarios [5,8,10,15,18,19] and algorithms [11,12,17].
In this survey paper, we also attempt to study the task offloading
problem, emphasizing, however, on novel algorithmic and control ap-
proaches. Thus, in contrast with recent surveys on task offloading, our
contribution is twofold; firstly, we provide a comprehensive survey
of task offloading within three subfields: (i) Optimization algorithms
(ii) Artificial Intelligence techniques and (iii) Control theory; secondly,
a categorization of the above techniques is provided based on their
objective function, the granularity level, the use of the Edge and/or
Cloud infrastructures and the incorporation of mobility in the overall
solution, depending on the type of the edge devices.
This paper is organized as follows. Section 2 presents an overview
of the various computing paradigms and relevant technologies evolved
in the last decade, along with some potential use cases for task of-
floading of interactive applications. Following, Section 2.2.6 formally
defines the task offloading problem along with the challenges in-
2
volved. Section 4 covers different task offloading solutions that have
been recently proposed, emphasizing on the mathematical models,
optimization techniques, machine learning algorithms and control the-
ory approaches. Section 5 presents the open challenges. Finally, we
conclude and provide suggestions for future work in Section 6.
2. Computing paradigms: Overview & use cases
As already discussed, over the last two decades Cloud Computing
has been the dominant service delivery paradigm. However, modern
applications come with strict requirements which cannot be met via
execution in remote Cloud resources (e.g., ultra low delay). Thus,
the current trend of resource provisioning is to augment the edge
of the network with computing capabilities. Towards this direction,
the emerging service model of Edge Computing promises to mitigate
the limitations of Cloud Computing. To clarify the ambiguity behind
the terminology and architectures used in the literature, this section
provides the fundamental elements of the various modern computing
infrastructures such as Cloud Computing, Mobile Cloud Computing,
Mobile Edge Computing and Fog Computing. Furthermore, emerging
use cases concerning task offloading at the Edge and/or Cloud are
presented.
2.1. Modern computing paradigms
2.1.1. Cloud computing
Cloud Computing has revolutionized the Internet and completely
transformed the way that applications, software and resources are
offered to the end users. According to NIST [20], Cloud Computing
is defined as ‘‘a model for enabling convenient, on-demand network
access to a shared pool of configurable computing resources (e.g., net-
works, servers, storage, applications and services) that can be rapidly
provisioned and released with minimal management effort or service
provider interaction”. The Cloud paradigm brings unique benefits. In
particular, with this model, computing resources are offered to the end
users on demand, in a self-service fashion, independent of the type of
the device and the location of the user. Furthermore, the computational
and network resources available at the Cloud can be shared and dynam-
ically scaled. This is achieved by adopting virtualization as the enabling
technology of the Cloud, allowing the resources to be allocated and
released with minimal interaction, while users pay for the service they
consume according to its usage [21].
Despite bringing numerous advantages, Cloud Computing poses
some serious limitations. These limitations, although they exist since
the beginning of the Cloud, they did not surface until recently. The
reason is that new communication technologies, new applications and
services have increased the data volume generated and at the same
time also increased the demands for low latency communications.
Hence, offering Cloud Computing resources in a centralized manner
far away from the users, can create serious delay bottlenecks. This
delay can be disastrous for mission critical applications, such as health
related applications, or time-critical applications like real-time control
in manufacturing. Another disadvantage is that forwarding the traffic
from the end devices to the Cloud, usually involves traversing the
core Internet. This can create three serious problems. Firstly, sending
hundreds of TB of data from the devices to the Cloud can certainly
create traffic hotspots in the Internet infrastructure, which can further
affect the communication delay. Secondly, the existence of various
different networks and administrative domains between the Cloud and
the front-end devices, can create an unstable and intermittent network
connectivity. Finally, the data, before being sent to the Cloud, proba-
bly have to traverse a backhaul network (e.g., cellular and satellite).
This backhaul network may be costly and lossy, creating additional
problems to this end-to-end communication. Of course, there may
be additional limitations, regarding for example the security aspects
of the communication, since this end device-to-Cloud communication
increases the surface of threat. However, in the particular survey we
emphasize only on the networking and data processing limitations.
Computer Networks 195 (2021) 108177F. Saeik et al.
2.1.2. Mobile cloud computing
Apparently, Cloud Computing can be used for offloading tasks from
mobile devices to a more powerful infrastructure. This approach cre-
ated the notion of Mobile Cloud Computing (MCC). A Mobile Cloud
is defined as a mobile device that can execute a resource-intensive
application on a distant high-performance compute server or compute
cluster and support thin client user interactions with the application
over the Internet [22–25]. MCC can be thus described as the inte-
gration of mobile devices with Cloud Computing technology. It offers
computing, storage, services and applications over the Internet and the
typical advantages found in a Cloud Computing environment such as
cost reduction and resource flexibility. In addition, MCC can potentially
save energy for mobile users by offloading high-energy consuming
applications to the Cloud [21].
However, such an approach still carries the typical Cloud limitations
presented above. Thus, the concept of MCC can be modified to offer
the necessary Cloud resources closer to the mobile devices. This new
flavor of MCC is called Cloudlet [26] and it allows the mobile devices
to offload their workload to a local ‘‘mini cloud”, consisting of multiple
multi-core hardware equipment directly connected to an Access Point
(AP) or Base Station (BS). Therefore, Cloudlet can be seen as a trusted,
resource-rich computer or computer cluster, which is connected to the
Internet and is available for use from mobile devices in proximity.
Due to the sheer proximity of Cloudlet, sharp interactive response for
immersive applications that magnify human cognition is much easier
to attain. Instead of depending on a remote server, a mobile user
instantiates a ‘‘Cloudlet” on the local network and uses it via a wireless
LAN. These proposed Cloudlets can be placed in common areas such as
railway stations, restaurants and coffee shops, so that mobile devices
could connect to them and act as a thin client. This opposes to the
use of a centralized Cloud server that would raise issues of latency and
bandwidth.
2.1.3. Fog computing
Fog Computing is another approach for expanding the Cloud Com-
puting concept to the edge of the network, thus enabling a new range
of apps and services [27,28]. Fog Computing was the first industry
initiative to explicitly define an architecture for applying utility Cloud
at the edge of the network, and was standardized by the OpenFog
consortium [29]. Specifically, the term Fog Computing was coined
by Cisco in 2012 and is defined as ‘‘the process of extending Cloud
Computing capabilities at the edge of the network. Fog incorporates
computing, storage and network resources close to the IoT layer to
facilitate the data processing’’. [27,30]. From the previous definition
it becomes evident that Fog Computing was introduced in order to
facilitate the monitoring, control and analysis of IoT devices in real
time, removing the long communication delay between the IoT devices
and the central analytics application servers in a remote Cloud.
Hence, Fog Computing expands Cloud Computing by installing lo-
calized computing facilities at the user’s premises, delivering Cloud
data to mobile users with fast local connections. The aim is to process
in part workload and services locally on Fog devices (such as hardened
routers, switches, IP video cameras), rather than being transmitted
to the Cloud [28]. As such, Fog Computing introduces an interme-
diate infrastructure layer between mobile users and Cloud, in order
to support low-latency and high-speed services. Moreover, Fog Com-
puting can support and promote applications that do not suit the
Cloud [31], such as (i) applications involving very low and consistent
latency, (ii) geographically distributed systems such as pipeline control
and sensor networks (iii) mobile applications like smart connected
vehicles and (iv) large-scale adaptive control systems, such as smart
energy delivery and smart traffic lights. As such, in the literature,
the typical applications usually combined with the Fog Computing
are mostly IoT related [32–34], cache networks [35] and immersive
media services (AR/VR, a 360-degree video and free-viewpoint video)
3
applications [36].
2.1.4. Edge computing
Fog Computing has managed to bypass many of the limitations of
Cloud Computing, increasing the performance of IoT and mobile ap-
plications in terms of task offloading. However, stringent requirements
such as ultra-low latency, user experience, stability and high reliability
have created the need for even higher localized information near end
users. Thus, Edge Computing is another similar concept, that can be
defined as a network layer encompassing end devices and their users, in
order to provide local computing capabilities on sensors, smart meters
or other network-accessible devices. Following the same mentality
where Cloud can be found in a distant location far away from the end
user and Fog can be found closer to the end user, Edge Computing
has also been associated with the term Mist Computing. As the name
suggests, Mist Computing covers the computational and communication
capacity available on the same level with the end devices. According
to NIST [37], Mist Computing is defined as a lightweight and primitive
type of Fog Computing which resides at the very edge of the network,
bringing the layer of Fog Computing closer to the smart end devices.
Mist Computing uses microcomputers and microcontrollers to feed into
nodes of Fog Computing and theoretically into centralized (Cloud)
Computing.
In light of this, Mobile Edge Computing (MEC) was developed as
a key technology to assist wireless networks with Cloud Computing-
like capabilities to provide low-latency and context-aware services
directly from the network Edge [38–48]. Mobile Edge Computing,
lately renamed as Multi-Access Edge Computing, was initiated under
the umbrella of the European Telecommunication Standards Institute
(ETSI) [47]. A key objective of the ETSI initiative is to standardize the
APIs between the mobile Edge platform and the application service sce-
narios (augmented reality (AR), mixed reality (MR), intelligent video
acceleration and Internet of Things gateway) and promote innovation
in an open environment [38]. ETSI’s reference architecture is largely
based on the concept of Network Function Virtualization (NFV), where
MEC applications can be offered as Virtualized Network Functions
(VNFs).
2.1.5. Computing paradigms comparison
Almost all of the paradigms discussed above have as a common
ground that they are offering remote computational and communi-
cation capabilities to the end devices. Furthermore, except from the
Cloud, the rest of the paradigms are able to offer these capabilities
at the edge of the network, as close to the end devices as possible.
Nonetheless, there are some differences between them.
First of all, in terms of available resources, as we move farther from
the end devices the available resources increase in quantity, with the
Cloud having practically infinite capacity. Since a public Cloud may
have dozens of data centers, each equipped with hundreds of servers,
around the world, there is no actual problem of resource depletion. In
contrast, in Fog, MCC, and MEC infrastructures, resource availability is
mostly limited due to the fact that they are comprised of micro-data
centers with few servers of lower capabilities than the ones that we
usually find in the Cloud. On top of that, in edge infrastructures, we
also find heterogeneous hardware resources with even lower resource
availability such as wireless routers and gateways, street cabinets and
Raspberry Pi’s.
Secondly, delay can be another factor of comparison between the
different infrastructures. As mentioned before, Cloud Computing is not
always a feasible solution for providing low latency communication. To
this end, the available infrastructure at the edge of the network is the
most favorable option to reduce the communication delay. Nonetheless,
since there are various levels of Edge at the WAN, LAN, or access net-
work, different levels of delays can be produced, regarding where the
computing resources are located. Obviously, going at the level of the
access network, i.e., the extreme Edge or Mist, the delay is minimized
since we do not have to account propagation and transmission delays
involved in traversing the LAN, WAN or a backhaul network. However,
Computer Networks 195 (2021) 108177F. Saeik et al.
in this case, another factor rises; that of the energy consumption. The
available hardware at the Mist is usually battery supplied, imposing the
double burden of both limited resources and limited lifetime.
Thirdly, some of the paradigms were conceived under the scope of
providing computational resources to specific applications. For exam-
ple, Fog Computing was introduced to facilitate some of the top IoT
application domains and vertical markets, such as energy, industry,
transportation, agriculture, and healthcare [29]. On the other hand,
MCC is mostly associated with providing remote computational re-
sources to mobile applications, while MEC introduces the necessary
flexibility to host multiple applications in the areas of video analyt-
ics, location services, IoT, augmented reality, optimized local content
distribution and data caching among others [49]. Particular emphasis
should be placed on the uniqueness of the MEC. In particular, even
though MEC is oriented to cellular Radio Access Networks (RAN), it can
be practically applied to any kind of access network. Furthermore, the
way that MEC has been defined and standardized by ETSI, promotes
an open environment where third-party developers, application and
service providers can all participate together towards expediting the
introduction of new applications targeting to respond to emerging user
requirements.
Finally, the decision of which paradigm to follow, usually includes
the requirements of security and confidentiality. Certainly, Cloud Com-
puting as a popular and successful technology, has many safeguards and
tools to provide a certain level of security and confidentiality. Nonethe-
less, several security threats still exist making Cloud-based security an
active open-challenge. Additionally, sending data to the Cloud over the
Internet can be susceptible to attacks. In contrast, by employing an
Edge infrastructure, the necessary security and confidentiality can be
attained since the data of the end devices usually stay within the local
network.
From the above, the pros and cons of each computing paradigm
can be extracted. Even though the MCC, Fog, and MEC can overcome
certain limitations of the Cloud, they usually cannot be offered as a
standalone solution. In other words, the notion of Edge Computing in
general, did not emerge to replace the Cloud but rather to complement
it. Thus, it is very important to create collaborative solutions (possibly
utilizing more than two computing paradigms) that will enable a
smooth continuum from the end device to the Cloud, with the goal to
satisfy the stringent requirements of novel and future applications.
2.2. Use cases
Following the above definitions of the computing paradigms and
the respective infrastructures, in this part of the survey we refer to
some typical applications that leverage task offloading at the available
resources at the edge of the network in order to increase their per-
formance. These real-world applications can range from simple data
processing to immersive multimedia applications. Following, we briefly
describe the role of task offloading in the particular set of applications.
2.2.1. Immersive applications
Current developments in computer vision have made possible the
launching of mixed reality applications, such as VR and AR, that can
offer immersive experience even in wireless environments. At the same
time, the development of increasingly advanced mobile devices such
as smart glasses, can help us identify objects, superimpose contextual
knowledge on our field of vision and create a three-dimensional view of
the surrounding environment. As these devices become smarter, more
and more sensor data in our environment can be aggregated, pro-
cessed and served, requiring however high bandwidth and low response
time communications. Hence, task offloading can be an advantageous
solution for this type of applications.
Specifically, both Cloud and Edge-based task offloading mechanisms
can be used in AR/VR applications [36,50–61]. The objectives include
reducing the energy consumption in mobile devices, increasing the
speed of computation intensive operations, reducing the average CPU
load to overcome computation intensive tasks and improving the user’s
4
Quality of Experience (QoE).
2.2.2. Autonomous vehicles
Similar to the immersive media services, autonomous vehicles is
another type of application that task offloading can be utilized. The key
objectives here are to reduce the latency and the transmission cost and
increase the efficiency of traffic management. Use case-oriented ser-
vices concerning autonomous driving include: Highway Pilot, Parking
Pilot, Fully Automated Vehicle and Vehicle on Demand [62].
Edge Computing is considered as the key technology in connected
vehicles, adding computation capabilities and geo-distributed services
to BSs and Edge devices distributed on the roadside. The idea is to
analyze data from proximate vehicles and roadside sensors and broad-
cast messages to drivers at a very low latency [63]. For example, in
an intelligent transportation system, low-level devices can be used for
the decision-making processes of the transportation [64]. Specifically,
the decision-making tasks can be distributed to Edge devices instead of
sending all the data to a centralized server. Moreover, task offloading
can enable real-time traffic management [65].
2.2.3. Robotics
Very complex robotic applications have been emerging during the
last decade, related among others to autonomous mobile agents, manip-
ulators and collaborative tasks. Efficient, safe and autonomous robot
operation in manufacturing, health care, learning and exploration,
requires running computation and memory intensive algorithms related
to image processing, planning, localization, mapping and autonomous
learning. Consequently, during the last few years, task offloading is
gaining attention that has lead to the new paradigms of Cloud, Edge
and Fog robotics [66–70].
Specifically, many offloading opportunities emerge in planning and
SLAM (simultaneous localization and mapping) algorithms for robotic
manipulators [71,72], mobile robots [73–77] and learning in gen-
eral [78,79], among others. It is worth noting that there are already
available commercial products that allow task offloading in robotic
applications [80–82].
2.2.4. Video streaming
In general, the video streaming use cases fall under the content
delivery network (CDN) [83] category. The key objective of CDN
networks is to reduce the cost and the number of bits transmitted over
the network, by maintaining an adequate QoE [84]. The mechanisms
to reduce the overall cost and traffic while providing a high QoE in
applications ranging from simple video streaming to HTTP, to Adap-
tive BitRate (ABR) and 360-degree video applications, can be further
improved by applying task offloading techniques.
Offloading can be implemented on Cloud-based solutions, where
appropriate resource allocation techniques can be used to increase
user satisfaction [85] or deployment costs of the CDN networks [84].
Nonetheless, task offloading at the Edge can supplement the achieved
performance. For example, multi-user mobile media delivery can be
enhanced by enabling the gateways (i.e., BSs) to perform appropriate
scheduling strategies [86]. An Edge infrastructure can also be used to
facilitate the caching and transcoding mechanisms in a distributed fash-
ion [87]. Regarding latency, data compression tasks can be offloaded
at the Edge [88], removing the burden of local compression models
and reducing at the same time the application response time [89].
Task offloading can be partially implemented by differentiating flows
based on their quality and performing the video compression at the
Edge, only for the high-quality video flows (e.g., 360-degree video
streaming) [90].
2.2.5. IoT
The impact of task offloading can be maximized in the context of IoT
applications. The reason is that the IoT devices are usually constrained
in terms of available resources and battery capacity. Inevitably, only
small and non resource demanding tasks can be executed locally. Task
offloading in IoT usually focuses on reducing task execution time,
Computer Networks 195 (2021) 108177F. Saeik et al.
response time and energy consumption. IoT use cases that can ben-
efit from task offloading can refer to health, agriculture, smart city,
industry and energy related applications among others.
IoT, from the very beginning, has been largely based on Cloud-
centric approaches in order to offload the tasks of data processing and
analysis of massive data produced from millions of IoT devices [91].
However, the long delays added from the Cloud, combined with the
introduction of new mission critical IoT applications, has pushed the
academic and industrial community to take advantage of the Edge con-
cept. Hence, local IoT Clouds have emerged, with the goal to maximize
the number of offloaded tasks that can be executed in close proximity
to the IoT devices [92] and to maximize the battery lifetime of the de-
vices [93,94]. However, the scalability issues of the IoT market which
is currently consisted of dozens of billions of devices often necessitates
a Cloud–Edge collaboration during the task offloading [95–97].
2.2.6. Physical disaster management
In case of disaster management, the process of task offloading
is crucial since it affects the efficiency of the rescue operations. In
addition, the network can be unstable and simply offloading tasks to
the Cloud could be difficult and require too much time. So, optimal
offloading strategies to local services, rather than remote Clouds, would
allow for precious time saving and preservation of battery of mobile
phones, sensors and autonomous agents in the field.
Unmanned Aerial Vehicles (UAVs), which possess great mobility
and versatility, are at the core of disaster management scenario by
providing situational awareness and computing resources. But, as they
are battery-powered, they cannot undertake the full computation of
all the involved data and need to offload tasks to near Edge Comput-
ing servers. This challenge is addressed in different papers [98–100].
In general, task offloading in the context of disaster remains little
explored [101,102].
3. Task offloading & challenges
In the previous sections, we have provided a short description of
the task offloading, its infrastructural components, and the importance
of this solution for new and emerging scenarios. In this part of the
survey, a more detailed definition of task offloading is provided, while
also, the typical objectives, the performance evaluation metrics, and the
challenges encountered during task offloading are presented.
Generally, task offloading can be defined as the transfer of resource-
intensive computational tasks to an external, resource-rich platform
such as the ones used in Cloud, Edge or Fog Computing. Offloading
the whole or part of the set of tasks to another processor or server,
can be used to accelerate resource-intensive and latency-sensitive ap-
plications [65,90,103]. Task offloading is a complex process and can
be affected by a number of different factors [24]. In particular, this
process involves application partitioning, offloading decision making
and distributed task execution [4,15,104]. A typical infrastructure in-
volved in an offloading scenario is illustrated in Fig. 1. From this,
it becomes evident that the network’s Edge infrastructure creates an
additional resource layer between the end devices and the external
platform. This layer is capable of reducing bandwidth consumption
on the backhaul, transport and Cloud networks, thus reducing any
communication delays, supporting applications with real-time require-
ments, improving the energy efficiency and consequently increasing the
lifetime of battery-powered devices. At this point, we should note that,
for the rest of the paper, we refer to the term Edge Computing as the
whole set of resources that can be found at the edge of the network,
including the Fog and Edge nodes.
5
3.1. Granularity levels of task offloading
Task offloading aims at optimizing the offloading of computation
intensive tasks from the end user device to a remote site, under various
computational, communication and mobility constraints. The process
of task offloading, as shown in Fig. 2, consists of (i) various hardware
components, such as end user devices and Edge/Cloud devices, (ii) mul-
tiple computing processes, including task splitting and computational
processing either locally or remotely and (iii) networking components
for transferring data between the hardware components involved.
In more detail, as Fig. 2 illustrates, a mobile device can execute
an application comprising of multiple tasks. The end device, through a
task splitting process, decides which of these tasks should be executed
locally and which ones should be offloaded to the Edge or Cloud
infrastructures. This decision is based on a plethora of factors that are
presented in the following sections, including the QoS requirements and
battery lifetime of the device, among others. Following, the tasks that
are to be executed remotely are transferred through the wireless access
network to the gateway and from there to a remote physical machine
(either at the Edge or Cloud), where they are executed following an
appropriate computational approach (e.g., creating a VM or container).
At the same time, the tasks that remain on the device are executed
locally using the available computational resources of the end device.
The last step is to combine the results of both local and remote executed
tasks to provide the final output of the application. Based on this
process, we describe the different types of task offloading according to
the task splitting decision taken, i.e., the granularity level, as follows:
3.1.1. Partial offloading at the edge
In this type of offloading, part of the computing tasks is processed
locally, while the rest is offloaded to the Edge. Partial offloading is
typically the most effective, since it can benefit from both local and
remote resources. Nonetheless, another level of complexity is added
since it needs to be decided and scheduled which tasks should be
offloaded while taking into account the possible energy and resource
constraints of the end device.
3.1.2. Full offloading at the edge
In this case, all of the computing tasks are offloaded and processed
at the Edge. Full offloading is usually translated into a simple resource
allocation problem, where tasks can be executed on virtual machines
or containers at the Edge. Energy-savings at the end devices can be
maximized, however we need to take into account other sources of
energy dissipation such as the transmission power of the device. Finally,
a precise network path from the device to the Edge site, where the
tasks are offloaded, has to be set up carefully, so as to comply with
the possible QoE/QoS constraints.
3.1.3. Partial/full offloading at the edge and cloud
During this type of offloading, a collaboration between the Edge and
Cloud infrastructures is established in order to execute the offloaded
tasks. This type of collaboration can be proved advantageous in large-
scale scenarios where the available Edge resources are not enough
to host all of the tasks offloaded from the end users. Herein, the
main challenge is the two-level task offloading decision. If a partial
offloading mechanism is followed, the first level of decision lies on
the local device, in order to decide the set of tasks that needs to be
offloaded at a remote location. In other words, the local device has to
decide which tasks can be executed locally in the device and which
ones should be offloaded either at the Edge or the Cloud. On the tasks
decided to be offloaded, a second-level of decision will be performed. In
particular, the Edge will perform a second task partitioning, regardless
of the type of offloading (i.e., partial or full) to determine the subset of
tasks to be executed at the Edge and the subset of tasks to be executed
at the Cloud. In the latter case, particular attention should be paid on
the transport network that facilitates the interconnection of the two
infrastructures and the delay constraints that it may impose.
