DR.

BABASAHEB AMBEDKAR
TECHNOLOGICALUNIVERSITY,LONERE

P.S.G.V.P.MANDAL’S
D. N. PATEL COLLEGE OF ENGINEERING
SHAHADA,DIST-NANDURBAR

SEMINAR
ON

“EMPOWERED EDGE”

SUBMITTEDBY

Mrs. ROSHANI SHARAD PATIL

(1951641293015)

UNDERTHEGUIDANCEOF
PROF.V.S.KALE
DEPARTMENT OF ELECTRICAL ENGINEERING
YEAR 2022-23
 P. S.G.V.P.MANDAL’S
D.N. PATEL COLLEGE OF ENGINEERING,
SHAHADA, DIST. NANDURBAR

This is to certify that

Mrs. ROSHANI SHARAD PATIL

(1951641293015)
Have been satisfactorily completed
Seminar entitled on
“EMPOWERED EDGE”

As a part of syllabus of

DR. BABASAHEB AMBEDKAR TECHNOLOGICAL

UNIVERSITY,LONERE

For the partial fulfillment of Bachelor of Technology in

ELECTRICAL ENGINEERING
In the Academic Year
2022-23

SEMINAR GUIDE H.O.D.

PROF.V.S.KALE PROF. Smt. K. A. PATEL

PRINCIPAL
EXTERNAL PROF. Dr. N. J. PATIL
 ACKNOWLEDGEMENT

It is a privilege for us to have been associated with Prof. V.S.KALE sir,our guide, during this
project work. We are thankful to him, for his constant inspiration and valuable guidance,
carefully reading and editing our work and always boosting our confidence to complete work.

We express our gratitude to Prof. Dr. N. J. Patil , Principal and Prof. Smt. K. A. Patel, Head
Department of Electrical Engineering, for their constant encouragement, co-operation ,valuable
guidance, and support. We express our sincere thanks to our academic and class coordinators
Prof. Pankaj R. PATIL and all the faculty members of the Electrical Department for their
unfailing inspiration.

We take this opportunity to thank all our classmates for their company during the course

work and for the useful discussions, we had with them.

We would be failing in our duties if we do not make a mention of our family members

including our parents and our siblings for providing moral support, without which this work

would not have been completed.

This kind of work cannot be finished without any others help, even some of them have

not aware of their contribution and importance in producing this report. It is a great pleasure

for us to take this opportunity to express our gratefulness to all.

NAME: - PATIL ROSHANI SHARAD

PRN NO:-1951641293015

ROLL ON:-13
 Empowered Edge

TABLE OF CONTENTS

Chapter Content Page

No. No.
1. INTRODUCTION 1-2

2. LITERATURE SURVEY 3-5

2.1 History 3
2.2 Basic Concepts 4
3 CONCEPT 6-11
3.1 Architecture 6-8
3.2 Working 8-10
3.3 Technologies 10-11
4 Functional and Technical Details 12
4.1 Features 12

4.2 Digital development 12

APPLICATIONS

5.1Offloading decision

5.2Ddpg-empowred vehicular edge computing and

5 Caching schme 12-20

5.3 Research challenge and open issues

5.4Edge lntelll Gence

5.5Edge lntellgence for 5g byond llot

6 ADVANTAGES AND DISADVANTAGES 21-22

6.1 Advantages
6.2 Disadvantages
7 CONCLUSION AND FUTURE SCOPE 23-24
7.1 Conclusion
7.2 Future Scope
REFERENCES 25
Empowered Edge
 Empowered Edge
c

LIST OF FIGURES

Sr. Page
Figure Name
No. No.

1 AI-empowered vehicular network architecture

 ABSTRACT

Recent advances in edge computing and caching have significant impacts on the
developments of vehicular net-works. Nevertheless, the heterogeneous requirements of on-
vehicle applications and the time-variability on popularity of contents, bring great
challenges for edge servers to efficiently utilize their resources. Moreover, the high mobility
of smart vehicles adds substantial complexity in jointly optimizing edge computing and
caching. Artificial Intelligence (AI) can greatly enhance the cognition and intelligence of
vehicular networks and thus assist in optimally allocating resources for the problems with
diverse, time-variant, and complex features. In this article, we propose a new architecture
that can dynamically orchestrate edge computing and caching resources to improve system
utility by making full use of AI based algorithms. Then we formulate a joint edge computing
and caching scheme to maximize system utility and develop a novel resource scheme by
exploiting deep reinforcement learning. Numerical results demonstrate the effectiveness of
the proposed scheme.
 Empowered Edge