Computer Networks 195 (2021) 108177F. Saeik et al.
Fig. 1. Infrastructure components during task offloading.
Fig. 2. Task offloading process.
3.2. Mobility of end devices
End device mobility is one of the most critical components when
it comes to task offloading decision. End devices can either be con-
sidered as static or mobile for the time window which spans between
initiating and finishing the offloading of their tasks. In the latter case,
6
mobility adds another level of splitting decision, as it needs to be
decided at which Edge site should the tasks be offloaded while the
user is on the move (or not). Even though mobility is considered
a challenge, it can generate a number of opportunities for the task
offloading. First of all, it can initiate a load balancing technique to
allow the system to provide the necessary services in distributed Edge
Computer Networks 195 (2021) 108177F. Saeik et al.
site scenarios [45,105,106]. Secondly, complementing mobility with
appropriate prediction solutions can enhance the system’s capacity, by
finding the potential next associated BS/AP of the user [87,107,108].
This can be even more beneficial in a dense scenario, where the system
can analyze the active users and their mobility patterns and allocate
the resources in an online manner to existing and newly requested
services. Moreover, mobility can benefit from handover mechanisms
that can enable service migrations between BSs and their Edge servers.
However, as the requirements of zero millisecond handover are studied
by the 5G community, mobility with prediction mechanisms is starting
to gain attention, in order to predict beforehand where the tasks should
be offloaded. This behavior can be decisive for the overall performance
in dense Edge deployments with multiple mobile users [109]. Based on
the above, we firstly describe the different types of end devices accord-
ing to their mobility level. Then, we present some mechanisms from
the literature on how the Edge infrastructures adapt to the potential
mobility of the offloading devices.
3.2.1. Static (low-mobility)
In this situation, the devices are considered static or relatively
static (low levels of mobility) during the task offloading procedure.
Apparently, this type of device mobility is considered trivial when it
comes to studying the task offloading problem, as it does not impose
any type of dynamicity to the network conditions. In certain cases,
purely non-mobile devices like stationary IoT sensors are engaged in
task offloading at the Edge, [110,111]. Other works apply this assump-
tion on mobile devices, to reduce the complexity of their proposed
offloading solutions; for example, the authors in [112] assume that
the statistics of the utilized wireless links remain unchanged during
the processing of the users’ tasks, reflecting a relatively static or low-
mobility scenario. In a similar manner, in [113] and [114], the time to
transfer the task response from a cell to another is excluded from the
total execution time of an offloaded mobile task, by assuming that the
end device stays in the same cell during the task offloading process.
3.2.2. Mobile (high-mobility)
On the other hand, when the devices involved in task offloading are
highly mobile, i.e., their movement has a direct effect on the network
conditions considered for task offloading and the respective resource
allocation, the problem becomes significantly more complex. Several
studies consider mobility at the Edge [65,115–118]. Specifically, the
focus is placed on the user contacts (inter-contacts) in which the user
can offload the task, based on the mobility pattern and an opportunistic
computing decision [119–123]. The opportunistic computing is taking
advantage of the contact patterns regulated by the mobility of the
devices (e.g., which Edge site the node is visiting and what type of
interactions occur on daily basis), in order to determine the amount of
computation to be offloaded to other devices. Opportunistic computing
takes also advantage of the contact patterns regulated by the mobility of
these devices, to determine the amount of computation to be offloaded
to other devices. Furthermore, in a mobility scenario, users may either
transfer their tasks to remote servers or peer devices, possibly through
the gateways or even via the Edge servers [124–127]. For instance, mo-
bility can influence the decision on which BS and Edge server to select
and when to perform the handover [124]. When trying to minimize
the execution delay, user mobility information needs to be combined
with the task characteristics and resource availability, in order to make
the best task scheduling decision [125]. This mobility information is
usually captured by trajectory prediction models [128,129] that can
actually uncover motion patterns of the users in real-time scenarios.
However, in most of the studied scenarios [6,8,105,130], the tasks
are usually offloaded to BSs that are in close proximity to the user’s
position and have sufficient resources to satisfy the time execution
requirements. For example, when the tasks are lightweight and can
be executed within a satisfactory time period, without migrating to a
7
neighbor Edge site, then the execution of the task should be processed
immediately and return to the mobile user. On the other hand, when
the task requires significantly longer time, then the task could be split
into sub-tasks and be transferred to neighboring BSs along the user’s
trajectory [6,8,131].
Most of the mobility tackling solutions integrate mechanisms to
obtain the device’s current and future positions, as well as tune the of-
floading infrastructure accordingly, to achieve the objectives described
in the following subsections. Using the type of the mobility tackling
mechanism as a criterion, mobility solutions can be categorized in the
following classes:
Proactive — Behavior Related: Nowadays, services running on
mobile devices, e.g., Google Location Services [132], constantly track
and log the historic mobility behavior of the user; the proliferation
of smartphone devices has made trajectory pattern crowdsourcing a
reality. What is more, distributing intelligence at the Edge has allowed
for logging the times an end user device connects and disconnects to a
smart access point, thus extracting users’ periodic movement patterns.
Based on this data, mobility information can be estimated and lever-
aged towards predicting the users’ position at any given moment [133,
134]. Specifically, this mobility information can be extracted in a
probabilistic way, by utilizing Mobility Markov Chains (MMC) to model
the historic behavior of a user as a discrete stochastic process; in this
way, the prediction of a user moving to a specific location depends on
their previous visited locations and the probability distribution of the
transitions between them [135]. Complementary to the regularity in
the users’ mobility patterns, a Markov model can be trained to estimate
the expected network quality and the expected staying time under the
coverage of each Edge server [136]. Then, one way to leverage the
extracted information is in favor of bringing the Edge Computing and
storage resources closer to the user; this can be achieved by proactively
installing the services that the users will consume in the Edge servers
located in the positions that they will most probably visit, thus reducing
the network delay during task offloading [137–139]. The probability
density function of the sojourn time, i.e., the time a user is expected to
spend within the coverage area of an Edge site, can also be exploited
towards predicting the user’s next location and seamlessly migrating
the service to be used for the task offloading appropriately [140]. For
example, MAGA [141] is a mobility-aware mechanism for partial task
offloading that falls in this category. Frequent mobility patterns are
inferred by a tailor matching subsequence method and then a genetic
algorithm is used for the offloading decision.
Proactive — Trajectory Related: Another way of proactively deal-
ing with mobility at the Edge is by exploiting the user’s ongoing
movement characteristics, i.e., trajectory, duration and speed. By ap-
plying these characteristics on specific translational motion models, one
can predict the location and the time of the next Edge server han-
dover [142]. Apart from utilizing motion models, periodically receiving
timestamped geolocation updates from a moving user, enables produc-
ing real-time travel information for route segments which can be used
for trajectory estimation [143]. In a cooperative Edge infrastructure
scenario, taking advantage of the mobility information can guide Edge
servers to route the collected offloaded tasks to adjacent servers at
the next location on the user’s moving direction. In this way, when
users arrive at the coverage area of the next Edge site, they receive
the product of their completed offloaded task, with the minimum addi-
tional delay [144]. As an example, a two-step offloading mechanism for
smart touristic services [145,146] is based on estimating the location
and density of users. Every mobile device takes the initial offloading
decision based on a dead reckoning technique and measurements of its
WiFi signal strength. Secondly, at the Edge side, a Kalman filter is used
to predict the number of users and a controller is responsible for the
final offloading decision and the allocation of resources to VMs.
Reactive: The evolution of the Edge–Cloud continuum and the
growing adoption it receives, has recently enabled network infrastruc-
tures to quickly and efficiently adapt to the rapidly changing user
environment, in real time. When it comes to task offloading on the
Computer Networks 195 (2021) 108177F. Saeik et al.
move, Edge servers can utilize a central agent, located at the Cloud, to
form a mobility-aware offloading infrastructure that tracks the users’
position and optimally routes the task and its response through the
closest server to the users’ locations [147]. In a similar manner, the
whole virtual server can migrate to the topologically closest Edge server
to the user, reactively, every time a relocation is detected [148]. For
instance, utilizing IP tracking, remote caching and the Software-Defined
Networking (SDN) paradigm, can set the ground for efficient and timely
task offloading as well; an SDN controller is able to track the user’s
network location, i.e., the Edge server in proximity and quickly react
to changes in it by rerouting the offloaded task’s response [149].
3.3. Task offloading objectives
When solving the task offloading problem, a number of differ-
ent objectives may be applied, as the different stakeholders and ac-
tors (e.g., Cloud providers, Edge providers, Mobile Users and Service
Providers) target a variety of goals. An objective function helps to
formally and mathematically formulate these goals and guide the of-
floading solution. Objectives of the task offloading problem can be
categorized as follows:
3.3.1. Delay
Task execution delay minimization is one of the main objectives
during the task offloading problem [42,65,115,150–153]. Regardless
of the type of task offloading, the overall goal is narrowed down to
reducing the total task execution delay. This delay, can be broken down
into a number of different delay contributors. The first source of delay
is the task execution delay, coming from the task that can be either
executed locally at the device, at the Edge or the Cloud. In the case
of offloading the task at a remote Edge or Cloud site, we need to
take into consideration the transmission and propagation delay at the
various layers of the infrastructure (access, Edge, transport and Cloud
networks), in both directions (i.e., sending the task and receiving the
response). On top of that, processing and queuing delay at the vari-
ous processing and forwarding devices should be taken into account.
Finally, an additional delay contributor can be the time to optimally
partition the task delay, during the task offloading decision [154]. The
delay objective can be expressed as either the minimization of the aver-
age delay of each task [155] or the total delay of all the involved tasks
of a mobile application. This objective is directly proportional to the
available resources [115,150] and the network conditions [151,154].
The total execution delay can also be used to assess the impact of
task offloading to the QoS achieved. Thus, according to the type of the
application used and the part of the infrastructure under consideration,
the task execution delay can be associated with: (i) the response time
(i.e., the time duration from when a user requests a service until the
service actually initiates [65,118,156]), (ii) the delay variation, in order
to reveal how robust the task offloading solution can be, both over time
and over dynamic traffic profiles, while also estimating the number of
SLA (Service Level Agreement) violations noticed [107,157,158] and
(iii) the network delay, including the four delay contributors (i.e., prop-
agation, transmission, processing and queuing delay) at the different
parts of the infrastructure [65,107,118,152,155,156,158–160].
3.3.2. Energy
The second most common objective during task offloading is the
minimization of the energy consumption [161–165]. This energy con-
sumption typically refers to the end devices [116,166–169]. The reason
is that mobile and IoT devices are usually battery powered, thus a major
concern is how to maximize the lifetime of the battery by reducing
the device’s energy consumption. Inevitably, it is reasonable to assume
that, by following a full offloading approach, the maximum energy sav-
ings can be pursued. However, a number of other energy contributors
need to be taken into account, even when a full offloading approach
is followed. First of all, during the offloading, the transmission power,
8
modulation and coding scheme, together with other radio parameters,
needs to be taken into account, as they contribute in significant energy
and consequently battery consumption [41,165]. This type of energy
consumption can increase, especially when network conditions are not
favorable. Secondly, by reviewing the full offloading from a complete,
network-wide view, one can easily understand that the problem is
simply pushed to the Edge and/or Cloud infrastructures. Thus, energy
consumption minimization needs to be pursued at all layers of this
end-to-end communication model [170,171].
To evaluate this objective, a number of different metrics can be
used; the most common is the average power consumption measured
by aggregating the power consumption on the hardware equipment
used [159,172]. Alternatively, energy consumption can be used, ex-
pressed as the power consumption over time. Normally a minimiza-
tion of power consumption leads to energy consumption minimization
as well [107,173–175]. Furthermore, as the end devices are usually
battery powered, the energy savings can be expressed as battery sav-
ings [170,176]. Finally, another way to provide the necessary en-
vironmental and economic sustainability is through minimizing the
electricity cost [177]. Electricity cost depends on location and time.
Hence, appropriate allocation of offloaded tasks potentially reduces
the electricity cost, cutting down on operational costs while providing
benefits to the environment [157,167].
3.3.3. Bandwidth/spectrum
The available bandwidth at the access network and how it can
be shared by multiple users in order to offload the tasks, is also
a significant constraint. However, due to the great influence it has
on the task offloading performance, it can also be considered as an
objective [166,176]. Due to the limited available bandwidth, especially
in IoT networks and dense cellular networks, the careful allocation of
spectrum becomes of utmost importance.
The objective of spectrum allocation is often associated with the
transmission rate and power level of each end user [176], as well as
the duration of the transmission of each device [151], in order to op-
timally share the available bandwidth. Thus, when trying to optimally
deploy the available spectrum, an efficient metric is to evaluate the
spectrum utilization in accordance with the number of offloaded tasks,
power transmission and bandwidth consumption [64,152,173,174]. In
view of the dynamic wireless conditions, the optimal scheduling of
the bandwidth needs to follow the time-varying channel state and be
also associated with the arrival rate of the tasks [166]. Throughput
is another typical metric applied to evaluate the overall spectrum
utilization, since it reveals how timely and efficiently the task can be
offloaded at the remote infrastructures [170,178].
3.3.4. Load balancing
How to carefully allocate and schedule the available physical hard-
ware resources is another objective to consider during task offload-
ing [12,17,42]. Specifically, there is a high interest in path optimiza-
tion, efficient resource usage and load balancing when solving the task
offloading problem. The goal is to provide the necessary scalability by
increasing the resource availability, increasing the number of offloaded
tasks concurrently deployed at the Edge and/or Cloud [17,179], max-
imizing the resource sharing and fairness among multiple users [42]
and facilitating the offloading of future tasks.
This objective can be translated into minimizing the overall resource
usage (e.g., minimizing the average percentage of the computational
and communication resource utilization), minimizing the maximum
resource utilization of the infrastructure or minimizing the variance
of resource utilization [95]. Load balancing can be applied either at
each layer of the infrastructure or between the different layers. For
example, the appropriate distribution of traffic between an Edge and a
Cloud infrastructure, can be considered as an alternative load balancing
objective [171,180].
Computer Networks 195 (2021) 108177F. Saeik et al.
i
m
b
a
m
p
t
i
A
a
w
s
l
t
d
d
f
d
d
a
b
t
s
a
d
C
t
3
a
i
t
c
3.3.5. Deployment cost
The task offloading problem can be often seen as a resource allo-
cation problem where appropriate resources at the Edge and/or Cloud
need to be reserved according to a deployment utility cost, in order
to execute the offloaded tasks in a virtualized environment. This de-
ployment cost can be modeled in various ways, each having a different
interpretation. For example, the deployment cost can be defined as the
aggregation of computational and communication resources that a set
of tasks needs in order to be provisioned. This is typical when mobile
or IoT applications needs to be complemented with specific network
functions (e.g., security, compression, QoS) that is otherwise difficult
for a user to achieve locally in his device [180–182].
In general, this cost can be expressed as the monetary cost for
computing processing (e.g., $/CPU hour), memory (e.g., $/GB) and net-
work bandwidth (e.g., $/GB/day), induced for using network resources.
These costs can be associated with the Cloud and Edge providers
and how and to whom they lease their infrastructure. Furthermore,
since this cost usually follows a ‘‘pay-as-you-go” model, the number
of physical resources used for the total number of tasks offloaded can
reflect the deployment cost [152,157].
3.3.6. Model accuracy
All the objectives described so far, are some typical objectives that
can be used regardless of the optimization solution/strategy followed.
However, when it comes to Artificial Intelligence (AI) and Machine
Learning (ML) -based approaches, an additional objective can be the
model accuracy for the prediction of the behavior of the applications
(e.g., mobility [183] and network related features [184]). Usually the
AI/ML objectives are used as secondary or complementing objectives of
the ones presented above [185]. In addition to that, regarding critical
applications that build or use ML models during the task offloading,
particular emphasis should be placed on the training time, inference
time and the cost of the required computational power.
Model accuracy can be optimized by using the appropriate ML
metrics. For example, regression metrics [186] can express how close
the predictions of a model are compared to the actual values, by using
the R-squared, Root Mean Squared Error (RMSE) and Mean Absolute
Error (MAE) metrics. In clustering models, the evaluation metrics [187]
measure the cohesion and separation of groups of observations. An
example of such an approach is the Sum of Squared Error (SSE) metric,
which aggregates the distance of each observation from its nearest
cluster. When using classification ML techniques [188], typical metrics
used to evaluate the accuracy are the precision (i.e., the percentage
of relevant observations among the retrieved observations), Recall
(i.e., the percentage of the total amount of relevant observations that
were actually retrieved) and F-Measure (i.e., the harmonic mean of
precision and recall).
3.3.7. Multi-objective
The typical objectives presented so far are usually conflicting, mak-
ing task offloading a very challenging problem. For example, aiming to
minimize the latency can lead to higher energy consumption on the end
device, by deciding to execute the tasks locally and vice versa. When
adopting load balancing objectives, offloaded tasks can be distributed
among different Edge sites or between the Edge and the Cloud, in
order to reduce the total delay and energy consumption. Similarly,
when minimizing the deployment cost, offloaded tasks may be gathered
in one single Edge site; this results in an uneven resource utilization
that can also create significant congestion in the infrastructure and
thus higher communication delays. Finally, when trying to optimize
the spectrum allocation independently of the available communication
and computational resources, it can result in poor offloading decisions
in terms of delay, deployment cost and load balancing. Hence, multi-
objective solutions can be used in order to explore the trade-off between
the various objectives. The most common multi-objective approaches
9
consider jointly minimizing the delay and energy consumption [117, f
173,189–196]. Energy and latency optimization can also be combined
with an optimal spectrum allocation through appropriate power and
channel allocation [162,167,168,174]. Regarding load balancing, it can
be jointly optimized with the delay, since both objectives are tightly
correlated [42].
3.4. Challenges of task offloading
Achieving the aforementioned objectives of task offloading entails
a series of challenges. In this section, we identify these challenges and
classify them into two main categories.
3.4.1. Network dynamics challenges
Dynamic Network Conditions: The mobile and IoT networks are
characterized as quickly varying access networks that create dynamic
network and traffic conditions. This is a significant challenge that adds
an extra level of complexity during the task offloading problem, since
it is very difficult to pre-specify the behavior of the network. Aspects
of noise, interference, fading and signal reflection can significantly
impact the wireless communications, aggressively altering the overall
throughput and delay of the wireless transmission. This necessitates an
analysis and prediction of the network conditions, in order to accu-
rately estimate when a task offloading decision positively affects the
performance. Besides this, prediction can be combined with a resource
allocation mechanism at the Edge and Cloud, since the amount of
resources required for the task execution is directly proportional to the
amount of traffic (i.e., the request rate) that will actually end up at the
Edge or Cloud.
Dynamic User Behavior: Another level of impediment and dynam-
cs, during task offloading, is added by the random behavior of the
obile users and how they employ their mobile applications. These
ehavioral aspects are very difficult to foresee and quantify, creating
s result arbitrary user-based traffic profiles. A categorization of the
obile applications based on the users’ preferences, the transmission
atterns, the spatial and temporal correlation of the user generated
raffic, as well as other traffic related characteristics, can be of utmost
mportance for the subsequent resource scheduling and allocation.
ccordingly, machine learning and data analytic techniques should be
pplied to estimate the users’ behavior and the rate of task generation.
Edge/Cloud Dynamics: Although the Edge and Cloud layers can
ork together in harmony, they still have their own dynamics. Cloud
ites are centralized while Edge sites are distributed having only a
ocal view of the network.This leads to different dynamics between
hese two layers. Specifically, contrary to Cloud, Edge has a spatial
ynamic exactly because of its location awareness. Additionally, end
evices can dynamically re-purpose and re-associate themselves to dif-
erent applications by offloading different type of tasks or simply new
evices could appear or disappear. This inevitably creates an additional
ynamic factor for the Edge. Obviously, the initial task offloading and
llocation decision over the Edge could be performed in an optimal way
y the Cloud, since it possesses this centralized system view; however,
he latter may not react timely to local dynamics. Therefore, Edge
ervers should meet the burden to locally decide to move services’ tasks
long time. Thus, a new challenge arises in order to (i) address these
ynamics, (ii) create a consistent view of the tasks to be executed that
loud and Edge should share, and (iii) place the different services at
he right places in due time.
.4.2. Resource allocation challenges
Task offloading is strongly affected by the resource allocation mech-
nisms that decide how and where the offloaded tasks will be executed
n a remote platform. Thus, the task offloading and resource alloca-
ion decisions are coupled and should be addressed jointly. The main
hallenges this issue creates follow.
Partitioning Decision: The decision of which task to offload is the
irst and most significant challenge to address, as it comprises the core
Computer Networks 195 (2021) 108177F. Saeik et al.
Table 1
Comparison of Mathematical Optimization (MO), Artificial Intelligence (AI), and Control
Theory (CT) approaches.
MO AI CT
Stability ✓
Low complexity ✓
Optimality ✓ ✓
Online training ✓
Reachability ✓ ✓
Real-time decision ✓ ✓ ✓
of the task offloading problem. As shown in Fig. 2, when new tasks
are generated by an application, an intelligent mechanism is required
in order to decide whether the task should be executed locally or to
be offloaded to a remote infrastructure. This partition decision of the
tasks is associated with the task execution delay, the transmission delay
and the energy consumption. A poor partitioning decision may result
in performance bottlenecks regarding the execution of the application.
Thus, a compromise between when and which tasks should be offloaded
to the Cloud/Edge has to be explored, taking also into consideration
any possible transmission costs in terms of energy, delay and money.
Resource Availability: The performance of an application is closely
dependent on the resources available at the end user/Cloud continuum.
In general, as we move towards the core of the infrastructure, the
available resources increase in amount, paying the price, however, of
a higher application delay. Hence, sharing these resources is a crucial
challenge which needs an efficient resource allocation and management
mechanism that will be able to guarantee the performance require-
ments. Thus, alongside the offloading decision, the resource allocation
mechanism should fulfill various functional and non-functional require-
ments. The primary goal of resource scheduling is the respect of the QoS
requirements of the application. Additionally, the resource allocation
should guarantee important properties, such as stability, reachability,
safety and robustness against internal uncertainties and external dis-
turbances. In terms of functional requirements, the resource allocation
strategy should be implemented with commercial or open-source re-
source orchestration tools, that enable scalability, interoperability and
the transparent development of the applications over heterogeneous
hardware and software technologies.
Performance Modeling: Measuring the performance of a task of-
floading solution is an additional challenge. The task offloading prob-
lem can be modeled as a system where the energy and/or delay are
the typical output variables and the available computing resources
(e.g., CPU, memory), incoming requests and network bandwidth are
the input variables. In most of the current studies, the proposed per-
formance models are single-input/single-output, empirically derived or
fixed. Although this assumption is realistic, the processing time of an
offloaded task depends on many time-varying parameters, which are
usually not easily measured. On the other hand, multi-input/multi-
output models are more accurate, but the identification process is usu-
ally strenuous. Specifically, the offloading decision performs adequately
only for specific operating conditions, being unable to guarantee sta-
bility under fluctuating workloads and heterogeneous communication
infrastructures, such as in IoT. Hence, this system model should be
adapted in order to include the performance metrics, expressed as state
variables, that can be regulated by the control parameters (i.e., the
input variables). This framework will be capable of capturing struc-
tural changes interpreted as discrete jumps in the dynamics, e.g., user
mobility, changes in network conditions and addition/removal of Edge
servers.
Task Management: As stated before, offloading tasks to the Cloud
follows a centralized approach in which the Cloud infrastructure serves
the whole set of tasks coming from the access network layer. In con-
trast, at the Edge layer, the infrastructure is usually distributed in
multiple geographically dispersed Edge sites. Obviously, this is one of
10
the core advantages of Edge Computing, as it creates a local efficiency
by executing the tasks and effectuating actuation in minimal time. How-
ever, at the same time, a meticulous design of the task management
control modules is required at the Edge. The placement of controllers
and their mapping to the sites that they will serve, the decision of which
task is going to be offloaded at each site and how the load is going to
be distributed between the sites, are some of the challenges that fall in
this category.
4. A taxonomy of task offloading approaches
This section lists the most prominent task offloading solutions that
have been proposed in the literature. These algorithmic solutions can
be divided into three categories: (i) Mathematical Optimization algo-
rithms; (ii) Artificial Intelligence techniques and (iii) Control Theory-
based approaches.
4.1. Methodology comparison
Before providing the most common approaches for each category,
it is worth investigating the advantages and disadvantages of each
category and the level of their efficiency. Accordingly, we present the
main characteristics of each category, while Table 1 summarizes the
main features supported.
Mathematical optimization is the most common solution category
applied in resource allocation and scheduling networking problems.
The reason lies in the fact that, traditionally, these types of problems
can be mathematically formulated and solved, using a great vari-
ety of existing solutions. Usually, the main goal for this category of
algorithms is to find the optimal solution among a set of possible solu-
tions. For example, in the context of task offloading the mathematical
optimization approaches will have to appropriately model the input
(e.g., Edge/Cloud infrastructure, end users, available resources, task
distribution size and duration) and, according to a certain objective,
decide when and where the tasks should be optimally offloaded.
The optimality can be achieved through exhaustive search optimiza-
tion solvers in the expense of a high complexity and execution time.
Nonetheless, mathematical optimization approaches can reduce their
time complexity when relaxing any hard constraints of the input and
altering their final goal into finding a sub-optimal but fast and real-
time solution. Even though these types of solutions can fit and be
used in real scenarios, they sometimes suffer from their static nature
and inability to adapt and model the dynamic challenges inherited
from the problem at hand. Under such circumstances, the algorithms
should be re-executed and re-customized every time a change happens
in time and/or space (e.g., dynamical arrivals of end users, mobility
and equipment failures).
AI task offloading mechanisms have also seen great progress nowa-
days. Data driven models, learning from batch or online data, provide
real-time task offloading decisions and elastic resource allocation. De-
cisions are made by generalizing historical data, recognizing in an
automatic way the prevailing data patterns and evaluating the possible
destination states of the main actors in the Cloud/Edge environment.
The state of Edge infrastructures includes the status of computing nodes
in terms of resource utilization, the number of application requests and
the user requirements contracted as SLAs.
Contemporary AI and ML task offloading is characterized by flexible
adaptation and automatic learning. ML models have the advantages
that they are not explicitly designed by human experts and they are
self-trained based on the available data. In addition, they can handle
multi-dimensional and multi-variety data in a unified way and they are
capable of identifying hidden patterns. The main weakness of AI-based
models lies in the case of significant inconsistency between training and
testing data properties, which may lead to performance degradation.
This means that the data should be selected and gathered with diligent
attention to detail and special emphasis should be given to the data
Computer Networks 195 (2021) 108177F. Saeik et al.
preparation tasks of synchronization, transformation, and normaliza-
tion. Lastly, we should take into consideration that the large amount
and high frequency of data make the training model a computational
heavy and resource-intensive process.
System Theory provides various models for describing the opera-
tion of a process. Additionally, Control Theory provides many formal
methodologies to analyze and control the performance of the pro-
cess. Both of them have been widely used for industrial processes
while, during the last decades, they have been introduced to com-
puter networks. System theory provides black- or gray-box training
algorithms, namely system identification, to compute multiple-input-
multiple-output (MIMO) models that capture the system dynamics of
continuous or discrete systems. However, system identification must
be performed offline and the computed model may have low accuracy.
The control-based task offloading solutions enable real-time decisions
against the dynamic network and workload conditions. Additionally,
apart from reaching an optimal operating point, the control-based
offloading solutions can guarantee various system properties, such as
stability and reachability. The guarantee of stability means that the
system will reach specific operating conditions and will remain on them
against any disturbance. The reachability property means that, given
the current system state, we can compute all possible destination states.
Although, the complexity of a controller increases with the complexity
of the system model (linear or not) and the properties to be guaranteed,
the design is an offline process and the real-time application of control
law is simple.
4.2. Mathematical optimization algorithms
The task offloading problem is usually formulated as a mathematical
problem, which tries to find an optimal or near-optimal solution.
The problem can be formulated by defining the objective function as
described in Section 3.3 and the optimization strategy used. These
strategies may include Mixed Integer Programming, heuristics, meta-
heuristics and game theory approaches, among others. Following, we
present the main optimization strategies found in the literature, while
a summary of them along with the objective of the study, algorithm
developed and type of offloading is listed in Table 2. Furthermore,
Fig. 3 illustrates the key components of the existing mathematical
optimization approaches, shedding light into the well and less explored
proposed solutions.
4.2.1. Mixed integer programming
Mixed Integer Programming (MIP) formulations provide a flexible
and mathematically precise way of formulating many real-world prob-
lems. Specifically, integer programming is a commonly used technique
for resource allocation and scheduling in wired and wireless networks.
The two main problem types that MIP addresses are: (i) network syn-
thesis and (ii) resource assignment problems [208]. MIP optimization
approaches facilitate also the introduction of a multi-objective function
optimizing more than one goals under various offloading constraints
(e.g., delay, energy and load balancing). Hence it can be often used as
an optimization strategy during the task offloading problem. Usually,
MIP provides a linear objective function (MILP), where at least one of
the variables takes integer/binary values. Even though these types of
problems can provide the optimal solution, they can be very complex
or even computationally intractable for large scale experimentation
scenarios. However, they can often be used as benchmark approaches
during the performance evaluation. For example, in the context of
task offloading, they can be used to minimize the weighted amount
of mobile energy consumption in a multi-user system, under latency
constraints [192]. Regarding delay, the objective function of the MIP
can include both the transmission and processing delay, especially for
IoT mission-critical applications in an Edge–Cloud collaboration [197].
In case the objective is non-linear, a Mixed Integer non-linear
(MINLP) or quadratic (MIQP) programming formulation is modeled.
11
Similarly, these types of problems allow the formation of multi-object-
ive functions. For instance, a system cost representing the weighted
cost of delay and energy consumption among all available users, can
be expressed as a non-linear objective function in a mixed Cloud–
Edge task offloading environment [198]. In case only the energy exists
in the objective functions, typically the latency requirements can be
imposed as constraints along with other various conditions (e.g., power
consumption levels, channel states and resource heterogeneity) [199].
4.2.2. Heuristics
Heuristic approaches can introduce fast but sub-optimal solutions.
The main advantage of heuristics is that they are simple algorithms
devised to address the problem at hand, with low execution time. In
contrast with MIP algorithms, they do not require specialized optimiza-
tion tools to be solved and can rather be expressed as pseudo-code, eas-
ily implementable in any programming language. To this end, heuristic
solutions are very popular to be applied in the task offloading problem.
These types of solutions can range from optimizing the offloading
decision of the user while minimizing the overall cost of energy, compu-
tation and delay, by applying appropriate relaxation and randomization
techniques [194], to optimizing the resource allocation at the Edge, by
considering a Cross-Entropy optimization approach [162]. Heuristics
can also be used to optimize non quantifiable parameters, such as QoE
in a Cloud–Edge collaboration [197], by satisfying the various compu-
tational and bandwidth restrictions, applying appropriate fairness and
popularity techniques [85].
The efficiency of the heuristic becomes much more evident when
the task offloading problem is modeled as a non-linear constrained
optimization problem, or when the experimentation covers large-scale
offloading scenarios [162]. In this case, greedy heuristics can be used
to estimate the exact solution [42,85,200].
4.2.3. Game theory
Lately, there has been also a use of game theoretic approaches to
deal with resource allocation problems. Through game theory, the task
offloading problem can be introduced as a resource allocation game.
For example, the problem of the partial task offloading in a multi-user,
Edge Computing infrastructure and a multi-channel wireless interfer-
ence environment, can be formulated as an offloading game [201].
This game tries to maximize the spectrum efficiency during offloading
by allocating the proper channel to each user/player. The specific
approach can be complemented with a second matching game that will
aim to maximize the efficiency of resource allocation at the Edge, by
appropriately selecting the right Edge servers [196]. A multi-step/slot
game theoretic approach can be followed in order to find the optimal
state, expressed as the Nash Equilibrium. Specifically, in each step
the end user/player can make a decision on whether to offload their
tasks in order to reach a potentially optimal offloading. A similar
slotted approach can be followed by treating each user as a player
with the goal to optimize the CPU-cycle frequency and offloading
decision, in order to maximize the energy efficiency. Game theory
has also been used in an Edge–Cloud interplay, where players can be
considered as the corresponding infrastructures [202]. In particular, by
formulating a Stackelberg game, a leader player is assigned the goal to
maximize a utility function expressed in order to obtain the optimal
revenue for the Edge and Cloud providers, while satisfying the delay
requirements. A different objective can be considered in an Edge–Cloud
collaboration, where the two infrastructures comprise the players of
the problem and try to minimize the overall energy consumption under
delay constraints [203], by using a potential game [209].
Computer Networks 195 (2021) 108177F. Saeik et al.
Table 2
Taxonomy of mathematical optimization task offloading algorithms.
Reference Optimization objective Algorithms Mobility Offloading
Delay Energy Bandwidth/
spectrum
Load
balancing
Deployment
cost
MIP Heuristic Game
theory
Contract
theory
Local
search
Static Mobile Partial Full Edge–
Cloud
[41] ✓ ✓ ✓ ✓ ✓
[42] ✓ ✓ ✓ ✓ ✓
[65] ✓ ✓ ✓ ✓ ✓
[85] ✓ ✓ ✓ ✓ ✓ ✓
[89] ✓ ✓ ✓ ✓ ✓
[116] ✓ ✓ ✓ ✓
[117] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[121] ✓ ✓ ✓ ✓ ✓ ✓
[150] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
[151] ✓ ✓ ✓ ✓ ✓
[152] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[154] ✓ ✓ ✓ ✓ ✓ ✓
[155] ✓ ✓ ✓ ✓ ✓
[161] ✓ ✓ ✓ ✓ ✓
[162] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[163] ✓ ✓ ✓ ✓ ✓
[164] ✓ ✓ ✓ ✓
[165] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[166] ✓ ✓ ✓ ✓ ✓
[167] ✓ ✓ ✓ ✓ ✓
[168] ✓ ✓ ✓ ✓ ✓ ✓
[169] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[170] ✓ ✓ ✓ ✓ ✓
[173] ✓ ✓ ✓ ✓ ✓ ✓
[174] ✓ ✓ ✓ ✓ ✓ ✓
[176] ✓ ✓ ✓ ✓ ✓
[189] ✓ ✓ ✓ ✓ ✓
[191] ✓ ✓ ✓ ✓ ✓ ✓
[192] ✓ ✓ ✓ ✓ ✓
[193] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
[194] ✓ ✓ ✓ ✓ ✓
[195] ✓ ✓ ✓ ✓ ✓
[196] ✓ ✓ ✓ ✓ ✓ ✓
[197] ✓ ✓ ✓ ✓ ✓ ✓
[198] ✓ ✓ ✓ ✓ ✓ ✓
[199] ✓ ✓ ✓ ✓ ✓
[200] ✓ ✓ ✓ ✓ ✓
[201] ✓ ✓ ✓ ✓
[202] ✓ ✓ ✓ ✓
[203] ✓ ✓ ✓ ✓ ✓
[204] ✓ ✓ ✓ ✓ ✓
[205] ✓ ✓ ✓ ✓ ✓
[206] ✓ ✓ ✓ ✓ ✓ ✓
[207] ✓ ✓ ✓ ✓ ✓
4.2.4. Contract theory
Naturally, task offloading introduces conflicts between the partici-
pating parties; for example, on the one hand, users and devices seek to
maximize energy and spectrum efficiency while on the other, small cells
and Edge servers try to minimize consumption of their own resources,
like battery capacity and computing power. Conflicts like these might
cause reluctance by third parties to participate, which could subse-
quently raise barriers to the development of attractive traffic offloading
solutions. Contract theory is an approach originating from real world
economics, that dictates the design of contracts to achieve cooperation
between the conflicting sides. In a broad sense, contract theory studies
the design of formal and informal agreements that motivate agents
with conflicting interests to take mutually beneficial actions. In the
wireless networks domain, the agents include the BS, service provider
and the spectrum owner, as well as the small cells, smart devices and
users [210]. The late boom of contract theory applications in task
offloading has managed to deal with many early challenges of the
field; for instance, combining contract theory with game theory and
a monetary rewards system, can eliminate the influence of information
asymmetry in a user–Edge server relationship [204]. Similar incentives
can be utilized when dealing with small-cell base stations (SBSs) and
heterogeneous ultra-dense networks (HetUDNs) [205]. When the goal is
to optimize bandwidth allocation in data offloading, dynamic program-
12
ming concepts can integrate with contract formulation, as in the UAV
to macro base station (MBS) offloading scenario, described in [206].
In the case of opportunistically offloading part of the cellular traffic to
coexisting networks, towards alleviating the overload problems caused
by traffic demands, a contract theory-based incentive mechanism can
motivate users to leverage their delay tolerance in exchange for service
cost [207].
4.2.5. Local search
Local search algorithms adopt mechanisms of perturbation to ex-
plore neighbor solutions in the search space, that allow to gradually
converge to local optimum solutions. Due to the problem-agnostic
nature of the local search algorithms, they can be used as a compo-
nent of heuristics and meta-heuristics in order to provide solutions
very close to optimality [211]. For instance, a one-dimensional local
search algorithm can be used for a partial task offloading solution,
with the goal to minimize the average execution delay, expressed as
a Markov chain process, under the energy constraints imposed by the
device [155]. When both energy and latency constitute the objective
of the partial task offloading problem, a univariate search technique
can be used [189]. This type of search allows to transform a non-
convex problem into a convex one, by finding a local optimum solution.
An iterative local search can further reduce the gap with an optimal
solution, where multiple iterations of the local search can result in
Computer Networks 195 (2021) 108177F. Saeik et al.
Fig. 3. Summary of mathematical optimization task offloading algorithms.
a better resource allocation of computing and channel resources in a
partial offloading scenario [174]. When the goal is to optimize the
energy sustainability of the end-user devices, by selecting the proper
Edge resources and at the same time to minimize the execution time of
the allocation, a simple bi-section search algorithm can be used [151].
4.3. AI-based optimization algorithms
In the above section, we provided traditional mathematical and
algorithmic approaches to derive the optimal or near-optimal solution,
in the context of task offloading. However, these approaches may suffer
from the following issues: (i) Most of the solutions investigated so far
fail to take into consideration the dynamic network conditions. Since
this is a random variable, it is difficult to estimate and reflect this
behavior during the allocation of the Edge and Cloud resources and
during the task partitioning decision; (ii) The traditional approaches are
rather opportunistic, addressing the challenges of the task offloading
in a short-term scale. However, in this manner, we cannot capture
the long-term time and space varying conditions in all the layers of
this end-to-end communication model. In other words, the solutions
presented above lack the ‘‘intelligence” to better adapt holistically to
the inherent challenges of the problem at hand. This prepares the
ground for using artificial intelligence techniques for the task offloading
problem.
Artificial Intelligence (AI) techniques include multi-disciplinary
techniques from machine learning, consensus-based and constraint-
based algorithms and they have been widely used in different computer
systems and network scenarios [8,185,200,212,213]. AI techniques are
becoming successful alternatives also for solving optimization problems
that include the mathematical formulation of uncertain, stochastic and
dynamic information, thus making them excellent candidates for the
task offloading problem. Furthermore, AI can potentially reduce the
complexity by enabling recursive feedback-based learning and local
interactions and thus faster speed in seeking sub-optimal solutions
than traditional techniques [8,185,212]. For example, during the task
13
offloading problem, by learning from data and tasks distributed across
the Edge infrastructure, AI can enable a smart, real-time, and dynamic
resource management framework [11,214–216]. On another perspec-
tive, AI techniques can also be applied to avoid costly data offloading,
by enabling data estimation or prediction, like in dual prediction
approaches [217].
Similar to traditional mathematical optimization solutions, when us-
ing AI, a problem can be formulated by defining the objective function
and the algorithmic strategy to be followed. Following, we present the
major AI techniques used in the literature to address the task offloading
problem. A summary of the related work with the objective of the
study, the algorithm developed and type of offloading, is listed in
Table 3, while Fig. 4 illustrates the distribution of the key components
of the existing AI optimization approaches.
4.3.1. Machine learning
As a sub-category of AI, Machine Learning (ML) gives devices or
computer systems the capability to learn useful patterns and behaviors
from historic data and make decisions about new ones. The models are
built without explicit programming; in the case of ML parameterized
models, such as Linear Discriminant Analysis, Logistic Regression and
Naive Bayes, the models are built by tuning a fixed number of parame-
ters of a predefined mapping function. In the case of non-parametric
models, such as the RBF-kernel Support Vector Machines, Decision
Trees and K-Nearest Neighbor, the models use a flexible number of
parameters with no prior knowledge about the data distribution and
mapping function. In both cases, a mapping function is a function
that maps the independent data variables to the dependent variables,
i.e., the variables the model predicts.
ML models can be divided in supervised, unsupervised and re-
inforcement learning ones, based on the available training data. A
prominent ML subfield is Deep Learning, which involves Artificial
Neural Networks (ANN) with multiple layers of representation. ML
models have been used successfully to overcome the challenges of task
offloading and resource allocation, as described below.
Computer Networks 195 (2021) 108177F. Saeik et al.
c
c
o
m
C
s
e
c
C
s
a
t
s
E
r
a
o
i
o
Table 3
Taxonomy of AI-based task offloading algorithms.
Reference Optimization objective Algorithms Mobility Granularity
Delay Energy Bandwidth/
spectrum
Load
balancing
Deployment
cost
Model
accuracy
Machine
learning
Population Constraint Static Mobile Partial Full Edge–
Cloud
[218] ✓ ✓ ✓ ✓ ✓ ✓
[219] ✓ ✓ ✓ ✓ ✓
[220] ✓ ✓ ✓ ✓ ✓ ✓
[221] ✓ ✓ ✓ ✓ ✓ ✓
[222] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[223] ✓ ✓ ✓ ✓ ✓
[224] ✓ ✓ ✓ ✓
[225] ✓ ✓ ✓ ✓
[226] ✓ ✓ ✓ ✓
[227] ✓ ✓ ✓ ✓ ✓
[228] ✓ ✓ ✓ ✓
[229] ✓ ✓ ✓ ✓
[230] ✓ ✓ ✓ ✓ ✓ ✓
[231] ✓ ✓ ✓ ✓ ✓ ✓
[232] ✓ ✓ ✓ ✓ ✓
[233] ✓ ✓ ✓ ✓
[234] ✓ ✓ ✓ ✓ ✓ ✓
[235] ✓ ✓ ✓ ✓
[236] ✓ ✓ ✓ ✓ ✓ ✓
[237] ✓ ✓ ✓ ✓
[238] ✓ ✓ ✓ ✓
[239] ✓ ✓ ✓ ✓ ✓ ✓
[240] ✓ ✓ ✓ ✓ ✓
[241] ✓ ✓ ✓ ✓ ✓
[242] ✓ ✓ ✓ ✓ ✓
[243] ✓ ✓ ✓ ✓ ✓ ✓
[244] ✓ ✓ ✓ ✓ ✓ ✓
[245] ✓ ✓ ✓ ✓ ✓
[246] ✓ ✓ ✓ ✓ ✓ ✓
[247] ✓ ✓ ✓ ✓
[248] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
[249] ✓ ✓ ✓ ✓ ✓
[250] ✓ ✓ ✓ ✓
[251] ✓ ✓ ✓ ✓ ✓
[252] ✓ ✓ ✓ ✓
[253] ✓ ✓ ✓ ✓ ✓
[254] ✓ ✓ ✓ ✓ ✓
[255] ✓ ✓ ✓ ✓ ✓
[256] ✓ ✓ ✓ ✓
[257] ✓ ✓ ✓ ✓ ✓
[258] ✓ ✓ ✓ ✓
[259] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[260] ✓ ✓ ✓ ✓ ✓ ✓
[261] ✓ ✓ ✓ ✓ ✓
[262] ✓ ✓ ✓ ✓ ✓ ✓
Supervised ML Models: Supervised ML models include classifi-
ation and regression models. In classification, the model predicts
lasses while in regression the model estimates continuous values. The
ffloading decision can be formulated with a multiclass classification
ethod and the resource allocation with a regression model [218].
lassification and Regression Trees (CART) [219] have been used to
elect the fittest Edge device for offloading, while minimizing time and
nergy by taking into consideration parameters such as the authenti-
ation, confidentiality, integrity, availability, capacity, speed and cost.
lassifiers such as JRIP and J48 [220] have been used for context-
ensitive offloading in a Mobile Cloud Computing environment using
robust profiling system. Logistic regression [221,222] has been used
o calculate the load of each Edge node and enhance a dynamic re-
ource allocation strategy. A different classification approach classifies
dge applications into classes of services [223], based on their QoS
equirements, and maps them to Edge and Cloud resources. The Apriori
lgorithm [224] has also been used to generate rules for every task, in
rder to select the Edge node that offers the minimum completion time.
Linear Regression [225] has been used to predict the total process-
ng duration of each task on each candidate Edge node, in order to
14
ffload entire tasks to one Edge node instead of a local execution. Linear
Regression [226] has also been used to predict over-loaded and under-
loaded nodes, in order to facilitate a live migration process of tasks.
Gaussian Process Regression [227] has been proposed to predict the
future workload of the tasks, allowing the deployment of new, delay
sensitive applications and reducing energy consumption, blocking of
requests and latency. The dynamic characteristics of applications and
the complex Edge/Cloud Computing environment, have been modeled
with the Support Vector Regressor [228] and the K-Nearest Neigh-
bor Regressor [229] for future load prediction and energy efficient
utilization of Edge servers respectively.
Unsupervised ML Models: Clustering models discover groups of
objects that are similar, close and dense or share some common prop-
erties. Clustering models are differentiated from Classification models
in that they do not require annotated data for training. Regarding task
offloading, clustering approaches have been used to group resources
based on the distance between Edge nodes, wireless and computational
resources [230,231] in order to minimize the response delay. In the
same fashion, Edge sites can be grouped for different task resource
demands [232] and Edge servers can be grouped using an analysis
of the allocated computing resources [233]. Unmanned aerial vehicles
are also clustered to enable efficient multi-modal and multi-task of-
floading [234] and IoT users according to their priorities [235]. The
Computer Networks 195 (2021) 108177F. Saeik et al.