INTRODUCTION chapter -1
With the rapid development of Internet of things (IoT) technology and the increasing number of vehicle
networks, the traditional vehicle ad hoc network (VANET) is gradually being integrated into the
Internet of vehicles (IoV). IoV is a new model that combines VANETs and vehicle remote information
processing to connect vehicles, people and things.In addition, it is a highly important field in intelligent
transportation systems (ITSs), as it covers intelligent transportation, cloud computing, vehicle
information services, logistics transportation services modern wireless technology, Internet access and
communication and other technologies The associate editor coordinating the review of this manuscript
and approving it for publication was Chao Chen. And applications. According to the forecast report from
Cisco, the global monthly mobile data usage in 2021 will be approximately 49 ex a bytes, and the
number of mobile devices will be 11.6 billion, increasing about approximately seven times between
2016 and 2021. With the explosion of mobile data, mobile phones are increasingly being used for
various computation-intensive applications, such as augmented reality; natural language processing;
face, hand gestures, and object recognition; and various forms of user configurations used for
recommendation; hence, mobile users enjoy a rich experience in the service network. Faced with the
surge of mobile data flow, reducing the delay of data transmission between vehicles and improving the
through- put of data transmission between vehicles are urgent problems .

communication technology and data acquisition technology to improve the safety and efficiency of the
traffic system, reduce accidents and reduce traffic congestion. Generally speaking, public
communication interfaces are divided into wireless networks (such as bluetooth and wi-fi) and cellular
networks (such as 3G, 4G and 5G). However, the limited network bandwidth in traditional cellular
networks limits the fast growth of the data transmission rate. In the emerging 5G network, the
application of D2D (device to device) communication technology promises to substantially improve the
spectrum efficiency to support data transmission between caching vehicles and mobile users. The
federal communications commission (FCC) authorized the 75-mhz band for the provision of vehicle-to-
vehicle wireless communications as dedicated short-range communications (DSRC). In addition, IEEE
standardizes the entire communication stack according to IEEE 802.11p as a wireless access to the
vehicle environment (WAVE) to support the interconnection between vehicles and between vehicles
and roads.In addition, various communication modes coexist in IoV, which include vehicle-to-vehicle
(V2V), vehicle-to-infrastructure (V2I), vehicle-to-sensor (V2S), vehicle-to-pedestrian (V2P), and vehicle-
to-network (V2N) communications, which form a dynamic mobile communication system. Fig. 1
illustrates the architecture of IoV. This enables the sharing and collection of information about vehicles,
roads and their surroundings. While the development of communication technology can alleviate a
certain amount of traffic congestion, the limited ability of the infrastructure to communicate, compute,
and store resources can lead to long delays and massive data transmission problems. In order to
overcome this challenge, combined with the deployment.resources on the edge of the wireless network,
the proposed

1
 Empowered Edge

vehicle edge network has attracted wide attention. Mobile edge computing (MEC) technology can over-
come the challenges of traditional mobile cloud computing (MCC). For example, centralized cloud
servers are located far away from the terminal devices, thereby resulting in low efficiency in
computation-intensive environments; the offloading of computing to the cloud consumes energy,
thereby reducing the service life of mobile batteries; and providing mobile users with complex
memoryutilization applications and higher data storage capacity is difficult . Reference studied the
multi-user computing offloading problem of mobile edge cloud computing under multi-channel wireless
interference, and put forward a distributed computing offloading algorithm by using the game theory
method. In addition, MEC can provide substantial value to mobile operators, service providers and end
users. The application scenarios of MEC span multiple fields, such as augmented reality, online games,
big data analysis and health monitoring in the medical Internet of things .With the emergence of IoV and
vehicle intelligence, vehicles are transforming from transport tools to intelligent terminals . In addition,
the variety and quantity of on-boarde quipment are increasing, and people’s requirements for
automobile service quality are constantly increasing. In the age of IoV, vehicle-mounted intelligent
modules can provide intelligent vehicle control, traffic management, accident prevention and navigation
capabilities, along with rich multimedia and mobile Internet application services and many emerging
interactive applications that improve the user experience, reduce operating costs and promote a safe
driving environment. Artificial intelligence (AI) can substantially improve the cognitive performance
and intelligence of vehicle networks, thereby contributing to the optimal allocation of resources for
problems with diverse, time-varying and complex characteristics [18]. Reinforcement learning (RL) is
an important branch of machine learning. It refers to the process of realizing objectives via multiple
steps and suitable decisions in a series of scenarios, which can be regarded as a multi-step sequential
decision problem . To overcome the problem of decentralized management of connected vehicles in a
distributed intelligent transportation system, reference designed an ant colony optimization algorithm
that is based on swarm intelligence (SI) for dynamic decision making of networked vehicles, which
enables vehicles to automatically and adaptively identify the best path to the destination. In, an
intelligent resource management strategy for joint communication mode selection, resource block
allocation and power control in D2D-V2V communication vehicle networks is proposed. The model-free
participant critical learning framework is used to effectively improve the learning efficiency and
identify the optimal strategy to ensure that the vehicle-to-vehicle link satisfies the communication
requirements of ultra-reliability and low latency while maximizing the overall network capacity. The
main contributions of this article are as follows: We introduce the architecture of IoV, the deployment
of V2X in vehicle-mounted communication, and the application of MEC and AI in IoV. We describe the
advantages and development history of MEC and the relationship between AI and DRL, and Wean alyze
the previous research on the application of AI to