Fig. 4. Summary of AI-based task offloading algorithms.
dependencies between tasks can be represented by a graph and, by
following a fuzzy clustering [236], makespan (i.e., the time difference
between the start and finish of tasks), monetary and energy costs can be
minimized. The K-means clustering method [237] can provide efficient
task scheduling, thus increasing the utilization of the Edge devices,
based on the type of resource requirements in terms of CPU, I/O and
communication. Lastly, a policy-based clustering approach [238] can
provide energy efficient task offloading solutions, by organizing the
interactions among the Edge nodes.
Deep Learning: Deep Learning has gained popularity in multiple
decision and scheduling problems because of highly accurate outcomes,
especially when large amounts of data are available. In Edge and
Cloud Computing, large amounts of data are being collected by re-
source monitoring tools, application logging mechanisms and network
sniffers [263]. A modular deep learning model can integrate different
sources of data, manipulate the data observations with hierarchical
layers of representations and extract generalized knowledge that goes
beyond the historical observations.
Deep Learning can provide timely and accurate task offloading
decisions, based on the resource usage of processing Edge nodes, the
workload and the QoS constraints defined in SLA [264]. A deep learn-
ing model works as a function approximator that takes as input the
current infrastructure and workload status and outputs the appropriate
processing nodes where each tasks will be offloaded. Further outputs
of the deep learning models include the decisions for vertical or hori-
zontal scaling up and VM migration, in order to guarantee the smooth
operation of tasks execution in a dynamic and quick-change computing
environment. Specifically, Deep Learning models have been imple-
mented to minimize the computation and task offloading overhead in
varying network conditions and limited computation resources [239].
A Deep Learning model that also addresses the challenges of speed,
power and security, while satisfying the quality of services, has been
proposed in order to determine the combination of different devices and
15
dynamic tasks [240]. In addition, close to the optimal joint offloading
decisions, bandwidth allocation can be generated with a distributed
deep learning-based offloading algorithm for Edge networks [241].
Furthermore, the challenge of energy efficient task offloading with
Deep Learning has been studied in the context of the internet of vehi-
cles [242], users’ equipment [243] and specifically for delay-sensitive
and computation intensive tasks in Edge Computing networks [244].
Deep Reinforcement Learning: Generally, reinforcement Learning
models take actions in an environment in order to maximize a cu-
mulative reward or minimize expected loss. Reinforcement Learning
is usually combined with Deep Learning in order to generalize with
previously unseen data in terms of environment, states and actions. In
the Edge Computing context, a Deep Reinforcement Learning model has
been implemented to make the binary offloading decision on whether
the offloading will take place partially locally or fully remotely to an
Edge server [245]. Deep Q-Networks [246,247] have been proposed
to automatically infer the offloading decisions, in order to optimize
the system performance. Furthermore, Deep Q-Networks have been
enhanced to capture the sequence of data with long short-term memory
layers [248], for mobile tasks in a large-scale heterogeneous Edge
environment and Gated Recurrent Unit layers [249] in multi-Edge
networks.
4.3.2. Population-based methods
Population-Based methods include a wide range of nature-inspired
algorithms and provide close to optimal solutions in combinatorial
problems, following a metaheuristic approach. Two main subcategories
of Population-Based methods are the Swarm Intelligence and the Evo-
lutionary Algorithms. Both of them have been proposed in order to
provide efficient solutions in task offloading challenges.
Swarm Intelligence (Consensus-based): The Swarm Intelligence
properties make it a practical design model for algorithms that solve
increasingly complex problems. In general, swarm algorithms strive
to allow the entire system to converge into a global consensus state,
while retaining the ability to perform assigned individual tasks in the
Computer Networks 195 (2021) 108177F. Saeik et al.
Fig. 5. Summary of control theory-based task offloading algorithms.
swarm. Ant Colony Optimization (ACO) and Particle Swarm Optimiza-
tion (PSO) are the most common Swarm Intelligence algorithms. ACO
can be applied for efficient task scheduling due to its strong global
search ability [250] and improve the response time of IoT applications
by distributing effectively the tasks over the Edge nodes [251]. On
the other hand, PSO can be used to minimize both the transmission
and the processing delay, as a means to minimize the total end-to-end
delay during a partial task offloading at the Edge [252]. In this case,
PSO can also incorporate other task offloading key mechanisms, such
as VM migration and transmission power management, to minimize
service delay as efficiently as possible, to provide high QoS for different
application profiles and to remain computationally feasible. Last but
not least, PSO has been used to jointly minimize energy consumption
and completion time for high-quality solutions [253,254].
Evolutionary Algorithms: Evolutionary Algorithms are based on
natural selection principles, such as reproduction, mutation, recom-
bination and selection. They perform a lot of iterations on a set of
candidate solutions, aiming for the closer to optimal solutions to sur-
vive as much as possible, while the unfit solutions tend to be discarded.
Evolutionary Algorithms [255] have been used in the deployment of
Edge nodes and the offloading strategies. A subclass of Evolutionary
Algorithms is the Genetic Algorithms (GA), which is characterized by
the crossover principle in the reproduction of candidate solutions. GAs
have been used for sequential task offloading and proactive fault tol-
erance [256]. Hybrid models, which combine GA with PSO, have also
been proposed [257] and they achieve close to optimal task offloading
of IoT applications, while minimizing the total makespan and energy
consumption.
4.3.3. Constraint satisfaction methods
The task offloading problem has also been re-defined as a Constraint
Satisfaction Problem (CSP) with multiple source of constraints such
as SLA, QoS, QoE, the heterogeneity of devices, the particularities of
VMs and the dynamicity of the task generation process. CSP [265]
16
is related to the artificial intelligence operations research and aims
to find feasible solutions by using methodologies such as constraint
propagation, local search, backtracking and various heuristics. Specifi-
cally, task properties, user mobility and network constraints have been
jointly formulated as a CSP [258], in order to reduce the task execution
delay in Mobile Edge Computing infrastructures. A CSP formulation
has also been used in combination with Min-conflicts scheduling algo-
rithm [259], for achieving the necessary load balancing of the Edge
resources and minimizing energy consumption. A more demanding
offloading use case is the distributed processing of data streams. In
this case, special emphasis is given to minimizing end-to-end latency
through the appropriate placement of the stream operators, either
on Cloud nodes or Edge devices. A CSP optimization framework has
also been proposed in order to minimize this latency and satisfy the
constraints of power, bandwidth and CPU utilization [260].
One prominent approach to address a CSP comes from the Con-
straint Programming (CP). CP [266] is a programming paradigm used
in solving complex problems, where instead of defining a sequence of
steps for the program to execute in order to obtain the result, one
defines the relationships between variables in the form of constraints
that must be met. Afterwards, by following the steps of branching
and exploration, CP finds feasible solutions to the problem. CP has
been proposed for a generic and easy-to-upgrade placement service
for Fog Computing with short resolution times and quality solutions.
Specifically, using Choco [261], a many times awarded constraint
solver, we can estimate close to optimal solutions in terms of network
infrastructure, applications graphs and metrics like the usage of storage,
network and energy resources. In addition, CP has been combined with
an event-based finite state model [262], in order to optimize mobile
battery life and guarantee QoS and cost minimization simultaneously.
4.4. Control theory-based algorithms
Originally, Control Theory was designed to regulate the behavior of
dynamic systems and keep the system output(s) following the desired
Computer Networks 195 (2021) 108177F. Saeik et al.
control signal, also called a reference. Control theory relies on feedback
mechanisms and is widely applied to computing systems. Because of
its nature to rely on feedback, which measures the difference between
the actual output and the desired reference, control theory has been
applied in the field of Edge Computing for implementing mechanisms
of efficient decision-making control [267–269], network design se-
lection [270,271], time-critical systems [75,76,272–275], admission
control [276], network management [277], switching Edge [278] and
network switching [279], among others. A summary of the control
theory approaches, along with the objective of each study, the algo-
rithm developed, type of offloading and the consideration of mobility,
is listed in Table 4. Moreover, Fig. 5 illustrates the distribution of the
key components of the existing control theory-based approaches.
4.4.1. Optimal control
The control theory foundations, specifically those based on linear
optimal control theory (i.e., Linear Quadratic Regulator (LQR) [281]),
consider the design of a selection strategy [270] in a heterogeneous
wireless network, with the objective to maximize the network resource
utilization, while meeting the constraints of the supported services.
Linear controllers are designed to meet the system’s constraints and
QoS metrics [276]. The high efficiency of the LQR controllers guar-
antees the control performance of the system. Skarin et al. [280]
consider the LQR for MIMO linear systems. Control theory can be
deemed of utmost importance in UAVs-related task offloading where
LQR-based controllers are used in order to achieve robust adaptive
attitude control [282].
4.4.2. State feedback control
Another advantage of the control-theoretic approach in task of-
floading is that it provides a methodology for the modeling, analysis
and evaluation of the system. Avgeris et al. [276] proposed a control
theory approach to study the adaptive resource allocation problem
for task offloading. The proposed two-level resource allocation and
admission management system for an Edge application cluster, gives
mobile users an alternative option for performing their tasks. The
proposed controllers allow mobile users to offload application-specific
tasks within the coverage area. However, it should be noted that
mobility of users within the proximity of the cluster is not taken
into consideration in this work. Kalatzis et al. [274] modeled the
performance of IoT-based applications with a switched system and
computed various equilibrium points that correspond to different op-
erating conditions. Based on these points, a simple scaling mechanism
was built to satisfy the varying workload demand. Extending [274],
SMOKE [100] is a scalable resource allocation mechanism for UAV-
based forest fire detection. UAVs are able to offload images to Edge
servers for further processing. In the case of wildfire, the workload of
the UAVs in the field increases significantly and the dynamic resource
allocation is essential to achieve the desired QoS. A group of linear
systems is used to model the container-based image processing services
and a state-feedback controller is designed to scale each container’s
computing resources.
4.4.3. Lyapunov Optimization approaches
Lyapunov optimization algorithms provide a unique property of
finding the sufficient conditions for stability in dynamical systems.
Due to the stability theory of dynamical systems, Lyapunov-based
optimization algorithms can be used in order to study the task offload-
ing problem. In particular, for minimizing the energy consumption of
mobile devices, there is a number of dynamic variables that need to be
fine-tuned and converged to an optimal value in order to minimize the
total energy consumption. Thus, Lyapunov optimization can be used to
find the necessary stability in the CPU-cycle frequency of the device,
transmission power, spectrum utilization and latency [139,170,178,
191], while satisfying the necessary task execution constraints [115].
Another variable that dynamically fluctuates during task offloading is
17
the resource availability. In this case, an online task offloading algo-
rithm, which leverages Lyapunov optimization methods and utilizes the
current system information, can be used to predict the user’s resource
availability [121]. The benefits are two-fold. First, network operators
with global network information are trusted to make the comprehen-
sive offloading decision for all the users. Second, the capabilities of
mobile devices are constantly improving and the multiplexing advan-
tage (due to the flexibility of resource availability between devices) can
be exploited to enable the execution of collaborative tasks for a wide
range of services.
4.4.4. Rest of the control approaches
This paragraph includes the control-based studies that cannot be
classified in the previous categories. Dlamini et al. [277] developed an
online algorithm for Edge network management based on predictive
control. This control mechanism aims at optimizing a two-objective
cost that includes energy consumption and QoS satisfaction. Sonmez
et al. [158] proposed a fuzzy workload orchestrator for the Edge
Computing environment. Here, a set of fuzzy rules assigns the offloaded
requests to a computational unit in a hierarchical Edge Computing
architecture. Spatharakis et al. [275] proposed a switching offloading
mechanism for robotic applications (i.e., path planning and localiza-
tion) within an Edge Computing setup. The offloading decision is
based on the uncertainty of the mobile robot’s position, the resource
availability at the Edge servers and the complexity of the path planning.
In addition, control theory can be applied to address the task
allocation problem [268] with a novel integrated Top-Down Bottom-
Up (TDBU) approach. In particular, the top-down module (i) observes
the bottom-up task preference decisions of the Edge devices and de-
cides the optimal task offloading strategy to ensure the overall system
performance; (ii) leverages top-down incentive schemes to implicitly
guide the Edge devices to pick the tasks that are most likely to finish in
time. Similarly, Wu et al. [269], proposed codeSpec, a decision making
control theory approach (on the Edge device) that periodically renews
its offloading decisions, at code level, with nearby IoT devices, in real-
time. CodeSpec, shifts the destined devices from inter-domain servers
to IoT devices nearby and only offloads binary code of user-specified re-
gions across different instruction set architectures (e.g., x86_64, ARMv7
and IA-32), using control theory.
5. Open challenges and future directions
The concept of task offloading has evolved from the simple idea
of migrating the computation intensive tasks of end-user devices at a
remote location. The concept of Edge Computing and the arrival of new
applications, enabled by recent trends in wireless communications such
as the IoT and 5G, have introduced tremendous innovation opportuni-
ties for task offloading. However, new technical and business challenges
arise. This section discusses future research directions and open issues
in the context of task offloading.
5.1. Heterogeneous networks
A Heterogeneous Network (HetNet) consists of a macro cell layout
with some possible Low Power Nodes (LPNs) placed throughout their
coverage zones. Task offloading in HetNet is suited for three cases,
as shown in Fig. 6: (i) Single-cell scenario, (ii) Contiguous cluster-cell
scenario and (iii) Non-contiguous cluster-cell scenario.
In the context of single-cell scenarios, new offloading decision vari-
ables can be the interference and less congested cells selection. In
the context of clustered-cells, the densest cell expands over several
neighbor cells. The devices in the edge of the cell can extend the com-
munication capacity (as well as other types of network management,
e.g., energy or mobility) through nearby devices, to neighboring cells.
In this case, the goal is to analyze the capability of a resource allocation
Computer Networks 195 (2021) 108177F. Saeik et al.
Table 4
Taxonomy of control theory-based task offloading algorithms.
Reference Optimization objective Algorithms Mobility Granularity
Delay Energy Bandwidth/
spectrum
Load
balancing
Deployment
cost
Optimal
control
State feedback
control
Lyapunov Rest control
approaches
Static Mobile Partial Full Edge–
Cloud
[94] ✓ ✓ ✓ ✓ ✓ ✓
[100] ✓ ✓ ✓ ✓
[115] ✓ ✓ ✓ ✓ ✓
[121] ✓ ✓ ✓ ✓ ✓ ✓
[139] ✓ ✓ ✓ ✓ ✓
[158] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[170] ✓ ✓ ✓ ✓ ✓
[178] ✓ ✓ ✓ ✓ ✓ ✓
[191] ✓ ✓ ✓ ✓ ✓ ✓
[268] ✓ ✓ ✓ ✓ ✓
[269] ✓ ✓ ✓ ✓ ✓ ✓ ✓
[274] ✓ ✓ ✓ ✓
[275] ✓ ✓ ✓ ✓
[276] ✓ ✓ ✓ ✓ ✓ ✓
[277] ✓ ✓ ✓ ✓
[280] ✓ ✓ ✓ ✓
Fig. 6. Overview of general heterogeneous networks scenarios.
technique to shift to a remote Edge location, involving inter-cell com-
munication and management aspects. A third scenario to be addressed
concerns the communication of devices (small-cell environments) that
are not adjacent, i.e., scenarios where the clustered cells may or may
not be neighboring cells. This scenario assists in understanding how a
resource allocation technique operates in terms of scalability, as well
as in terms of supporting challenges such as long delays, or network
partitions (as the small-cells are not in contiguous clusters).
Heterogeneous Dense Networks (HDHNs) consisting of clustered
cells, require algorithms that are able to extend capacity over a distance
of several cells towards the crowded cells. For the optimization of
tasks with several small and distributed dense servers (either Edge or
Cloud), however, the algorithms should draw capacity from cells within
a short distance from the dense ones, such that only a few cells that are
located close to the dense cells are affected. Initial studies in this area
deal with simple scenarios of one end user and one Edge server in a
single cell, or few end devices, one or more Edge servers and a central
Cloud, while the results show the feasibility of task offloading with the
combination of Edge and Cloud communications [41,200]. However,
if multiple end-user devices reuse the spectrum to connect to multiple
Edge and Cloud servers, imposing as a constraint to not degrade the
QoE and the service continuity, the effect of signaling overhead, smooth
handover and dynamic resource allocation on the offloading, becomes
more significant. Such effects have not been thoroughly explored in the
aforementioned studies.
5.2. Real-time distributed resource allocation
The optimization procedures during task offloading are primordial
in order to handle crucial operations such as intelligent resource al-
location and service continuity, by making independent and rational
strategic decisions and smartly adapting to the environment. In this
18
scenario, access networks are usually centralized, with all the traffic
going through a central node (i.e., a BS). Furthermore, by offloading
the tasks to a distributed Edge infrastructure, a separate backhaul
connection is required which can increase the installation and energy
costs for the mobile operators. Finding the right resources to offload the
tasks to, in such distributed scenarios, is a critical objective, especially
for heterogeneous networks. On top of that, the dynamic behavior
of the user adds another level of complexity when the appropriate
Edge site needs to be selected in order to achieve the task offload-
ing objectives, such as low latency and resource optimization, while
maintaining at the same time the user association information with
the adjacent BS. Developing a real-time task offloading algorithm that
considers the user interactions and a distributed Edge infrastructure, in
order to improve the service delivery in dynamic scenarios, is one of
the greatest currently open challenges. Moreover, in the heterogeneous
networks, efficient real time allocation schemes that learn based on
historic performance and adapt online to application’s statistics, are still
in their infancy.
5.3. Mobility-induced network dynamics
In many situations, the dynamic movement behavior of the users
becomes the deciding factor on whether to offload the task or not.
Even though few existing researches aim to take mobility into ac-
count, the particular case is still considered an open challenge. For
example, developing algorithms by learning the user’s behavior and
network dynamics in parallel, in order to reduce communication and
computational costs, are of utmost importance for new and emerging
applications. Beyond 2020, there will be a growing demand for high
user mobility applications such as drone-based applications, high speed
trains, moving hotspots and 3D connectivity. Current solutions, would
be difficult to be applied in such extremes scenarios, not only in terms
Computer Networks 195 (2021) 108177F. Saeik et al.
of accuracy but also in terms of minimum performance requirements
(e.g over 500 km/h high speed mobility, high throughput and ultra low
latency).
5.4. Node, resource and application heterogeneity
Another critical challenge is dealing the heterogeneity of the avail-
able infrastructure in terms of hardware and available resources. Both
mobile and Edge devices are characterized by a great heterogeneity in
terms of hardware, software and resource capabilities specifications.
Furthermore, the existence of a large range of applications with differ-
ent performance requirements, that are readily available at the same
end device, can affect or limit the efficiency of the task offloading
solution used. All these factors are key components during task offload-
ing. Thus, maintaining adequate service delivery and service continuity,
while addressing the task offloading in such a heterogeneous scenario,
as a research topic, is still in its infancy.
5.5. Moving edge resources closer to the end devices
There exist several situations in which task offloading is indispens-
able but where the end devices are not able to directly offload data
to an Edge server (e.g., that are not in range or do not have enough
energy to reach it). In such cases, it may be useful to send off mobile
Edge resources close to these end devices and adapt to their mobility
and needs. However, obviously, mobile Edge resources might not be
as powerful as stationary ones and might also be limited in terms of
the services they can offer and their autonomy. Thus, the challenges
imposed to task offloading here will be (i) how to trigger the sending
of mobile resources, (ii) what kind of tasks to offload and (iii) from
what end devices.
5.6. Security and privacy
Naturally, task offloading involves a huge amount of data out-
sourced to third party Edge infrastructures, thus security and privacy
concepts are of paramount importance. These concepts can be ad-
dressed from different angles, i.e., (i) end user device, (ii) Edge data
center and (iii) the actual data transmission over the network. Lately, a
great increase in the variety of sophisticated attacks on end user devices
has been observed, which constitutes the main target for attackers.
Regarding the Edge infrastructures, threats are mainly focused on the
data transmission between the different nodes of the network. Proposed
solutions include various steganography and homomorphic encryption
techniques, as well as hardware-based secure execution. However,
when used individually, most of these solutions have limitations in their
applications; e.g., encryption keys may be too large hence dramatically
increasing the amount of transmitted and stored data, while computa-
tion on encrypted data is still in early research stages. Undoubtedly,
Edge-related security and privacy threats are advancing in a quick
manner, making it challenging to adapt to and deflect. Centralized
monolithic security systems need to evolve as well into agile distributed
solutions that combine more than one techniques, to fit better to the
Edge Computing paradigm. Hence, task offloading solutions should
be enhanced by taking into consideration security and confidentiality
constraints.
5.7. Fault tolerance
Apart from security and privacy, an important factor contributing
to building trust towards task offloading at the Edge is fault tolerance.
As thoroughly discussed in the previous sections, mobility support is
one of the most important requirements during task offloading and
this is because autonomy of communication and freedom of movement
19
are crucial criteria when it comes to users’ satisfaction. Still, there
are certain obstacles in achieving seamless connectivity and uninter-
rupted access to an Edge server while moving. For example, network
bandwidth and data exchange rates may vary or connection might be
lost. Thus, task offloading should be enhanced with fault tolerance
techniques to guarantee the successful transmission and execution of
the task, as well as minimize application response time and energy
consumption in end-user devices.
5.8. Control-related challenges
Although control theory is widely applied in Cloud elasticity and
resource allocation problems [283], there are still open challenges on
task offloading and Edge Computing that can be addressed by control
techniques. Apart from respecting the QoS requirements, control the-
ory is able to guarantee important system properties, e.g., stability,
positive invariant sets and ultimate boundedness [284], against the
inherent system’s uncertainties and external disturbances. Intermittent
connectivity, the innate management features of the virtualization
technologies and the limited resources at the edge of the network,
lead to a highly volatile dynamic environment that necessitates ad-
vanced modeling and control methodologies. Regarding the modeling
of the offloading-based applications, control theory provides many
modeling alternatives involving switching systems [145,284], Linear
Parameter Varying (LPV) systems [285,286] and Fuzzy Takagi–Sugeno
systems [287,288], that allow the natural incorporation of uncertainties
and disturbances in the performance model.