vehicle edge networks. We study the architecture of MEC-based IoV and discuss the use of MEC in IoV.
In addition, the characteristics of MEC, FC and MCC are analyzed, and the key technologies for
supporting MEC are introduced. In addition, the previous studies on efficient MEC calculation for IoV
are analyzed. We consider the theoretical characteristics of AI; analyze DRL, which is a key method for
realizing AI, and demonstrate the architecture of AI in IoV. In addition, We introduce the effective key AI
algorithms for calculating the offload and resource allocation in IoV, and We analyze the previous AI
research on IoV. We combine the application of AI and MEC technology in IoV and analyez the key
technologies that support the application of AI in vehicle edge networks. In addition, the previous
studies on edge caches and on joint computing resources and caches are introduced. Finally, the future
development directions and research challenges of IoV are discussed. This article reviews the
architecture, implementation technology and application of IoV that is based on AI and MEC. We explore

1
 Empowered Edge

the characteristics of IoV development, the communication mode, and the impact of combining AI and
MEC technology on the construction of intelligent IoV. This article is divided into the following parts:
Section II describes the architecture of MEC in IoV and introduces the development history of edge
computing and the characteristics of MEC. In addition, the research on computing offloading of MEC in
IoV is analyzed. Section III mainly studies the application of AI InIoV, expounds on the architecture in
which AI is combined with IoV, and discusses DRL, which is an important technique for realizing AI. The
key algorithms and applications of AI in IoV are analyzed. In Section IV, the significance of the
combination of AI and MEC technology in IoV is discussed, the key technologies in AI-based vehicle edge
networks are studied, and relevant studies on IoV are analyzed. In Section V, the challenges that are
faced by IoV and the future development directions are discussed. Finally Section VI summarizes the
study.

2
 Empowered Edge

2. LITERATURE REVIEW CHAPTER-2

2.1 History
The beginnings of decentralized computing

The origin of edge computing can be traced back to the 1990s, when Akamai launched its
content delivery network (CDN). The idea back then was to introduce nodes at locations
geographically closer to the end user for the delivery of cached content such as images and
videos.

In 1997, in their work "Agile application

aware adaptation for mobility," Nobel et al. demonstrated how different types of applications
(web browsers, video, and speech recognition) running on resource constrained mobile
devices can offload certain tasks to powerful servers (surrogates). The goal was to relieve the
load on the computing resources. And, as proposed in a later work, to improve the battery life
of mobile devices. Today, for example, speech-recognition services from Google, Apple, and
Amazon work in a similar way. In 2001, with reference to pervasive
computing,Satyanarayanan et al. generalized this approach in their paper "Pervasive
computing: vision and challenges."

In 2001 scalable and decentralized

distributed applications used, as proposed, different peer-to-peer (so-called distributed hash

tables) overlay networks. These self-organizing overlay networks enable efficient and fault-
tolerant routing, object location, and load balancing. Moreover, these systems also make it
possible to exploit the network proximity of underlying physical connections in the internet,
thereby avoiding long-distance links between peers. This not only decreases the overall
network load but also improves the latency of applications.

Edge computing is gaining more and more popularity in the loT domain. In 2018, it was one of
the top technology trends forming the foundation for the next generation of digital businesses.
In parallel, given the massive amounts of data and the need to optimize computational
resources, we are also seeing an increasing tendency to send data to the cloud.

3
 Empowered Edge

2.2 Basic Concepts

The world's data is expected to grow 61% to 175 zettabytes by 2025.[13] The increase of loT devices at
the edge of the network is producing a massive amount of data - storing and using all that data in cloud
data centers pushes network bandwidth requirements to the limit.[14] Despite the improvements of
network technology, data centers cannot guarantee acceptable transfer rates and response times,
which, however, often is a critical requirement for many applications. [15] Furthermore, devices at the
edge constantly consume data coming from the cloud, forcing companies to decentralize data storage
and service provisioning, leveraging physical proximity to the end user.

In a similar way, the aim of edge computing is to move the computation away from data centers towards
the edge of the network, exploiting smart objects, mobile phones, or network gateways to perform tasks
and provide services on behalf of the cloud.[16] By moving services to the edge, it is possible to provide
content caching, service delivery, persistent data storage, and loT management resulting in better
response times and transfer rates. At the same time, distributing the logic to different network nodes
introduces new issues and challenges.

2.2.1 Privacy and security

The distributed nature of this paradigm introduces a shift in security schemes used in cloud computing.
In edge computing, data may travel between different distributed nodes connected through the Internet
and thus requires special encryption mechanisms. independent of the cloud. Edge nodes may also be
resource-constrained devices, limiting the choice in terms of security methods. Moreover, a shift from
centralized top-down infrastructure to a decentralized trust model is required.17) On the other hand,
by. keeping and processing data at the edge, it is possible to increase privacy by minimizing the
transmission of sensitive information to the cloud. Furthermore, the ownership of collected data shifts
from service providers to end-users.[18]

2.2.2 Scalabilit

Scalability in a distributed network must face different issues. First, it must take into account the
heterogeneity of the devices, having different performance and energy constraints, the highly dynamic
condition, and the reliability of the connections compared to more robust infrastructure of cloud data
centers. Moreover, security requirements may introduce further latency in the communication between
nodes, which may slow down the scaling process. [15]

2.2.3 Reliability

Management of failovers is crucial in order to keep a service alive. If a single node goes down and is
unreachable, users should still be able to access a service without interruptions. Moreover, edge
computing systems must provide actions to recover from a failure and alerting the user about the
incident. To this aim, each device must maintain the network topology of the entire distributed system,

4
 Empowered Edge

so that detection of errors and recovery become easily applicable. Other factors that may influence this
aspect are the connection technologies in use, which may provide different levels of reliability, and the
accuracy of the data produced at the edge that could be unreliable due to particular environment
conditions. [15] As an example an edge computing device, such as a voice assistant may continue to
provide service to local users even during cloud service or internet outages. [18]

2.2.4 Speed

Edge computing brings analytical computational resources close to the end users and therefore can
increase the responsiveness and throughput of applications. A well-designed edge platform would
significantly outperform a traditional cloud-based system. Some applications rely on short response
times, making edge computing a significantly more feasible option than cloud computing. Examples
range from loT to autonomous driving,[19] anything health or human/ public safety relevant, 201 or
involving human perception such as facial recognition, which typically takes a human between 370-
620ms to perform.[211 Edge computing is more likely to be able to mimic the same perception speed as
humans, which is useful in applications such as augmented reality where the headset should preferably
recognize who a person is at the same time as the wearer does.