5.9. Controller design for cyber–physical systems (CPS)
Another active and very practical challenge in the context of IoT
and mobile-enabled computing, is the controller design for cyber–
physical systems (CPS), found e.g., in manufacturing, transportation
and collaborative robotics. In the context of dynamic networks and
remote computing, a joint co-design decision making strategy for task
offloading, resource allocation and controller design for CPS, is more
appropriate when compared to separate layers of controllers for the
infrastructure resources and the system to be controlled. This new
generation of controllers will be made possible by merging two sets
of models, namely (a) the performance model described above and
(b) the process model (having, for example, variables related to po-
sition, orientation and velocity of mobile devices, lighting conditions,
room temperature and mode of operation of sensors). The co-designed
controllers should address many of the non-idealities of the dynamic
networks found during task offloading, where resources must be used
parsimoniously, in balance with the constraints and the overall objec-
tive [289]. It is worth mentioning that in control engineering, shared
and imperfect communication networks between the controller and the
sensor/actuator have been studied extensively, generating the branch
of Networked Control Systems (NCS) [290]. Several developed methods
address time delays and packet dropouts of NCS, utilizing perturba-
tion theory, Lyapunov stability theory and hybrid systems analysis,
including probabilistic methods involving Markov chains and stochas-
tic automata [291–293]. A breakthrough will be the emergence of
event-triggered and self-triggered control, that allows asynchronous
sampling, thus reducing the network traffic, while at the same time
providing guaranteed trade-offs of the degradation of the closed-loop
system performance [294]. Consequently, in the context of task offload-
ing, a timely challenge is to provide co-design control formulations
that develop dynamic task offloading as well as control design algo-
rithms for CPS, taking simultaneously into account the schedulability,
available and requested network resources, Edge resources and energy
consumption. It is anticipated that such control algorithms will improve
performance, utilization of the underlying infrastructure as well as
resilience and robustness of the systems to be controlled.
Computer Networks 195 (2021) 108177F. Saeik et al.
d
i
n
t
A
d
c
&
–
C
r
D
c
i
A
N
R
6. Conclusion
In this paper, we have presented a detailed and comprehensive
study of the task offloading problem and we have extensively ana-
lyzed the main components and different computing paradigms of an
end-to-end communication path. This path starts at the end device
(i.e., mobile/IoT devices) and leverages the benefits of the added
computational resources at the edge of the network, before ending
up to the Cloud. Nonetheless, offloading parts of an application from
an end device to a remote Edge or Cloud site, arises a number of
challenges mostly related to the dynamic network behavior and the
resource allocation problem to be solved. Between these challenges, the
type of end devices in terms of mobility is expected to be one of the
main characteristics of task offloading that can dictate the scheduling
and allocation of the offloaded tasks. Prominent solutions addressing
these challenges have given the necessary momentum to task offloading
in order to be considered as a viable solution not only to existing, but
to emerging and upcoming applications such as immersive applications,
autonomous vehicles and robotics.
The benefits of task offloading are numerous, allowing to increase
the QoS and QoE of the application, while extending the battery
lifetime of the end devices. To achieve that, specific emphasis is given
on the optimization models used along with the different objectives,
in order to partition and allocate the tasks of end devices into an end-
to-end communication path that creates a user to Cloud continuum.
Since this problem contains various dynamic parameters in terms of
network and user behavior, appropriate AI and ML techniques to
anticipate traffic demands have also been presented. Control theory
techniques have also been investigated as an alternative way to address
the uncertainties of this dynamic problem, with the aim to ensure the
necessary stability of the task offloading systems. Our survey concluded
with some interesting open challenges that will shape and transform the
task offloading problem for future network scenarios and applications.
These future directions emphasize on control co-design, dynamic and
real-time allocation in a heterogeneous Edge environment, and secure
and highly mobile network platforms.
CRediT authorship contribution statement
Firdose Saeik: Conceptualization, Investigation, Writing – original
raft, Writing – review & editing. Marios Avgeris: Conceptualization,
Investigation, Writing – original draft, Writing – review & editing. Dim-
trios Spatharakis: Conceptualization, Investigation, Writing – origi-
al draft, Writing – review & editing. Nina Santi: Conceptualization,
Investigation, Writing – original draft, Writing – review & editing.
Dimitrios Dechouniotis: Conceptualization, Investigation, Writing –
original draft, Writing – review & editing. John Violos: Conceptualiza-
ion, Investigation, Writing – original draft, Writing – review & editing.
ris Leivadeas: Conceptualization, Investigation, Writing – original
raft, Writing – review & editing. Nikolaos Athanasopoulos: Con-
eptualization, Investigation, Writing – original draft, Writing – review
editing. Nathalie Mitton: Conceptualization, Investigation, Writing
original draft, Writing – review & editing. Symeon Papavassiliou:
onceptualization, Investigation, Writing – original draft, Writing –
eview & editing.
eclaration of competing interest
The authors declare that they have no known competing finan-
ial interests or personal relationships that could have appeared to
nfluence the work reported in this paper.
cknowledgment
This work was supported in part by the CHIST-ERA-2018-DRUID-
ET project, France.
20
eferences
[1] A. Gupta, R.K. Jha, A survey of 5G network: Architecture and emerging
technologies, IEEE Access 3 (2015) 1206–1232.
[2] M. Series, IMT Vision–framework and overall objectives of the future
development of IMT for 2020 and beyond, Recomm. ITU (2015) 2083–0.
[3] M. Altamimi, A. Abdrabou, K. Naik, A. Nayak, Energy cost models of smart-
phones for task offloading to the cloud, IEEE Trans. Emerg. Top. Comput. 3
(3) (2015) 384–398.
[4] X. Ma, Y. Zhao, L. Zhang, H. Wang, L. Peng, When mobile terminals meet the
cloud: computation offloading as the bridge, IEEE Netw. 27 (5) (2013) 28–33.
[5] W. Yu, F. Liang, X. He, W.G. Hatcher, C. Lu, J. Lin, X. Yang, A survey on the
edge computing for the Internet of Things, IEEE Access 6 (2017) 6900–6919.
[6] P. Mach, Z. Becvar, Mobile edge computing: A survey on architecture and
computation offloading, IEEE Commun. Surv. Tutor. 19 (3) (2017) 1628–1656.
[7] C. Jiang, X. Cheng, H. Gao, X. Zhou, J. Wan, Toward computation offloading
in edge computing: A survey, IEEE Access 7 (2019) 131543–131558.
[8] Y. Mao, C. You, J. Zhang, K. Huang, K.B. Letaief, A survey on mobile edge
computing: The communication perspective, IEEE Commun. Surv. Tutor. 19
(4) (2017) 2322–2358.
[9] C.-H. Hong, B. Varghese, Resource management in fog/edge computing: a
survey on architectures, infrastructure, and algorithms, ACM Comput. Surv. 52
(5) (2019) 1–37.
[10] S. Wang, X. Zhang, Y. Zhang, L. Wang, J. Yang, W. Wang, A survey on mobile
edge networks: Convergence of computing, caching and communications, IEEE
Access 5 (2017) 6757–6779.
[11] T.K. Rodrigues, K. Suto, H. Nishiyama, J. Liu, N. Kato, Machine learning meets
computation and communication control in evolving edge and cloud: Challenges
and future perspective, IEEE Commun. Surv. Tutor. (2019).
[12] J. Wang, J. Pan, F. Esposito, P. Calyam, Z. Yang, P. Mohapatra, Edge cloud
offloading algorithms: Issues, methods, and perspectives, ACM Comput. Surv.
52 (1) (2019) 1–23.
[13] Q. Pham, F. Fang, V.N. Ha, M.J. Piran, M. Le, L.B. Le, W. Hwang, Z.
Ding, A survey of multi-access edge computing in 5G and beyond: Funda-
mentals, technology integration, and state-of-the-art, IEEE Access 8 (2020)
116974–117017.
[14] A. Yousefpour, C. Fung, T. Nguyen, K. Kadiyala, F. Jalali, A. Niakanlahiji, J.
Kong, J.P. Jue, All one needs to know about fog computing and related edge
computing paradigms: A complete survey, J. Syst. Archit. 98 (2019) 289–330.
[15] L. Lin, X. Liao, H. Jin, P. Li, Computation offloading toward edge computing,
Proc. IEEE 107 (8) (2019) 1584–1607.
[16] X. Shan, H. Zhi, P. Li, Z. Han, A survey on computation offloading for mobile
edge computing information, in: 2018 IEEE 4th International Conference on
Big Data Security on Cloud (BigDataSecurity), IEEE International Conference
on High Performance and Smart Computing, (HPSC) and IEEE International
Conference on Intelligent Data and Security (IDS), IEEE, 2018, pp. 248–251.
[17] Y. Ai, M. Peng, K. Zhang, Edge computing technologies for Internet of Things:
a primer, Digit. Commun. Netw. 4 (2) (2018) 77–86.
[18] W.Z. Khan, E. Ahmed, S. Hakak, I. Yaqoob, A. Ahmed, Edge computing: A
survey, Future Gener. Comput. Syst. 97 (2019) 219–235.
[19] L. Mendiboure, M.-A. Chalouf, F. Krief, Edge computing based applications in
vehicular environments: Comparative study and main issues, J. Comput. Sci.
Tech. 34 (4) (2019) 869–886.
[20] P. Mell, T. Grance, et al., The NIST Definition of Cloud Computing, Computer
Security Division, National Information Technology Laboratory, 2011.
[21] L. Mei, W.K. Chan, T. Tse, A tale of clouds: Paradigm comparisons and some
thoughts on research issues, in: 2008 IEEE Asia-Pacific Services Computing
Conference, IEEE, 2008, pp. 464–469.
[22] N. Fernando, S.W. Loke, W. Rahayu, Dynamic mobile cloud computing: Ad hoc
and opportunistic job sharing, in: 2011 Fourth IEEE International Conference
on Utility and Cloud Computing, IEEE, 2011, pp. 281–286.
[23] G. Huerta-Canepa, D. Lee, A virtual cloud computing provider for mobile
devices, in: Proceedings of the 1st ACM Workshop on Mobile Cloud Computing
& Services: Social Networks and beyond, ACM, 2010, p. 6.
[24] A.U.R. Khan, M. Othman, S.A. Madani, S.U. Khan, A survey of mobile cloud
computing application models, IEEE Commun. Surv. Tutor. 16 (1) (2014)
393–413.
[25] N.I.M. Enzai, M. Tang, A taxonomy of computation offloading in mobile
cloud computing, in: 2014 2nd IEEE International Conference on Mobile Cloud
Computing, Services, and Engineering, 2014, pp. 19–28.
[26] M. Satyanarayanan, V. Bahl, R. Caceres, N. Davies, The case for vm-based
cloudlets in mobile computing, IEEE Pervas. Comput. (2009).
[27] F. Bonomi, R. Milito, J. Zhu, S. Addepalli, Fog computing and its role in the
Internet of Things, in; Proceedings of the First Edition of the MCC Workshop
on Mobile Cloud Computing, 2012, pp. 13–16.
[28] S. Antonio, Cisco delivers vision of fog computing to accelerate value from
billions of connected devices, Cisco (2014).
[29] O. Consortium, et al., Openfog reference architecture for fog computing, Archit.
Work. Group (2017).
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb1
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb1
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb1
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb2
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb2
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb2
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb3
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb3
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb3
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb3
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb3
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb4
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb4
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb4
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb5
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb5
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb5
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb6
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb6
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb6
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb7
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb7
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb7
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb8
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb8
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb8
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb8
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb8
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb9
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb9
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb9
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb9
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb9
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb10
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb10
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb10
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb10
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb10
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb11
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb11
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb11
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb11
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb11
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb12
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb12
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb12
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb12
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb12
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb13
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb14
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb14
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb14
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb14
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb14
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb15
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb15
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb15
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb16
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb17
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb17
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb17
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb18
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb18
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb18
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb19
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb19
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb19
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb19
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb19
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb20
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb20
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb20
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb21
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb21
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb21
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb21
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb21
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb22
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb22
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb22
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb22
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb22
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb23
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb23
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb23
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb23
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb23
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb24
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb24
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb24
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb24
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb24
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb26
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb26
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb26
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb28
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb28
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb28
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb29
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb29
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb29
Computer Networks 195 (2021) 108177F. Saeik et al.
[30] R. Cisco, M.Y. Upc, M. Nemirovsky, Fog computing, in: Proc. Cloud Assist.
Serveys Eur. Conf. Bled, 2012, pp. 1–15.
[31] F. Bonomi, R. Milito, P. Natarajan, J. Zhu, Fog computing: A platform for
Internet of Things and analytics, in: Big Data and Internet of Things: A Roadmap
for Smart Environments, Springer, 2014, pp. 169–186.
[32] M. Chiang, T. Zhang, Fog and IoT: An overview of research opportunities, IEEE
Internet Things J. 3 (6) (2016) 854–864.
[33] X. Hou, Y. Li, M. Chen, D. Wu, D. Jin, S. Chen, Vehicular fog computing: A
viewpoint of vehicles as the infrastructures, IEEE Trans. Veh. Technol. 65 (6)
(2016) 3860–3873.
[34] S.A. Soleymani, A.H. Abdullah, M. Zareei, M.H. Anisi, C. Vargas-Rosales, M.K.
Khan, S. Goudarzi, A secure trust model based on fuzzy logic in vehicular ad
hoc networks with fog computing, IEEE Access 5 (2017) 15619–15629.
[35] M.S. Elbamby, M. Bennis, W. Saad, Proactive edge computing in latency-
constrained fog networks, in: 2017 European Conference on Networks and
Communications, EuCNC, IEEE, 2017, pp. 1–6.
[36] D. You, T.V. Doan, R. Torre, M. Mehrabi, A. Kropp, V. Nguyen, H. Salah, G.T.
Nguyen, F.H. Fitzek, Fog computing as an enabler for immersive media: Service
scenarios and research opportunities, IEEE Access 7 (2019) 65797–65810.
[37] M. Iorga, L. Feldman, R. Barton, M. Martin, N. Goren, C. Mahmoudi, The Nist
Definition of Fog Computing, Tech. Rep., National Institute of Standards and
Technology, 2017.
[38] Y.C. Hu, M. Patel, D. Sabella, N. Sprecher, V. Young, Mobile edge computing—A
key technology towards 5G, ETSI White Paper 11 (11) (2015) 1–16.
[39] M. ETSI, Mobile edge computing (mec); framework and reference architecture,
in: ETSI, DGS MEC, vol. 3, 2016.
[40] D. Sabella, A. Vaillant, P. Kuure, U. Rauschenbach, F. Giust, Mobile-edge
computing architecture: The role of MEC in the Internet of Things, IEEE
Consumer Electron. Mag. 5 (4) (2016) 84–91.
[41] K. Zhang, Y. Mao, S. Leng, Q. Zhao, L. Li, X. Peng, L. Pan, S. Mahar-
jan, Y. Zhang, Energy-efficient offloading for mobile edge computing in 5G
heterogeneous networks, IEEE Access 4 (2016) 5896–5907.
[42] I. Ketykó, L. Kecskés, C. Nemes, L. Farkas, Multi-user computation offloading
as multiple knapsack problem for 5G mobile edge computing, in: 2016
European Conference on Networks and Communications, EuCNC, IEEE, 2016,
pp. 225–229.
[43] K. Zhang, Y. Mao, S. Leng, A. Vinel, Y. Zhang, Delay constrained offloading
for mobile edge computing in cloud-enabled vehicular networks, in: 2016 8th
International Workshop on Resilient Networks Design and Modeling, RNDM,
IEEE, 2016, pp. 288–294.
[44] T. Taleb, S. Dutta, A. Ksentini, M. Iqbal, H. Flinck, Mobile edge computing
potential in making cities smarter, IEEE Commun. Mag. 55 (3) (2017).
[45] T.X. Tran, M.-P. Hosseini, D. Pompili, Mobile edge computing: Recent efforts
and five key research directions, IEEE COMSOC MMTC Commun.-Frontiers
(2017).
[46] T. Taleb, K. Samdanis, B. Mada, H. Flinck, S. Dutta, D. Sabella, On multi-access
edge computing: A survey of the emerging 5G network edge cloud architecture
and orchestration, IEEE Commun. Surv. Tutor. 19 (3) (2017) 1657–1681.
[47] S. Kekki, W. Featherstone, Y. Fang, P. Kuure, A. Li, A. Ranjan, D. Purkayastha,
F. Jiangping, D. Frydman, G. Verin, et al., MEC In 5G networks, ETSI White
Paper 28 (2018) 1–28.
[48] V. Bahl, Emergence of micro datacenter (cloudlets/edges) for mobile computing,
in: Microsoft Devices & Networking Summit 2015, 2015.
[49] European Telecommunication Standards Institute (ETSI), Multi-access Edge
Computing (MEC), https://www.etsi.org/technologies/multi-access-edge-
computing.
[50] E. Miluzzo, T. Wang, A.T. Campbell, Eyephone: activating mobile phones
with your eyes, in: Proceedings of the Second ACM SIGCOMM Workshop on
Networking, Systems, and Applications on Mobile Handhelds, ACM, 2010, pp.
15–20.
[51] R. Kemp, N. Palmer, T. Kielmann, H. Bal, Cuckoo: a computation offloading
framework for smartphones, in: International Conference on Mobile Computing,
Applications, and Services, Springer, 2010, pp. 59–79.
[52] R. Kemp, N. Palmer, T. Kielmann, H. Bal, Opportunistic communication for
multiplayer mobile gaming: Lessons learned from photoshoot, in: Proceedings
of the Second International Workshop on Mobile Opportunistic Networking,
ACM, 2010, pp. 182–184.
[53] S. Bohez, J. De Turck, T. Verbelen, P. Simoens, B. Dhoedt, Mobile, collaborative
augmented reality using cloudlets, in: 2013 International Conference on MOBILe
Wireless MiddleWARE, Operating Systems, and Applications, IEEE, 2013, pp.
45–54.
[54] T. Verbelen, P. Simoens, F. De Turck, B. Dhoedt, Leveraging cloudlets for
immersive collaborative applications, IEEE Pervas. Comput. 12 (4) (2013)
30–38.
[55] J.-M. Chung, Y.-S. Park, J.-H. Park, H. Cho, Adaptive cloud offloading of aug-
mented reality applications on smart devices for minimum energy consumption,
Ksii Trans. Internet Inf. Syst. 9 (8) (2015).
21
[56] U. Drolia, R. Martins, J. Tan, A. Chheda, M. Sanghavi, R. Gandhi, P.
Narasimhan, The case for mobile edge-clouds, in: 2013 IEEE 10th International
Conference on Ubiquitous Intelligence and Computing and 2013 IEEE 10th
International Conference on Autonomic and Trusted Computing, IEEE, 2013,
pp. 209–215.
[57] J. Dolezal, Z. Becvar, T. Zeman, Performance evaluation of computation
offloading from mobile device to the edge of mobile network, in: 2016 IEEE
Conference on Standards for Communications and Networking, CSCN, IEEE,
2016, pp. 1–7.
[58] I.-S. Comşa, G.-M. Muntean, R. Trestian, An innovative machine-learning-based
scheduling solution for improving live UHD video streaming quality in highly
dynamic network environments, IEEE Trans. Broadcast. (2020).
[59] S. Mangiante, G. Klas, A. Navon, Z. GuanHua, J. Ran, M.D. Silva, VR is on
the edge: How to deliver 360 videos in mobile networks, in: Proceedings of
the Workshop on Virtual Reality and Augmented Reality Network, 2017, pp.
30–35.
[60] J. Chakareski, VR/AR immersive communication: Caching, edge computing, and
transmission trade-offs, in: Proceedings of the Workshop on Virtual Reality and
Augmented Reality Network, 2017, pp. 36–41.
[61] Q. Liu, S. Huang, J. Opadere, T. Han, An edge network orchestrator for mobile
augmented reality, in: IEEE INFOCOM 2018-IEEE Conference on Computer
Communications, IEEE, 2018, pp. 756–764.
[62] E. Fraedrich, R. Cyganski, I. Wolf, B. Lenz, User Perspectives on Au-
tonomous Driving: A Use-Case-Driven Study in Germany, Geographisches
Institut, Humboldt-Universität zu Berlin, 2016.
[63] U. Puetzschler, LTE and Car2x: Connected cars on the way to 5G, in: Mobile
Broadband SIG, vol. 6, 2016.
[64] Y. Sahni, J. Cao, S. Zhang, L. Yang, Edge mesh: A new paradigm to
enable distributed intelligence in Internet of Things, IEEE Access 5 (2017)
16441–16458.
[65] X. Wang, Z. Ning, L. Wang, Offloading in internet of vehicles: A fog-enabled
real-time traffic management system, IEEE Trans. Ind. Inf. 14 (10) (2018)
4568–4578.
[66] G. Hu, W.P. Tay, Y. Wen, Cloud robotics: architecture, challenges and
applications, IEEE Netw. 26 (3) (2012) 21–28.
[67] B. Kehoe, S. Patil, P. Abbeel, K. Goldberg, A survey of research on cloud robotics
and automation, IEEE Trans. Autom. Sci. Eng. 12 (2) (2015) 398–409.
[68] S. Jordan, T. Haidegger, L. Kovács, I. Felde, I. Rudas, The rising prospects
of cloud robotic applications, in: 2013 IEEE 9th International Conference on
Computational Cybernetics, ICCC, IEEE, 2013, pp. 327–332.
[69] D. Song, A.K. Tanwani, K. Goldberg, B. Siciliano, Networked-, Cloud-and
Fog-Robotics, Springer, 2019.
[70] A.K. Tanwani, R. Anand, J.E. Gonzalez, K. Goldberg, RILaaS: Robot Inference
and Learning as a Service, IEEE Robotics Autom. Lett. (2020).
[71] K. Bekris, R. Shome, A. Krontiris, A. Dobson, Cloud automation: Precomputing
roadmaps for flexible manipulation, IEEE Robot. Autom. Mag. 22 (2) (2015)
41–50.
[72] J. Glover, D. Rus, N. Roy, Probabilistic models of object geometry for
grasp planning, in: Proceedings of Robotics: Science and Systems IV, Zurich,
Switzerland, 2008, pp. 278–285.
[73] L. Riazuelo, J. Civera, J.M. Montiel, C2tam: A cloud framework for cooperative
tracking and mapping, Robot. Auton. Syst. 62 (4) (2014) 401–413.
[74] L. Turnbull, B. Samanta, Cloud robotics: Formation control of a multi robot sys-
tem utilizing cloud infrastructure, in: 2013 Proceedings of IEEE Southeastcon,
IEEE, 2013, pp. 1–4.
[75] G. Mohanarajah, V. Usenko, M. Singh, R. D’Andrea, M. Waibel, Cloud-based
collaborative 3D mapping in real-time with low-cost robots, IEEE Trans. Autom.
Sci. Eng. 12 (2) (2015) 423–431.
[76] N. Tian, A.K. Tawani, K. Goldberg, S. Sojoudi, Motion segmentation and
synthesis for latency mitigation in a cloud robotic tele-operation system, in:
International Symposium on Robotics Research, 2019.
[77] S.L. Bowman, N. Atanasov, K. Daniilidis, G.J. Pappas, Probabilistic data asso-
ciation for semantic slam, in: 2017 IEEE International Conference on Robotics
and Automation, ICRA, IEEE, 2017, pp. 1722–1729.
[78] L. Riazuelo, M. Tenorth, D. Marco, M. Salas, L. Mosenlechner, L. Kunze,
M. Beetz, J. Tardos, L. Montano, J. Montiel, Roboearth web-enabled and
knowledge-based active perception, in: IROS Workshop on AI-Based Robotics,
2013.
[79] A.K. Tanwani, N. Mor, J. Kubiatowicz, J.E. Gonzalez, K. Goldberg, A fog
robotics approach to deep robot learning: Application to object recognition and
grasp planning in surface decluttering, in: 2019 International Conference on
Robotics and Automation, ICRA, IEEE, 2019, pp. 4559–4566.
[80] A.W. Services, AWS robomaker [online], 2021, https://aws.amazon.com/
robomaker/. (Accessed 20 January 2021).
[81] Google, Cloud robotics core [online], 2021, https://googlecloudrobotics.github.
io/core/, (Accessed 20 January 2021).
[82] Rapyuta robotics [online], 2021, https://www.rapyuta-robotics.com. (Accessed
20 January 2021).
[83] J. Dilley, B. Maggs, J. Parikh, H. Prokop, R. Sitaraman, B. Weihl, Globally
distributed content delivery, IEEE Internet Comput. 6 (5) (2002) 50–58.
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb31
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb31
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb31
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb31
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb31
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb32
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb32
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb32
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb33
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb33
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb33
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb33
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb33
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb34
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb34
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb34
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb34
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb34
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb35
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb35
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb35
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb35
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb35
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb36
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb36
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb36
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb36
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb36
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb37
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb37
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb37
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb37
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb37
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb38
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb38
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb38
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb39
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb39
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb39
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb40
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb40
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb40
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb40
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb40
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb41
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb41
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb41
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb41
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb41
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb42
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb43
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb44
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb44
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb44
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb45
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb45
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb45
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb45
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb45
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb46
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb46
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb46
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb46
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb46
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb47
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb47
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb47
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb47
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb47
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb48
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb48
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb48
https://www.etsi.org/technologies/multi-access-edge-computing
https://www.etsi.org/technologies/multi-access-edge-computing
https://www.etsi.org/technologies/multi-access-edge-computing
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb50
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb51
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb51
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb51
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb51
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb51
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb52
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb53
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb54
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb54
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb54
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb54
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb54
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb55
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb55
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb55
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb55
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb55
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb56
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb57
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb58
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb58
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb58
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb58
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb58
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb61
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb61
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb61
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb61
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb61
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb62
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb62
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb62
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb62
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb62
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb63
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb63
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb63
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb64
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb64
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb64
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb64
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb64
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb65
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb65
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb65
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb65
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb65
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb66
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb66
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb66
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb67
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb67
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb67
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb68
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb68
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb68
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb68
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb68
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb69
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb69
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb69
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb70
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb70
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb70
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb71
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb71
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb71
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb71
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb71
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb73
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb73
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb73
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb74
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb74
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb74
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb74
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb74
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb75
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb75
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb75
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb75
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb75
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb77
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb77
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb77
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb77
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb77
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb79
https://aws.amazon.com/robomaker/
https://aws.amazon.com/robomaker/
https://aws.amazon.com/robomaker/
https://googlecloudrobotics.github.io/core/
https://googlecloudrobotics.github.io/core/
https://googlecloudrobotics.github.io/core/
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb83
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb83
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb83
Computer Networks 195 (2021) 108177F. Saeik et al.
[84] C. Papagianni, A. Leivadeas, S. Papavassiliou, A cloud-oriented content delivery
network paradigm: Modeling and assessment, IEEE Trans. Dependable Secure
Comput. 10 (5) (2013) 287–300.
[85] K. Bilal, A. Erbad, M. Hefeeda, Crowdsourced multi-view live video streaming
using cloud computing, IEEE Access 5 (2017) 12635–12647.
[86] J.O. Fajardo, I. Taboada, F. Liberal, Improving content delivery efficiency
through multi-layer mobile edge adaptation, IEEE Netw. 29 (6) (2015) 40–46.
[87] T.X. Tran, P. Pandey, A. Hajisami, D. Pompili, Collaborative multi-bitrate video
caching and processing in mobile-edge computing networks, in: 2017 13th
Annual Conference on Wireless on-Demand Network Systems and Services,
WONS, IEEE, 2017, pp. 165–172.
[88] J. Ren, G. Yu, Y. Cai, Y. He, Latency optimization for resource allocation
in mobile-edge computation offloading, IEEE Trans. Wireless Commun. 17 (8)
(2018) 5506–5519.
[89] Y.-H. Kao, B. Krishnamachari, M.-R. Ra, F. Bai, Hermes: Latency optimal
task assignment for resource-constrained mobile computing, IEEE Trans. Mob.
Comput. 16 (11) (2017) 3056–3069.
[90] S. Yang, Y. He, X. Zheng, Fovr: Attention-based VR streaming through
bandwidth-limited wireless networks, in: 2019 16th Annual IEEE International
Conference on Sensing, Communication, and Networking, SECON, IEEE, 2019,
pp. 1–9.
[91] J. Gubbi, R. Buyya, S. Marusic, M. Palaniswami, Internet of Things (IoT): A
vision, architectural elements, and future directions, Future Gener. Comput.
Syst. 29 (7) (2013) 1645–1660.
[92] I. Farris, L. Militano, M. Nitti, L. Atzori, A. Iera, Federated edge-assisted mobile
clouds for service provisioning in heterogeneous IoT environments, in: 2015
IEEE 2nd World Forum on Internet of Things, WF-IoT, IEEE, 2015, pp. 591–596.
[93] C. Zhu, X. Li, L. Song, L. Xiang, Development of a theoretically based thermal
model for lithium ion battery pack, J. Power Sources 223 (2013) 155–164.
[94] Y. Nan, W. Li, W. Bao, F.C. Delicato, P.F. Pires, Y. Dou, A.Y. Zomaya, Adaptive
energy-aware computation offloading for cloud of things systems, IEEE Access
5 (2017) 23947–23957.
[95] I. Farris, L. Militano, M. Nitti, L. Atzori, A. Iera, Mifaas: A mobile-IoT-federation-
as-a-service model for dynamic cooperation of IoT cloud providers, Future
Gener. Comput. Syst. 70 (2017) 126–137.
[96] S.K. Sharma, X. Wang, Live data analytics with collaborative edge and cloud
processing in wireless IoT networks, IEEE Access 5 (2017) 4621–4635.
[97] A. Yousefpour, G. Ishigaki, R. Gour, J.P. Jue, On reducing IoT service delay
via fog offloading, IEEE Internet Things J. 5 (2) (2018) 998–1010.
[98] K. Kim, C.S. Hong, Optimal task-UAV-edge matching for computation offloading
in uav assisted mobile edge computing, in: 2019 20th Asia-Pacific Network
Operations and Management Symposium, APNOMS, IEEE, 2019, pp. 1–4.
[99] S. Chen, Q. Wang, J. Chen, T. Wu, An intelligent task offloading algorithm
(iTOA) for UAV network, in: 2019 IEEE Globecom Workshops, GC Wkshps,
IEEE, 2019, pp. 1–6.
[100] M. Avgeris, D. Spatharakis, D. Dechouniotis, N. Kalatzis, I. Roussaki, S.
Papavassiliou, Where there is fire there is smoke: a scalable edge computing
framework for early fire detection, Sensors 19 (3) (2019) 639.
[101] J. Wang, M. Conrad Meyer, Y. Wu, Y. Wang, Maximum data-resolution
efficiency for fog-computing supported spatial big data processing in disaster
scenarios, IEEE Trans. Parallel Distrib. Syst. 30 (2019) 1826–1842.
[102] J. Zhou, J. Fan, J. wang, J. Zhu, Task offloading for social sensing applications
in mobile edge computing, in: 2019 Seventh International Conference on
Advanced Cloud and Big Data, CBD, IEEE, 2019, pp. 333–338.
[103] G. Ananthanarayanan, P. Bahl, P. Bodík, K. Chintalapudi, M. Philipose, L.
Ravindranath, S. Sinha, Real-time video analytics: The killer app for edge
computing, Computer 50 (10) (2017) 58–67.
[104] J. Wang, J. Pan, F. Esposito, P. Calyam, Z. Yang, P. Mohapatra, Edge cloud
offloading algorithms: Issues, methods, and perspectives, ACM Comput. Surv.
52 (1) (2019) 2:1–2:23, http://doi.acm.org/10.1145/3284387.
[105] T.X. Tran, A. Hajisami, P. Pandey, D. Pompili, Collaborative mobile edge
computing in 5G networks: New paradigms, scenarios, and challenges, IEEE
Commun. Mag. 55 (4) (2017) 54–61.
[106] J. Violos, S. Pelekis, A. Berdelis, S. Tsanakas, K. Tserpes, T. Varvarigou,
Predicting visitor distribution for large events in smart cities, in: 2019 IEEE
International Conference on Big Data and Smart Computing, BigComp, 2019,
pp. 1–8 (ISSN: 2375-9356).
[107] T.X. Tran, D. Pompili, Joint task offloading and resource allocation for multi-
server mobile-edge computing networks, IEEE Trans. Veh. Technol. 68 (1)
(2018) 856–868.
[108] J. Violos, E. Psomakelis, K. Tserpes, F. Aisopos, T. Varvarigou, Leveraging
user mobility and mobile app services behavior for optimal edge resource
utilization, in: Proceedings of the International Conference on Omni-Layer
Intelligent Systems, COINS ’19, Association for Computing Machinery, Crete,
Greece, 2019, pp. 7–12.
[109] L. Yang, H. Zhang, M. Li, J. Guo, H. Ji, Mobile edge computing empowered
energy efficient task offloading in 5G, IEEE Trans. Veh. Technol. 67 (7) (2018)
6398–6409.
[110] Z. Xu, X. Liu, G. Jiang, B. Tang, A time-efficient data offloading method with
privacy preservation for intelligent sensors in edge computing, EURASIP J.
Wireless Commun. Networking 2019 (1) (2019) 1–12.
22
[111] F. Liu, Z. Huang, L. Wang, Energy-efficient collaborative task computation
offloading in cloud-assisted edge computing for IoT sensors, Sensors 19 (5)
(2019) 1105.
[112] M. Chen, B. Liang, M. Dong, Multi-user multi-task offloading and resource
allocation in mobile cloud systems, IEEE Trans. Wireless Commun. 17 (10)
(2018) 6790–6805.
[113] Y. Cao, T. Jiang, C. Wang, Optimal radio resource allocation for mobile task
offloading in cellular networks, IEEE Netw. 28 (5) (2014) 68–73.
[114] Y. Cao, C. Long, T. Jiang, S. Mao, Share communication and computation
resources on mobile devices: a social awareness perspective, IEEE Wirel.
Commun. 23 (4) (2016) 52–59.
[115] D. Huang, P. Wang, D. Niyato, A dynamic offloading algorithm for mobile
computing, IEEE Trans. Wireless Commun. 11 (6) (2012) 1991–1995.
[116] S. Sardellitti, S. Barbarossa, G. Scutari, Distributed mobile cloud comput-
ing: Joint optimization of radio and computational resources, in: 2014 IEEE
Globecom Workshops, GC Wkshps, IEEE, 2014, pp. 1505–1510.
[117] O. Munoz, A. Pascual-Iserte, J. Vidal, Optimization of radio and computational
resources for energy efficiency in latency-constrained application offloading,
IEEE Trans. Veh. Technol. 64 (10) (2014) 4738–4755.
[118] J. Plachy, Z. Becvar, E.C. Strinati, Dynamic resource allocation exploiting
mobility prediction in mobile edge computing, in: 2016 IEEE 27th Annual Inter-
national Symposium on Personal, Indoor, and Mobile Radio Communications,
PIMRC, IEEE, 2016, pp. 1–6.
[119] T. Truong-Huu, C.-K. Tham, D. Niyato, To offload or to wait: An opportunistic
offloading algorithm for parallel tasks in a mobile cloud, in: 2014 IEEE 6th
International Conference on Cloud Computing Technology and Science, IEEE,
2014, pp. 182–189.
[120] M. Chen, Y. Hao, Y. Li, C.-F. Lai, D. Wu, On the computation offloading at
ad hoc cloudlet: architecture and service modes, IEEE Commun. Mag. 53 (6)
(2015) 18–24.
[121] L. Pu, X. Chen, J. Xu, X. Fu, D2D fogging: An energy-efficient and incentive-
aware task offloading framework via network-assisted D2D collaboration, IEEE
J. Sel. Areas Commun. 34 (12) (2016) 3887–3901.
[122] S. Zhou, Y. Sun, Z. Jiang, Z. Niu, Exploiting moving intelligence: Delay-
optimized computation offloading in vehicular fog networks, IEEE Commun.
Mag. 57 (5) (2019) 49–55.
[123] W. Li, X. You, Y. Jiang, J. Yang, L. Hu, Opportunistic computing offloading in
edge clouds, J. Parallel Distrib. Comput. 123 (2019) 69–76.
[124] Y. Sun, S. Zhou, J. Xu, EMM: Energy-aware mobility management for mobile
edge computing in ultra dense networks, IEEE J. Sel. Areas Commun. 35 (11)
(2017) 2637–2646.
[125] Z. Wang, Z. Zhao, G. Min, X. Huang, Q. Ni, R. Wang, User mobility aware task
assignment for mobile edge computing, Future Gener. Comput. Syst. 85 (2018)
1–8.
[126] G. Ahani, D. Yuan, BS-assisted task offloading for D2D networks with pres-
ence of user mobility, in: 2019 IEEE 89th Vehicular Technology Conference
(VTC2019-Spring), IEEE, 2019, pp. 1–5.
[127] Z. Ning, P. Dong, X. Wang, J.J. Rodrigues, F. Xia, Deep reinforcement learning
for vehicular edge computing: An intelligent offloading system, ACM Trans.
Intell. Syst. Technol. (TIST) 10 (6) (2019) 1–24.
[128] T.M.T. Do, O. Dousse, M. Miettinen, D. Gatica-Perez, A probabilistic kernel
method for human mobility prediction with smartphones, Pervasive Mob.
Comput. 20 (2015) 13–28.
[129] B. Li, H. Zhang, H. Lu, User mobility prediction based on Lagrange’s inter-
polation in ultra-dense networks, in: 2016 IEEE 27th Annual International
Symposium on Personal, Indoor, and Mobile Radio Communications, PIMRC,
IEEE, 2016, pp. 1–6.
[130] H.A. Alameddine, S. Sharafeddine, S. Sebbah, S. Ayoubi, C. Assi, Dynamic task
offloading and scheduling for low-latency IoT services in multi-access edge
computing, IEEE J. Sel. Areas Commun. 37 (3) (2019) 668–682.
[131] S. Wu, W. Xia, W. Cui, Q. Chao, Z. Lan, F. Yan, L. Shen, An efficient offload-
ing algorithm based on support vector machine for mobile edge computing
in vehicular networks, in: 2018 10th International Conference on Wireless
Communications and Signal Processing, WCSP, IEEE, 2018, pp. 1–6.
[132] M. Maanoja, M. Weckström, Location services, Google Patents, US Patent 7,
069, 023, 2006.
[133] S. Deng, L. Huang, D. Hu, J.L. Zhao, Z. Wu, Mobility-enabled service selection
for composite services, IEEE Trans. Serv. Comput. 9 (3) (2016) 394–407.
[134] S. Deng, L. Huang, J. Taheri, J. Yin, M. Zhou, A.Y. Zomaya, Mobility-aware
service composition in mobile communities, IEEE Trans. Syst. Man Cybern.:
Syst. 47 (3) (2017) 555–568.
[135] S. Gambs, M.-O. Killijian, M.N. del Prado Cortez, Next place prediction using
mobility Markov chains, in: Proceedings of the First Workshop on Measurement,
Privacy, and Mobility, 2012, pp. 1–6.
[136] K. Lee, I. Shin, User mobility model based computation offloading decision for
mobile cloud, J. Comput. Sci. Eng. 9 (3) (2015) 155–162.
[137] X. Sun, N. Ansari, Adaptive avatar handoff in the cloudlet network, IEEE Trans.
Cloud Comput. 7 (3) (2019) 664–676.
[138] M. Chen, Y. Hao, M. Qiu, J. Song, D. Wu, I. Humar, Mobility-aware caching
and computation offloading in 5G ultra-dense cellular networks, Sensors 16 (7)
(2016) 974.
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb84
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb84
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb84
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb84
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb84
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb85
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb85
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb85
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb86
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb86
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb86
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb87
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb88
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb88
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb88
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb88
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb88
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb89
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb89
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb89
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb89
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb89
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb90
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb91
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb91
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb91
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb91
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb91
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb92
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb92
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb92
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb92
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb92
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb93
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb93
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb93
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb94
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb94
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb94
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb94
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb94
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb95
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb95
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb95
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb95
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb95
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb96
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb96
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb96
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb97
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb97
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb97
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb98
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb98
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb98
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb98
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb98
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb99
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb99
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb99
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb99
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb99
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb100
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb100
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb100
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb100
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb100
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb101
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb101
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb101
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb101
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb101
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb102
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb102
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb102
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb102
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb102
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb103
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb103
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb103
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb103
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb103
http://doi.acm.org/10.1145/3284387
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb105
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb105
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb105
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb105
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb105
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb107
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb107
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb107
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb107
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb107
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb108
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb109
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb109
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb109
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb109
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb109
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb110
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb110
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb110
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb110
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb110
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb111
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb111
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb111
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb111
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb111
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb112
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb112
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb112
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb112
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb112
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb113
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb113
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb113
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb114
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb114
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb114
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb114
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb114
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb115
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb115
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb115
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb116
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb116
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb116
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb116
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb116
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb117
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb117
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb117
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb117
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb117
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb118
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb119
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb120
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb120
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb120
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb120
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb120
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb121
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb121
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb121
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb121
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb121
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb122
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb122
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb122
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb122
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb122
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb123
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb123
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb123
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb124
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb124
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb124
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb124
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb124
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb125
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb125
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb125
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb125
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb125
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb126
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb126
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb126
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb126
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb126
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb127
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb127
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb127
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb127
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb127
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb128
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb128
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb128
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb128
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb128
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb129
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb130
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb130
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb130
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb130
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb130
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb131
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb133
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb133
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb133
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb134
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb134
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb134
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb134
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb134
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb136
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb136
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb136
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb137
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb137
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb137
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb138
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb138
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb138
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb138
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb138
Computer Networks 195 (2021) 108177F. Saeik et al.
[139] J. Xu, L. Chen, P. Zhou, Joint service caching and task offloading for mobile
edge computing in dense networks, in: IEEE INFOCOM 2018-IEEE Conference
on Computer Communications, IEEE, 2018, pp. 207–215.
[140] D. Wang, Z. Liu, X. Wang, Y. Lan, Mobility-aware task offloading and migration
schemes in fog computing networks, IEEE Access 7 (2019) 43356–43368.
[141] Y. Shi, S. Chen, X. Xu, MAGA: A mobility-aware computation offloading
decision for distributed mobile cloud computing, IEEE Internet Things J. 5 (1)
(2017) 164–174.
[142] F. Yu, H. Chen, J. Xu, DMPO: Dynamic mobility-aware partial offloading in
mobile edge computing, Future Gener. Comput. Syst. 89 (2018) 722–735.
[143] V.A. Siris, D. Kalyvas, Enhancing mobile data offloading with mobility predic-
tion and prefetching, ACM SIGMOBILE Mob. Comput. Commun. Rev. 17 (1)
(2013) 22–29.
[144] C. Yang, Y. Liu, X. Chen, W. Zhong, S. Xie, Efficient mobility-aware task
offloading for vehicular edge computing networks, IEEE Access 7 (2019)
26652–26664.
[145] D. Spatharakis, I. Dimolitsas, D. Dechouniotis, G. Papathanail, I. Fotoglou,
P. Papadimitriou, S. Papavassiliou, A scalable edge computing architecture
enabling smart offloading for location based services, Pervasive Mob. Comput.
67 (2020) 101217.
[146] G. Papathanail, I. Fotoglou, C. Demertzis, A. Pentelas, K. Sgouromitis, P.
Papadimitriou, D. Spatharakis, I. Dimolitsas, D. Dechouniotis, S. Papavassiliou,
COSMOS: An orchestration framework for smart computation offloading in edge
clouds, in: NOMS 2020-2020 IEEE/IFIP Network Operations and Management
Symposium, IEEE, 2020, pp. 1–6.
[147] M. Akter, F.T. Zohra, A.K. Das, Q-MAC: Qos and mobility aware optimal re-
source allocation for dynamic application offloading in mobile cloud computing,
in: 2017 International Conference on Electrical, Computer and Communication
Engineering, ECCE, IEEE, 2017, pp. 803–808.
[148] C. Puliafito, E. Mingozzi, C. Vallati, F. Longo, G. Merlino, Companion fog
computing: Supporting things mobility through container migration at the edge,
in: 2018 IEEE International Conference on Smart Computing, SMARTCOMP,
2018, pp. 97–105.
[149] W. Junior, A. França, K. Dias, J.N. de Souza, Supporting mobility-aware
computational offloading in mobile cloud environment, J. Netw. Comput. Appl.
94 (2017) 93–108.
[150] L. Yang, J. Cao, H. Cheng, Y. Ji, Multi-user computation partitioning for
latency sensitive mobile cloud applications, IEEE Trans. Comput. 64 (8) (2014)
2253–2266.
[151] S. Bi, Y.J. Zhang, Computation rate maximization for wireless powered mobile-
edge computing with binary computation offloading, IEEE Trans. Wireless
Commun. 17 (6) (2018) 4177–4190.
[152] T. Zhao, S. Zhou, X. Guo, Z. Niu, Tasks scheduling and resource allocation in
heterogeneous cloud for delay-bounded mobile edge computing, in: 2017 IEEE
International Conference on Communications, ICC, IEEE, 2017, pp. 1–7.
[153] Q. Liu, T. Han, N. Ansari, Joint radio and computation resource management
for low latency mobile edge computing, in: 2018 IEEE Global Communications
Conference, GLOBECOM, IEEE, 2018, pp. 1–7.
[154] L. Yang, J. Cao, Y. Yuan, T. Li, A. Han, A. Chan, A framework for partitioning
and execution of data stream applications in mobile cloud computing, ACM
SIGMETRICS Perform. Eval. Rev. 40 (4) (2013) 23–32.
[155] J. Liu, Y. Mao, J. Zhang, K.B. Letaief, Delay-optimal computation task schedul-
ing for mobile-edge computing systems, in: 2016 IEEE International Symposium
on Information Theory, ISIT, IEEE, 2016, pp. 1451–1455.
[156] M. Jia, W. Liang, Z. Xu, M. Huang, Y. Ma, Qos-aware cloudlet load balancing
in wireless metropolitan area networks, IEEE Trans. Cloud Comput. (2018).
[157] C. Wang, C. Liang, F.R. Yu, Q. Chen, L. Tang, Computation offloading and
resource allocation in wireless cellular networks with mobile edge computing,
IEEE Trans. Wireless Commun. 16 (8) (2017) 4924–4938.
[158] C. Sonmez, A. Ozgovde, C. Ersoy, Fuzzy workload orchestration for edge
computing, IEEE Trans. Netw. Serv. Manag. (2019).
[159] C.-F. Liu, M. Bennis, H.V. Poor, Latency and reliability-aware task offloading
and resource allocation for mobile edge computing, in: 2017 IEEE Globecom
Workshops, GC Wkshps, IEEE, 2017, pp. 1–7.
[160] L. Wang, L. Jiao, J. Li, M. Mühlhäuser, Online resource allocation for
arbitrary user mobility in distributed edge clouds, in: 2017 IEEE 37th Interna-
tional Conference on Distributed Computing Systems, ICDCS, IEEE, 2017, pp.
1281–1290.
[161] S. Cao, X. Tao, Y. Hou, Q. Cui, An energy-optimal offloading algorithm of
mobile computing based on hetnets, in: 2015 International Conference on
Connected Vehicles and Expo, ICCVE, IEEE, 2015, pp. 254–258.
[162] Y. Zhao, S. Zhou, T. Zhao, Z. Niu, Energy-efficient task offloading for mul-
tiuser mobile cloud computing, in: 2015 IEEE/CIC International Conference on
Communications in China, ICCC, IEEE, 2015, pp. 1–5.
[163] Y. Wang, M. Sheng, X. Wang, L. Wang, W. Han, Y. Zhang, Y. Shi, Energy-
optimal partial computation offloading using dynamic voltage scaling, in: 2015
IEEE International Conference on Communication Workshop, ICCW, IEEE, 2015,
pp. 2695–2700.
[164] C. You, K. Huang, Multiuser resource allocation for mobile-edge computation
offloading, in: 2016 IEEE Global Communications Conference, GLOBECOM,
IEEE, 2016, pp. 1–6.
23
[165] P. Di Lorenzo, S. Barbarossa, S. Sardellitti, Joint optimization of radio resources
and code partitioning in mobile edge computing, 2013, arXiv preprint arXiv:
1307.3835.
[166] W. Labidi, M. Sarkiss, M. Kamoun, Energy-optimal resource scheduling and
computation offloading in small cell networks, in: 2015 22nd International
Conference on Telecommunications, ICT, IEEE, 2015, pp. 313–318.
[167] M. Kamoun, W. Labidi, M. Sarkiss, Joint resource allocation and offloading
strategies in cloud enabled cellular networks, in: 2015 IEEE International
Conference on Communications, ICC, IEEE, 2015, pp. 5529–5534.
[168] S. Sardellitti, G. Scutari, S. Barbarossa, Joint optimization of radio and compu-
tational resources for multicell mobile-edge computing, IEEE Trans. Signal Inf.
Process. Netw. 1 (2) (2015) 89–103.
[169] W. Labidi, M. Sarkiss, M. Kamoun, Joint multi-user resource scheduling and
computation offloading in small cell networks, in: 2015 IEEE 11th Inter-
national Conference on Wireless and Mobile Computing, Networking and
Communications, WiMob, IEEE, 2015, pp. 794–801.
[170] Y. Mao, J. Zhang, K.B. Letaief, Dynamic computation offloading for mobile-edge
computing with energy harvesting devices, IEEE J. Sel. Areas Commun. 34 (12)
(2016) 3590–3605.
[171] K. Singh, A. Awasthi, Quality, Reliability, Security and Robustness in
Heterogeneous Networks, Springer, 2013.
[172] A. Kiani, N. Ansari, Optimal code partitioning over time and hierarchical
cloudlets, IEEE Commun. Lett. 22 (1) (2017) 181–184.
[173] Y. Mao, J. Zhang, K.B. Letaief, Joint task offloading scheduling and transmit
power allocation for mobile-edge computing systems, in: 2017 IEEE Wireless
Communications and Networking Conference, WCNC, IEEE, 2017, pp. 1–6.
[174] J. Zhang, X. Hu, Z. Ning, E.C.-H. Ngai, L. Zhou, J. Wei, J. Cheng, B. Hu,
Energy-latency tradeoff for energy-aware offloading in mobile edge computing
networks, IEEE Internet Things J. 5 (4) (2017) 2633–2645.
[175] J. Guo, Z. Song, Y. Cui, Z. Liu, Y. Ji, Energy-efficient resource allocation for
multi-user mobile edge computing, in: GLOBECOM 2017-2017 IEEE Global
Communications Conference, IEEE, 2017, pp. 1–7.
[176] F. Samie, V. Tsoutsouras, L. Bauer, S. Xydis, D. Soudris, J. Henkel, Computation
offloading and resource allocation for low-power IoT edge devices, in: 2016
IEEE 3rd World Forum on Internet of Things, WF-IoT, IEEE, 2016, pp. 7–12.
[177] A. Leivadeas, C. Papagianni, S. Papavassiliou, Going Green with the Networked
Cloud: Methodologies and Assessment, John Wiley & Sons, Inc., 2015.
[178] S. Li, N. Zhang, S. Lin, L. Kong, A. Katangur, M.K. Khan, M. Ni, G. Zhu, Joint
admission control and resource allocation in edge computing for Internet of
Things, IEEE Netw. 32 (1) (2018) 72–79.
[179] L. Tong, Y. Li, W. Gao, A hierarchical edge cloud architecture for mobile com-
puting, in: IEEE INFOCOM 2016-the 35th Annual IEEE International Conference
on Computer Communications, IEEE, 2016, pp. 1–9.
[180] A. Leivadeas, G. Kesidis, M. Ibnkahla, I. Lambadaris, VNF placement
optimization at the edge and cloud, Future Internet 11 (3) (2019) 69.
[181] A. Leivadeas, G. Kesidis, M. Falkner, I. Lambadaris, A graph partitioning game
theoretical approach for the VNF service chaining problem, IEEE Trans. Netw.
Serv. Manag. 14 (4) (2017) 890–903.
[182] A. Leivadeas, M. Falkner, I. Lambadaris, M. Ibnkahla, G. Kesidis, Balancing
delay and cost in virtual network function placement and chaining, in: 2018
IEEE International Conference on Network Softwarization and Workshops,
NetSoft, IEEE, 2018, pp. 433–440.
[183] I. Skarga-Bandurova, M. Derkach, I. Kotsiuba, The information service for
delivering arrival public transport prediction, in: 2018 IEEE 4th International
Symposium on Wireless Systems Within the International Conferences on
Intelligent Data Acquisition and Advanced Computing Systems, IDAACS-SWS,
2018, pp. 191–195.
[184] A. Hameed, A. Leivadeas, IoT Traffic multi-classification using network and
statistical features in a smart environment, in: EEE International Workshop on
Computer Aided Modeling and Design of Communication Links and Networks,
CAMAD, IEEE, 2020, pp. 1–7.
[185] W. Sun, J. Liu, Y. Yue, AI-enhanced offloading in edge computing: when
machine learning meets industrial IoT, IEEE Netw. 33 (5) (2019) 68–74.
[186] A. Botchkarev, A new typology design of performance metrics to measure
errors in machine learning regression algorithms, Interdisciplinary J. Inf. Knowl.
Manag. 14 (2019) 045–076.
[187] J.-O. Palacio-Niño, F. Berzal, Evaluation metrics for unsupervised learning
algorithms, 2019, arXiv preprint arXiv:1905.05667.
[188] J. Lever, M. Krzywinski, N. Altman, Classification evaluation, Nature Methods
13 (8) (2016) 603–604, https://www.nature.com/articles/nmeth.3945.
[189] Y. Wang, M. Sheng, X. Wang, L. Wang, J. Li, Mobile-edge computing: Partial
computation offloading using dynamic voltage scaling, IEEE Trans. Commun.
64 (10) (2016) 4268–4282.
[190] M. Deng, H. Tian, B. Fan, Fine-granularity based application offloading policy
in cloud-enhanced small cell networks, in: 2016 IEEE International Conference
on Communications Workshops, ICC, IEEE, 2016, pp. 638–643.
[191] Y. Mao, J. Zhang, S. Song, K.B. Letaief, Power-delay tradeoff in multi-
user mobile-edge computing systems, in: 2016 IEEE Global Communications
Conference, GLOBECOM, IEEE, 2016, pp. 1–6.
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb139
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb139
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb139
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb139
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb139
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb140
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb140
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb140
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb141
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb141
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb141
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb141
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb141
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb142
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb142
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb142
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb143
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb143
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb143
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb143
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb143
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb144
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb144
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb144
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb144
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb144
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb145
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb146
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb147
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb149
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb149
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb149
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb149
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb149
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb150
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb150
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb150
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb150
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb150
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb151
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb151
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb151
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb151
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb151
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb152
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb152
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb152
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb152
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb152
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb153
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb153
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb153
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb153
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb153
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb154
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb154
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb154
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb154
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb154
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb155
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb155
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb155
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb155
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb155
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb156
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb156
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb156
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb157
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb157
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb157
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb157
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb157
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb158
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb158
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb158
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb159
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb159
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb159
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb159
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb159
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb160
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb161
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb161
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb161
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb161
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb161
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb162
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb162
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb162
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb162
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb162
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb163
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb164
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb164
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb164
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb164
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb164
http://arxiv.org/abs/1307.3835
http://arxiv.org/abs/1307.3835
http://arxiv.org/abs/1307.3835
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb166
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb166
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb166
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb166
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb166
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb167
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb167
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb167
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb167
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb167
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb168
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb168
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb168
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb168
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb168
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb169
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb170
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb170
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb170
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb170
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb170
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb171
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb171
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb171
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb172
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb172
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb172
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb173
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb173
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb173
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb173
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb173
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb174
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb174
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb174
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb174
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb174
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb175
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb175
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb175
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb175
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb175
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb176
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb176
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb176
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb176
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb176
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb177
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb177
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb177
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb178
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb178
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb178
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb178
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb178
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb179
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb179
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb179
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb179
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb179
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb180
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb180
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb180
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb181
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb181
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb181
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb181
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb181
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb182
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb184
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb185
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb185
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb185
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb186
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb186
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb186
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb186
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb186
http://arxiv.org/abs/1905.05667
https://www.nature.com/articles/nmeth.3945
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb189
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb189
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb189
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb189
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb189
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb190
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb190
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb190
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb190
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb190
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb191
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb191
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb191
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb191
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb191
Computer Networks 195 (2021) 108177F. Saeik et al.
[192] C. You, K. Huang, H. Chae, B.-H. Kim, Energy-efficient resource allocation
for mobile-edge computation offloading, IEEE Trans. Wireless Commun. 16 (3)
(2016) 1397–1411.
[193] M.-H. Chen, B. Liang, M. Dong, Joint offloading and resource allocation for
computation and communication in mobile cloud with computing access point,
in: IEEE INFOCOM 2017-IEEE Conference on Computer Communications, IEEE,
2017, pp. 1–9.
[194] M.-H. Chen, B. Liang, M. Dong, A semidefinite relaxation approach to mobile
cloud offloading with computing access point, in: 2015 IEEE 16th International
Workshop on Signal Processing Advances in Wireless Communications, SPAWC,
IEEE, 2015, pp. 186–190.
[195] Y. Liu, F.R. Yu, X. Li, H. Ji, V.C. Leung, Hybrid computation offloading in fog
and cloud networks with non-orthogonal multiple access, in: IEEE INFOCOM
2018-IEEE Conference on Computer Communications Workshops, INFOCOM
WKSHPS, IEEE, 2018, pp. 154–159.
[196] Q.-V. Pham, T. Leanh, N.H. Tran, B.J. Park, C.S. Hong, Decentralized computa-
tion offloading and resource allocation for mobile-edge computing: A matching
game approach, IEEE Access 6 (2018) 75868–75885.
[197] Z. Ning, P. Dong, X. Kong, F. Xia, A cooperative partial computation offloading
scheme for mobile edge computing enabled Internet of Things, IEEE Internet
Things J. (2018).
[198] J. Du, L. Zhao, J. Feng, X. Chu, Computation offloading and resource allocation
in mixed fog/cloud computing systems with min-max fairness guarantee, IEEE
Trans. Commun. 66 (4) (2018) 1594–1608.
[199] S. Li, Y. Tao, X. Qin, L. Liu, Z. Zhang, P. Zhang, Energy-aware mobile edge
computation offloading for IoT over heterogenous networks, IEEE Access 7
(2019) 13092–13105.
[200] H. Guo, J. Liu, J. Zhang, Computation offloading for multi-access mobile edge
computing in ultra-dense networks, IEEE Commun. Mag. 56 (8) (2018) 14–19.
[201] X. Chen, L. Jiao, W. Li, X. Fu, Efficient multi-user computation offloading
for mobile-edge cloud computing, IEEE/ACM Trans. Netw. 24 (5) (2015)
2795–2808.
[202] Y. Liu, C. Xu, Y. Zhan, Z. Liu, J. Guan, H. Zhang, Incentive mechanism for
computation offloading using edge computing: A Stackelberg game approach,
Comput. Netw. 129 (2017) 399–409.
[203] H. Guo, J. Liu, Collaborative computation offloading for multiaccess edge
computing over fiber–wireless networks, IEEE Trans. Veh. Technol. 67 (5)
(2018) 4514–4526.
[204] M. Zeng, Y. Li, K. Zhang, M. Waqas, D. Jin, Incentive mechanism design
for computation offloading in heterogeneous fog computing: A contract-based
approach, in: 2018 IEEE International Conference on Communications, ICC,
2018, pp. 1–6.
[205] J. Du, E. Gelenbe, C. Jiang, H. Zhang, Y. Ren, Contract design for traffic
offloading and resource allocation in heterogeneous ultra-dense networks, IEEE
J. Sel. Areas Commun. 35 (11) (2017) 2457–2467.
[206] Z. Hu, Z. Zheng, L. Song, T. Wang, X. Li, UAV offloading: Spectrum trading con-
tract design for UAV-assisted cellular networks, IEEE Trans. Wireless Commun.
17 (9) (2018) 6093–6107.
[207] Y. Li, J. Zhang, X. Gan, L. Fu, H. Yu, X. Wang, A contract-based incentive
mechanism for delayed traffic offloading in cellular networks, IEEE Trans.
Wireless Commun. 15 (8) (2016) 5314–5327.
[208] M.G.C. Resende, P. Pardalos, Handbook of Optimization in Telecommunication,
Springer, 2006.
[209] D. Monderer, L.S. Shapley, Potential games, Games Economic Behav. 14 (1)
(1996) 124–143.
[210] Y. Zhang, M. Pan, L. Song, Z. Dawy, Z. Han, A survey of contract theory-based
incentive mechanism design in wireless networks, IEEE Wirel. Commun. 24 (3)
(2017) 80–85.
[211] M. Gendreau, J.Y. Potvin, Handbook of Metaheuristics, Int. Series in Operations
Research & Management Science, 2010.
[212] K.B. Letaief, W. Chen, Y. Shi, J. Zhang, Y.-J.A. Zhang, The roadmap to 6G: AI
empowered wireless networks, IEEE Commun. Mag. 57 (8) (2019) 84–90.
[213] F.A. Salaht, F. Desprez, A. Lebre, C. Prud’Homme, M. Abderrahim, Service
placement in fog computing using constraint programming, in: 2019 IEEE
International Conference on Services Computing, SCC, IEEE, 2019, pp. 19–27.
[214] A.I. Orhean, F. Pop, I. Raicu, New scheduling approach using reinforcement
learning for heterogeneous distributed systems, J. Parallel Distrib. Comput. 117
(2018) 292–302.
[215] Y.-R. Shiue, K.-C. Lee, C.-T. Su, Real-time scheduling for a smart factory using
a reinforcement learning approach, Comput. Ind. Eng. 125 (2018) 604–614.
[216] S. Wang, T. Tuor, T. Salonidis, K.K. Leung, C. Makaya, T. He, K. Chan, Adaptive
federated learning in resource constrained edge computing systems, IEEE J. Sel.
Areas Commun. 37 (6) (2019) 1205–1221.
[217] L.C. Monteiro, F.C. Delicato, L. Pirmez, P.F. Pires, C. Miceli, Dpcas: Data
prediction with cubic adaptive sampling for wireless sensor networks, in:
International Conference on Green, Pervasive, and Cloud Computing, Springer,
2017, pp. 353–368.
[218] B. Yang, X. Cao, J. Bassey, X. Li, T. Kroecker, L. Qian, Computation offloading
in multi-access edge computing networks: A multi-task learning approach, in:
ICC 2019 – 2019 IEEE International Conference on Communications, ICC, 2019,
pp. 1–6 (ISSN: 1938-1883).
24
[219] D. Rahbari, M. Nickray, Task offloading in mobile fog computing by
classification and regression tree, Peer-to-Peer Netw. Appl. 13 (1) (2020)
104–122.
[220] W. Junior, E. Oliveira, A. Santos, K. Dias, A context-sensitive offloading system
using machine-learning classification algorithms for mobile cloud environment,
Future Gener. Comput. Syst. 90 (2019) 503–520.
[221] H. Bashir, S. Lee, K.H. Kim, Resource allocation through logistic regression
and multicriteria decision making method in IoT fog computing, Trans. Emerg.
Telecommun. Technol. n/a (n/a) (2019) e3824, _eprint: https://onlinelibrary.
wiley.com/doi/pdf/10.1002/ett.3824, https://onlinelibrary.wiley.com/doi/abs/
10.1002/ett.3824.
[222] Y. Jararweh, M.B. Issa, M. Daraghmeh, M. Al-Ayyoub, M.A. Alsmirat, Energy
efficient dynamic resource management in cloud computing based on logistic
regression model and median absolute deviation, Sustain. Comput.: Inform. Syst.
19 (2018) 262–274.
[223] J.C. Guevara, R.d.S. Torres, N.L.S. da Fonseca, On the classification of fog
computing applications: A machine learning perspective, J. Netw. Comput.
Appl. 159 (2020) 102596.
[224] L. Liu, D. Qi, N. Zhou, Y. Wu, A task scheduling algorithm based on classifi-
cation mining in fog computing environment, Wirel. Commun. Mob. Comput.
(ISSN: 1530-8669) 2018 (2018) e2102348, https://www.hindawi.com/journals/
wcmc/2018/2102348/.
[225] K. Kim, J. Lynskey, S. Kang, C.S. Hong, Prediction based sub-task offloading
in mobile edge computing, in: 2019 International Conference on Information
Networking, ICOIN, 2019, pp. 448–452 (ISSN: 1976-7684).
[226] F. Farahnakian, P. Liljeberg, J. Plosila, LiRCUP: Linear regression based cpu
usage prediction algorithm for live migration of virtual machines in data
centers, in: 2013 39th Euromicro Conference on Software Engineering and
Advanced Applications, 2013, pp. 357–364 (ISSN: 2376-9505).
[227] R.A.C. da Silva, N.L.S.d. Fonseca, Resource allocation mechanism for a fog-cloud
infrastructure, in: 2018 IEEE International Conference on Communications, ICC,
2018, pp. 1–6 (ISSN: 1938-1883).
[228] R. Hu, J. Jiang, G. Liu, L. Wang, CPU load prediction using support vector
regression and Kalman Smoother for Cloud, in: 2013 IEEE 33rd International
Conference on Distributed Computing Systems Workshops, 2013, pp. 88–92
(ISSN: 2332-5666).
[229] F. Farahnakian, T. Pahikkala, P. Liljeberg, J. Plosila, Energy aware consolidation
algorithm based on k-nearest neighbor regression for cloud data centers, in:
2013 IEEE/ACM 6th International Conference on Utility and Cloud Computing,
2013, pp. 256–259.
[230] H. Cheng, W. Xia, F. Yan, L. Shen, Balanced clustering and joint re-
sources allocation in cooperative fog computing system, in: 2019 IEEE Global
Communications Conference, GLOBECOM, 2019, pp. 1–6 (ISSN: 2576-6813).
[231] M. Bouet, V. Conan, Mobile edge computing resources optimization: A
geo-clustering approach, IEEE Trans. Netw. Serv. Manag. 15 (2) (2018)
787–796.
[232] Y. Li, N.T. Anh, A.S. Nooh, K. Ra, M. Jo, Dynamic mobile cloudlet clustering for
fog computing, in; 2018 International Conference on Electronics, Information,
and Communication, ICEIC, 2018, pp. 1–4.
[233] G. Li, Q. Lin, J. Wu, Y. Zhang, J. Yan, Dynamic computation offloading
based on graph partitioning in mobile edge computing, IEEE Access 7 (2019)
185131–185139.
[234] L. Hu, Y. Tian, J. Yang, T. Taleb, L. Xiang, Y. Hao, Ready player one: UAV-
clustering-based multi-task offloading for vehicular VR/AR gaming, IEEE Netw.
33 (3) (2019) 42–48.
[235] X. Liu, J. Yu, J. Wang, Y. Gao, Resource allocation with edge computing in IoT
networks via machine learning, IEEE Internet Things J. 7 (4) (2020) 3415–3426.
[236] A.A.A. Gad-Elrab, A.Y. Noaman, Fuzzy clustering-based task allocation approach
using bipartite graph in cloud-fog environment, in: Proceedings of the 16th
EAI International Conference on Mobile and Ubiquitous Systems: Computing,
Networking and Services, MobiQuitous ’19, Association for Computing Machin-
ery, New York, NY, USA, 2019, pp. 454–463, https://doi.org/10.1145/3360774.
3360833.
[237] I. Ullah, H.Y. Youn, Task classification and scheduling based on K-means clus-
tering for edge computing, Wirel. Pers. Commun. 113 (4) (2020) 2611–2624,
https://doi.org/10.1007/s11277-020-07343-w.
[238] A. Bozorgchenani, D. Tarchi, G.E. Corazza, An energy-aware offloading clus-
tering approach (eaoca) in fog computing, in: 2017 International Symposium
on Wireless Communication Systems, ISWCS, 2017, pp. 390–395 (ISSN:
2154-0225).
[239] S. Yu, X. Wang, R. Langar, Computation offloading for mobile edge computing:
A deep learning approach, in: 2017 IEEE 28th Annual International Symposium
on Personal, Indoor, and Mobile Radio Communications, PIMRC, 2017, pp. 1–6
(ISSN: 2166-9589).
[240] D.S. Rani, M. Pounambal, Deep learning based dynamic task offloading in
mobile cloudlet environments, Evol. Intell. (ISSN: 1864-5917) (2019) https:
//doi.org/10.1007/s12065-019-00284-9.
[241] L. Huang, X. Feng, A. Feng, Y. Huang, L.P. Qian, Distributed deep learning-
based offloading for mobile edge computing networks, Mob. Netw. Appl. (2018)
https://doi.org/10.1007/s11036-018-1177-x.
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb192
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb192
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb192
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb192
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb192
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb193
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb194
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb195
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb196
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb196
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb196
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb196
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb196
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb197
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb197
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb197
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb197
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb197
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb198
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb198
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb198
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb198
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb198
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb199
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb199
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb199
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb199
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb199
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb200
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb200
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb200
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb201
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb201
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb201
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb201
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb201
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb202
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb202
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb202
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb202
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb202
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb203
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb203
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb203
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb203
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb203
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb205
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb205
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb205
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb205
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb205
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb206
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb206
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb206
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb206
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb206
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb207
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb207
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb207
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb207
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb207
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb208
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb208
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb208
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb209
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb209
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb209
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb210
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb210
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb210
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb210
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb210
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb211
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb211
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb211
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb212
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb212
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb212
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb213
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb213
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb213
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb213
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb213
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb214
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb214
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb214
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb214
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb214
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb215
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb215
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb215
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb216
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb216
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb216
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb216
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb216
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb217
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb219
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb219
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb219
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb219
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb219
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb220
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb220
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb220
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb220
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb220
https://onlinelibrary.wiley.com/doi/pdf/10.1002/ett.3824
https://onlinelibrary.wiley.com/doi/pdf/10.1002/ett.3824
https://onlinelibrary.wiley.com/doi/pdf/10.1002/ett.3824
https://onlinelibrary.wiley.com/doi/abs/10.1002/ett.3824
https://onlinelibrary.wiley.com/doi/abs/10.1002/ett.3824
https://onlinelibrary.wiley.com/doi/abs/10.1002/ett.3824
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb222
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb223
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb223
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb223
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb223
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb223
https://www.hindawi.com/journals/wcmc/2018/2102348/
https://www.hindawi.com/journals/wcmc/2018/2102348/
https://www.hindawi.com/journals/wcmc/2018/2102348/
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb231
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb231
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb231
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb231
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb231
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb233
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb233
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb233
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb233
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb233
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb234
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb234
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb234
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb234
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb234
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb235
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb235
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb235
https://doi.org/10.1145/3360774.3360833
https://doi.org/10.1145/3360774.3360833
https://doi.org/10.1145/3360774.3360833
https://doi.org/10.1007/s11277-020-07343-w
https://doi.org/10.1007/s12065-019-00284-9
https://doi.org/10.1007/s12065-019-00284-9
https://doi.org/10.1007/s12065-019-00284-9
https://doi.org/10.1007/s11036-018-1177-x
Computer Networks 195 (2021) 108177F. Saeik et al.
[242] X. Wang, X. Wei, L. Wang, A deep learning based energy-efficient computational
offloading method in internet of vehicles, China Commun. 16 (3) (2019) 81–91.
[243] Z. Ali, L. Jiao, T. Baker, G. Abbas, Z.H. Abbas, S. Khaf, A deep learning ap-
proach for energy efficient computational offloading in mobile edge computing,
IEEE Access 7 (2019) 149623–149633.
[244] X. Zhu, S. Chen, S. Chen, G. Yang, Energy and delay co-aware computation
offloading with deep learning in fog computing networks, in: 2019 IEEE 38th
International Performance Computing and Communications Conference, IPCCC,
2019, pp. 1–6 (ISSN: 2374-9628).
[245] L. Huang, S. Bi, Y.J. Zhang, Deep reinforcement learning for online computation
offloading in wireless powered mobile-edge computing networks, IEEE Trans.
Mob. Comput. (2019) 1.
[246] R. Zhao, X. Wang, J. Xia, L. Fan, Deep reinforcement learning based mobile
edge computing for intelligent Internet of Things, Phys. Commun. (2020)
101184.
[247] K. Zhang, Y. Zhu, S. Leng, Y. He, S. Maharjan, Y. Zhang, Deep learning
empowered task offloading for mobile edge computing in urban informatics,
IEEE Internet Things J. 6 (5) (2019) 7635–7647.
[248] H. Lu, C. Gu, F. Luo, W. Ding, X. Liu, Optimization of lightweight task
offloading strategy for mobile edge computing based on deep reinforcement
learning, Future Gener. Comput. Syst. 102 (2020) 847–861, http://www.
sciencedirect.com/science/article/pii/S0167739X19308209.
[249] J. Baek, G. Kaddoum, Heterogeneous task offloading and resource allocations
via deep recurrent reinforcement learning in partial observable multi-fog
networks, IEEE Internet Things J. (2020) 1.
[250] J. Liu, X. Wei, T. Wang, J. Wang, An ant colony optimization fuzzy clustering
task scheduling algorithm in mobile edge computing, in: J. Li, Z. Liu, H.
Peng (Eds.), Security and Privacy in New Computing Environments, in: Lecture
Notes of the Institute for Computer Sciences, Social Informatics and Telecom-
munications Engineering, Springer International Publishing, Cham, 2019, pp.
615–624.
[251] M.K. Hussein, M.H. Mousa, Efficient task offloading for IoT-based applica-
tions in fog computing using ant colony optimization, IEEE Access 8 (2020)
37191–37201.
[252] T.G. Rodrigues, K. Suto, H. Nishiyama, N. Kato, A PSO model with VM
migration and transmission power control for low service delay in the
multiple cloudlets ECC scenario, in: 2017 IEEE International Conference on
Communications, ICC, IEEE, 2017, pp. 1–6.
[253] L.N.T. Huynh, Q.-V. Pham, X.-Q. Pham, T.D.T. Nguyen, M.D. Hossain, E.-N.
Huh, Efficient computation offloading in multi-tier multi-access edge computing
systems: A particle swarm optimization approach, Appl. Sci. 10 (1) (2020) 203,
https://www.mdpi.com/2076-3417/10/1/203.
[254] Y. Zhang, Y. Liu, J. Zhou, J. Sun, K. Li, Slow-movement particle swarm
optimization algorithms for scheduling security-critical tasks in resource-limited
mobile edge computing, Future Gener. Comput. Syst. 112 (2020) 148–161,
http://www.sciencedirect.com/science/article/pii/S0167739X19333552.
[255] D. Zhang, F. Haider, M. St-Hilaire, C. Makaya, Model and algorithms for the
planning of fog computing networks, IEEE Internet Things J. 6 (2) (2019)
3873–3884.
[256] A.A. Al-habob, O.A. Dobre, A. Garcia Armada, Sequential task scheduling for
mobile edge computing using genetic algorithm, in: 2019 IEEE Globecom
Workshops, GC Wkshps, 2019, pp. 1–6.
[257] V. Yadav, B.V. Natesha, R.M.R. Guddeti, GA-PSO: Service allocation in fog
computing environment using hybrid bio-inspired algorithm, in: TENCON 2019
– 2019 IEEE Region 10 Conference, TENCON, 2019, pp. 1280–1285 (ISSN:
2159-3450).
[258] Z. Wang, Z. Zhao, G. Min, X. Huang, Q. Ni, R. Wang, User mobility aware task
assignment for mobile edge computing, Future Gener. Comput. Syst. 85 (2018)
1–8, http://www.sciencedirect.com/science/article/pii/S0167739X17318587.
[259] M.B. Kamal, N. Javaid, S.A.A. Naqvi, H. Butt, T. Saif, M.D. Kamal, Heuristic
min-conflicts optimizing technique for load balancing on fog computing, in:
Advances in Intelligent Networking and Collaborative Systems, 2019, pp.
207–219.
[260] G. Amarasinghe, M.D. de Assuņ c ao, A. Harwood, S. Karunasekera, A data
stream processing optimisation framework for edge computing applications, in:
2018 IEEE 21st International Symposium on Real-Time Distributed Computing,
ISORC, 2018, pp. 91–98 (ISSN: 2375-5261).
[261] F. Ait Salaht, F. Desprez, A. Lebre, C. Prud’homme, M. Abderrahim, Service
placement in fog computing using constraint programming, in: 2019 IEEE
International Conference on Services Computing, SCC, 2019, pp. 19–27 (ISSN:
2474-2473).
[262] S. Vakilinia, D. Qiu, M.M. Ali, Optimal multi-dimensional dynamic resource al-
location in mobile cloud computing, EURASIP J. Wireless Commun. Networking
2014 (1) (2014) 201, https://doi.org/10.1186/1687-1499-2014-201.
[263] M.C. Calzarossa, L. Massari, D. Tessera, Workload characterization: A survey
revisited, ACM Comput. Surv. 48 (3) (2016) 48:1–48:43, https://doi.org/10.
1145/2856127.
[264] I. Bouras, F. Aisopos, J. Violos, G. Kousiouris, A. Psychas, T. Varvarigou, G.
Xydas, D. Charilas, Y. Stavroulas, Mapping of quality of service requirements
25
to resource demands for iaas, in: CLOSER, 2019.
[265] C. Lecoutre, Constraint Networks: Targeting Simplicity for Techniques and
Algorithms, John Wiley & Sons, 2013.
[266] F. Rossi, P. Van Beek, T. Walsh, Handbook of Constraint Programming, Elsevier,
2006.
[267] X. Yin, A. Jindal, V. Sekar, B. Sinopoli, A control-theoretic approach for
dynamic adaptive video streaming over HTTP, in: Proceedings of the 2015
ACM Conference on Special Interest Group on Data Communication, 2015, pp.
325–338.
[268] D.Y. Zhang, D. Wang, An integrated top-down and bottom-up task allocation
approach in social sensing based edge computing systems, in: IEEE INFOCOM
2019-IEEE Conference on Computer Communications, IEEE, 2019, pp. 766–774.
[269] C. Wu, Y. Zhang, Y. Deng, Toward fast and distributed computation migration
system for edge computing in IoT, IEEE Internet Things J. 6 (6) (2019)
10041–10052.
[270] Y. Li, Edge Computing-Based Access Network Selection for Heterogeneous
Wireless Networks (Ph.D. dissertation), Universite Rennes, 2017.
[271] A. Koike, N. Higo, Y. Sueda, Proxy-based network function to assist robotic
feedback control system, in: 2018 IEEE International Symposium on Local and
Metropolitan Area Networks, LANMAN, IEEE, 2018, pp. 116–118.
[272] S. Host, W. Tärneberg, P. Ödling, M. Kihl, M. Savi, M. Tornatore, Network
requirements for latency-critical services in a full cloud deployment, in: 2016
24th International Conference on Software, Telecommunications and Computer
Networks, SoftCOM, IEEE, 2016, pp. 1–5.
[273] M. Maheswaran, T. Yang, S. Memon, A fog computing framework for au-
tonomous driving assist: architecture, experiments, and challenges, 2019, arXiv
preprint arXiv:1907.09454.
[274] N. Kalatzis, M. Avgeris, D. Dechouniotis, K. Papadakis-Vlachopapadopoulos,
I. Roussaki, S. Papavassiliou, Edge computing in IoT ecosystems for UAV-
enabled early fire detection, in: 2018 IEEE International Conference on Smart
Computing, SMARTCOMP, IEEE, 2018, pp. 106–114.
[275] D. Spatharakis, M. Avgeris, N. Athanasopoulos, D. Dechouniotis, S. Papavassil-
iou, A switching offloading mechanism for path planning and localization in
robotic applications, in: 2020 International Conferences on Internet of Things
(IThings) and IEEE Green Computing and Communications (GreenCom) and
IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data
(SmartData) and IEEE Congress on Cybermatics (Cybermatics), IEEE, 2020, pp.
77–84.
[276] M. Avgeris, D. Dechouniotis, N. Athanasopoulos, S. Papavassiliou, Adaptive
resource allocation for computation offloading: A control-theoretic approach,
ACM Trans. Internet Technol. (TOIT) 19 (2) (2019) 1–20.
[277] T. Dlamini, Á.F. Gambín, D. Munaretto, M. Rossi, Online resource management
in energy harvesting BS sites through prediction and soft-scaling of computing
resources, in: 2018 IEEE 29th Annual International Symposium on Personal,
Indoor and Mobile Radio Communications, PIMRC, IEEE, 2018, pp. 1820–1826.
[278] L. Zhang, S. Wang, R.N. Chang, QCSS: a QoE-aware control plane for adaptive
streaming service over mobile edge computing infrastructures, in: 2018 IEEE
International Conference on Web Services, ICWS, IEEE, 2018, pp. 139–146.
[279] F. Basic, A. Aral, I. Brandic, Fuzzy handoff control in edge offloading, in: 2019
IEEE International Conference on Fog Computing, ICFC, IEEE, 2019, pp. 87–96.
[280] P. Skarin, J. Eker, M. Kihl, K. Årzén, Cloud-assisted model predictive control,
in: 2019 IEEE International Conference on Edge Computing, EDGE, 2019, pp.
110–112.
[281] R.E. Kalman, et al., Contributions to the theory of optimal control, Bol. Soc.
Mat. Mexicana 5 (2) (1960) 102–119.
[282] Z. Zhou, J. Yu, X. Dong, Q. Li, Z. Ren, Robust adaptive attitude control of the
quad-rotor UAV based on the LQR and NESO technique, in: 2018 IEEE 14th
International Conference on Control and Automation, ICCA, IEEE, 2018, pp.
745–750.
[283] A. Ullah, J. Li, Y. Shen, A. Hussain, A control theoretical view of cloud
elasticity: taxonomy, survey and challenges, Cluster Comput. 21 (4) (2018)
1735–1764.
[284] D. Dechouniotis, N. Leontiou, N. Athanasopoulos, G. Bitsoris, S. Denazis, ACRA:
A unified admission control and resource allocation framework for virtualized
environments, in: 2012 8th International Conference on Network and Service
Management (Cnsm) and 2012 Workshop on Systems Virtualiztion Management
(Svm), IEEE, 2012, pp. 145–149.
[285] P.S. Saikrishna, R. Pasumarthy, N.P. Bhatt, Identification and multivariable
gain-scheduling control for cloud computing systems, IEEE Trans. Control Syst.
Technol. 25 (3) (2016) 792–807.
[286] D. Dechouniotis, N. Leontiou, N. Athanasopoulos, A. Christakidis, S. Denazis,
A control-theoretic approach towards joint admission control and resource
allocation of cloud computing services, Int. J. Netw. Manag. 25 (3) (2015)
159–180.
[287] S. Rashidi, S. Sharifian, Cloudlet dynamic server selection policy for mobile
task off-loading in mobile cloud computing using soft computing techniques, J.
Supercomput. 73 (9) (2017) 3796–3820.
[288] N. Leontiou, D. Dechouniotis, S. Denazis, S. Papavassiliou, A hierarchical
control framework of load balancing and resource allocation of cloud computing
services, Comput. Electr. Eng. 67 (2018) 235–251.
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb242
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb242
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb242
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb243
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb243
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb243
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb243
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb243
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb245
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb245
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb245
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb245
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb245
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb246
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb246
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb246
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb246
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb246
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb247
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb247
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb247
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb247
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb247
http://www.sciencedirect.com/science/article/pii/S0167739X19308209
http://www.sciencedirect.com/science/article/pii/S0167739X19308209
http://www.sciencedirect.com/science/article/pii/S0167739X19308209
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb249
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb249
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb249
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb249
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb249
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb250
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb251
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb251
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb251
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb251
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb251
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb252
https://www.mdpi.com/2076-3417/10/1/203
http://www.sciencedirect.com/science/article/pii/S0167739X19333552
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb255
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb255
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb255
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb255
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb255
http://www.sciencedirect.com/science/article/pii/S0167739X17318587
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb259
https://doi.org/10.1186/1687-1499-2014-201
https://doi.org/10.1145/2856127
https://doi.org/10.1145/2856127
https://doi.org/10.1145/2856127
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb264
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb264
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb264
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb264
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb264
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb265
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb265
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb265
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb266
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb266
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb266
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb268
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb268
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb268
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb268
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb268
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb269
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb269
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb269
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb269
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb269
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb270
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb270
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb270
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb271
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb271
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb271
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb271
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb271
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb272
http://arxiv.org/abs/1907.09454
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb274
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb275
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb276
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb276
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb276
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb276
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb276
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb277
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb278
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb278
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb278
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb278
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb278
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb279
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb279
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb279
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb281
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb281
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb281
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb282
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb283
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb283
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb283
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb283
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb283
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb284
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb285
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb285
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb285
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb285
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb285
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb286
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb287
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb287
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb287
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb287
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb287
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb288
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb288
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb288
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb288
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb288
Computer Networks 195 (2021) 108177F. Saeik et al.
[289] D. Dechouniotis, N. Athanasopoulos, A. Leivadeas, N. Mitton, R. Jungers, S.
Papavassiliou, Edge computing resource allocation for dynamic networks: The
DRUID-net vision and perspective, Sensors 20 (8) (2020) 2191.
[290] W. Zhang, M.S. Branicky, S.M. Phillips, Stability of networked control systems,
IEEE Control Syst. Mag. 21 (1) (2001) 84–99.
[291] J.P. Hespanha, P. Naghshtabrizi, Y. Xu, A survey of recent results in networked
control systems, Proc. IEEE 95 (1) (2007) 138–162.
[292] B. Sinopoli, L. Schenato, M. Franceschetti, K. Poolla, M.I. Jordan, S.S. Sastry,
Kalman filtering with intermittent observations, IEEE Trans. Autom. Control 49
(9) (2004) 1453–1464.
[293] D. Simon, D. Robert, O. Sename, Robust control/scheduling co-design: applica-
tion to robot control, in: 11th IEEE Real Time and Embedded Technology and
Applications Symposium, IEEE, 2005, pp. 118–127.
[294] W. Heemels, K.H. Johansson, P. Tabuada, An introduction to event-triggered
and self-triggered control, in: 2012 IEEE 51st IEEE Conference on Decision and
Control, CDC, IEEE, 2012, pp. 3270–3285.
Firdose Saeik is currently pursuing a Ph.D. at ETS. Canada.
He received the M.Sc. degree from VIT university, India, in
2010. He has more than 9 years of work experience with
multiple roles as System Engineer, Junior Researcher, and
Research Engineer. His research interests are in the field
of Mobile Edge Computing, Quality of Experience, Next-
generation mobile services and applications such as Virtual
Reality (VR) and Augmented Reality (AR).
Marios Avgeris is currently a Ph.D student in the NET-
MODE Lab at the National Technical University of Athens
(NTUA). He received his Diploma in Electrical & Com-
puter Engineering (ECE) from NTUA, Greece, in 2016.
His research interests are control theory, edge and cloud
computing, IoT, semantic web technologies and network
monitoring.
Dimitrios Spatharakis is currently a Ph.D. student in the
NETMODE Lab at the National Technical University of
Athens (NTUA). He received a Diploma in Electrical &
Computer Engineering (ECE) from NTUA, Greece, in 2018.
His research interests focus on IoT, cyber–physical systems,
edge computing and cloud computing
Nina Santi is a Ph.D. student under the supervision of
Nathalie Mitton in the Inria FUN team. Their focus is on
small computing devices like electronic tags and sensor
networks. She has received the M.Sc. degrees in Computer
Science from University of Lille, France, in 2020.
Dimitrios Dechouniotis is currently research associate with
NETMODE Lab of the National Technical University of
Athens (NTUA). From 2007 to 2016, he was non-tenured
Lecturer at the EE Dept. of Technical Educational Institute
of Western Greece, Greece. He received his diploma in ECE
from University of Patras in 2004, the M.Sc. degree in
Control Systems and Robotics from NTUA in 2009, and the
Ph.D. degree in ECE from University of Patras in 2014.
His research interests lie in the area of cloud computing,
Internet of Things, mobile cloud computing and control
theory.
26
John Violos is research associate in the Dept. of Soft-
ware Engineering and Information Technology at ETS. His
previous positions were research associate at National Tech-
nical University of Athens, sessional lecturer at Harokopio
University of Athens and visiting lecturer at National and
Kapodistrian University of Athens. He was a member in
the European Commission’s Digital Single Market working
group on the code of conduct for switching and porting
data between cloud service providers. His research interests
include Deep Learning, Machine Learning, Cloud and Edge
computing.
Aris Leivadeas is currently an Assistant Professor with the
Dept. of Software and Information Technology Engineering
at ETS. From 2015 to 2018 he was a postdoc in the Dept.
of SCE, at Carleton University. In parallel, Aris worked as
an intern at Ericsson and then at Cisco in Ottawa, Canada.
He received his diploma in ECE from University of Patras in
2008, the M.Sc. degree in Engineering from King’s College
London in 2009, and the Ph.D degree in ECE from NTUA in
2015. His research interests include Cloud Computing, IoT,
and network optimization and management. He received
the best paper award in ICPE’18 and the best presentation
award in HPSR’20.
Nikolaos Athanasopoulos is a Lecturer at the School of
Electronics, Electrical Engineering and Computer Science
at Queens University Belfast. He received a Diploma and
a Ph.D. in Electrical and Computer Engineering from the
University of Patras, Greece and has held postdoctoral
researcher positions at TU/e and UCLouvain. He has been
an IKY and a Marie Curie Fellow. His interests are in control
theory, focusing on hybrid systems and set-based methods
with applications in edge/cloud computing and resource
allocation.
Nathalie Mitton received the M.Sc. and PhD. degrees in
Computer Science from INSA Lyon in 2003 and 2006
respectively. She received her Habilitation à diriger des
recherches (HDR) in 2011 from Université Lille 1. She
is currently an Inria full researcher since 2006 and from
2012, she is the scientific head of the Inria FUN team
which is focused on small computing devices like elec-
tronic tags and sensor networks. Her research interests
focus on self-organization from PHY to routing for wireless
constrained networks. She has published her research in
more than 30 international revues and more than 100
international conferences. She is involved in the setup
of the FIT IoT LAB platform (http://fit-equipex.fr/, https:
//www.iot-lab.info), the H2020 CyberSANE and VESSE-
DIA projects and in several program and organization
committees such as Infocom 2021&2020&2019, PerCom
2020&2019, DCOSS 2021&2020&2019, Adhocnow (since
2015), ICC (since 2015), Globecom (since 2017), VTC (since
2016), etc. She also supervises several Ph.D. students and
engineers.
Symeon Papavassiliou is currently a professor in the
School of Electrical and Computer Engineering at the Na-
tional Technical University of Athens (NTUA). From 1995
to 1999, he was a senior technical staff member at AT&T
Laboratories, New Jersey. In August 1999 he joined the ECE
Dept at the New Jersey Institute of Technology, USA, where
he was an associate professor until 2004. He has an estab-
lished record of publications in his field of expertise, with
more than 350 technical journal and conference published
papers, while he has received several scientific awards and
distinctions. His main research interests lie in the areas
of optimization and performance evaluation of mobile and
distributed systems, wireless networks and complex systems.
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb289
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb289
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb289
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb289
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb289
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb290
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb290
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb290
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb291
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb291
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb291
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb292
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb292
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb292
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb292
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb292
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb293
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb293
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb293
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb293
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb293
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb294
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb294
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb294
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb294
http://refhub.elsevier.com/S1389-1286(21)00232-2/sb294
http://fit-equipex.fr/
https://www.iot-lab.info
https://www.iot-lab.info
Task offloading in Edge and Cloud Computing: A survey on mathematical, artificial intelligence and control theory solutions
Introduction
Computing paradigms: Overview & use cases
Modern computing paradigms
Cloud computing
Mobile cloud computing
Fog computing
Edge computing
Computing paradigms comparison
Use cases
Immersive applications
Autonomous vehicles
Robotics
Video streaming
IoT
Physical disaster management
Task offloading & challenges
Granularity levels of task offloading
Partial offloading at the edge
Full offloading at the edge
Partial/full offloading at the edge and cloud
Mobility of end devices
Static (low-mobility)
Mobile (high-mobility)
Task offloading objectives
Delay
Energy
Bandwidth/spectrum
Load balancing
Deployment cost
Model accuracy
Multi-objective
Challenges of task offloading
Network dynamics challenges
Resource allocation challenges
A taxonomy of task offloading approaches
Methodology comparison
Mathematical optimization algorithms
Mixed integer programming
Heuristics
Game theory
Contract theory
Local search
AI-based optimization algorithms
Machine learning
Population-based methods
Constraint satisfaction methods
Control theory-based algorithms
Optimal control
State feedback control
Lyapunov Optimization approaches
Rest of the control approaches
Open challenges and future directions
Heterogeneous networks
Real-time distributed resource allocation
Mobility-induced network dynamics
Node, resource and application heterogeneity
Moving edge resources closer to the end devices
Security and privacy
Fault tolerance
Control-related challenges
Controller design for cyber–physical systems (CPS)
Conclusion
CRediT authorship contribution statement
Declaration of competing interest
Acknowledgment
References