5
 Empowered Edge

3. CONCEPT Chapter-3
3.1 Architecture
The variability in terms of computing and caching capabilities, the diverse requirements of
different vehicular applications, time-varying content popularity, and the randomness of
duration that a special vehicle stays in the coverage of an RSU or a BS, makes it challenging for
vehicular networks to offer such flexibility while also satisfying diverse performance and
specific requirements. To address this, we propose a novel AI-empowered vehicular network
architecture with cognitive and intelligent abilities for smart vehicular edge computing and
caching, as shown in Fig. 1.In the proposed architecture, RSUs are located along a road and act
as edge servers to provide communication, computing, and caching capabilities. Unlike existing
RSUs, these RSUs are deployed with AI functions and can provide diverse resources for smart
vehicles to implement attractive applications, such as navigation, video streaming, and smart
traffic lights. Smart vehicles are key computation task generators and content requesters
moving on the road with a certain speed and direction. The computation tasks and requested
content of smart vehicles often need to be processed and obtained in time. RSUs and smart
vehicles are located within the coverage area of an existing BS. Due to a larger coverage, the BS
is usually deployed stronger computing and caching resources than RSUs. Thus, when RSUs
cannot support computation requirements of smart vehicles, such computation intensive tasks
can be offloaded to the BS. Because of relatively high caching capability, the BS often caches
large-size and less popular contents while RSUs often cache small-size and more important
contents, such as the latest news and emergencyTo exploit AI, we regard the vehicular network
as environment and deploy an intelligent agent on the BS. The information of vehicular
environment, such as real-time behavior/demand of vehicles and wireless channel condition,
can be cognized based on AI methods. The agent can interact with the environment through
state, action, and reward. For meeting the requirements of smart vehicles, the agent
automatically designs sophisticated actions based on current state, which include cross-layer
offloading, cooperative multipoint caching and delivery, and V2V edge caching. . The variability
in terms of computing and caching capabiities , the diverse requirements of different vehicular
applications , time-varying content popularity, and the randomness of duration that a special
vehicle stays in the coverage of an RSU or a BS, makes it challenging for vehicular networks to
offersuch flexibility while also satisfying diverse performance and specific requirements. To
address this, we propose a novel AI-empowered vehicular network architecture with cognitive
and intelligent abilities for smart vehicular edge computing and caching, as shown in Fig. 1.In
the proposed architecture, RSUs are located along a r oadand act as edge servers to provide
communication, computing, and caching capabilities. Unlike existing RSUs, these RSU are
deployed with AI functions and can provide diversere sources for smart vehicles to implement
attractive applications , such as navigation, video streaming, and smart traffic lights. Smart
vehicles are key computation task generators and content requesters moving on the road with
a certain speed and direction. The computation tasks and requested content of smart vehicles
often need to be processed and obtained in time. RSUs and smart vehicles are located within
the coverage area of an existing BS. Due to a larger coverage, the BS is usually deployed

6
 Empowered Edge

stronger computing and caching resources than RSUs. Thus, when RSUs cannot support
computation requirements of smart vehicles, such computation in tensivetasks can be of
floaded to the BS. Because of relatively high caching capability, the BS often caches large-size
and less popular contents while RSUs often cache small-size and more important contents, such
as the latest news and emergencywarning.To exploit AI, we regard the vehicular network as
environment and deploy an intelligent agent on the BS. fo rmation of vehicular environment,
such as real-time be-havior /demand of vehicles and wireless channel condition,can be
cognized based on AI methods [12]. The agent caninteract with the environment through state,
action, and reward .For meeting the requirements of smart vehicles, the agentautomatically
designs sophisticated actions based on current state, which include cross-layer of floading,
cooperative multi-point caching and delivery, and V2V edge caching.

1] Cross-layer Of floading: To provide sufficient computation resources for vehicles, in this

architecture, we combine heterogeneous network with MEC to establish a two-tier cross-layer
of floading. Heterogeneous network brings small cel lbase stations, such as access points and
RSUs, into cellular networks to form a two-tier cross-layer infrastructure. MEC introduces
computing capability to small cell base stations. The two-tier cross-layer offloading is able to
process the t asks that demand ultra-low latency and ultra-high computation resources. If the
task can be arbitrarily divided into severalparts, the two-tier cross-layer offloading will
support partial offloading and parallel computing [2]. Specifically , as show nin Fig. 1, a red
smart vehicle generates a computation-intensivetask with a stringent deadline and the vehicle
determines tooffload it to the nearest RSU for computing. Since the available computation
resources of the RSU is not sufficient to process the task, the RSU further of floads the task to
the nearest BS forcolla boratively computing. That is, the computation-intensivetask is divided
into three parts: the first part is processe dlocally, the second part is processed by the RSU, and
the third part is processed by the BS. Since the backhaul link between RSUs and the BS is wired
line, the capacity of backhaul link can guarantee a robust cross-layer of floading.

To implement cross-layer o ffloading, it is necessary to makea fast and dynamic of floading

policies. Leveraging in-deepcognition of AI, the proposed architecture can intelligently make of
floading decisions and resource allocations for cross-layer off loading. For example, the BS can
obtain current states of RSUs and diverse requirements of vehicles in advance by utilizing AI
algorithms. Then, the BS can generate the dynamic ffloading policy for each vehicle, and
specific resource allocations for each RSU and itself based on these parameters.

2) Cooperative Multi-point Caching and Delivery: Caching popular contents at RSUs enables
one-hop fast V2R content delivery and the alleviation of backhaul congestion [12].However,
due to the limitation of caching resources, a single RSU cannot store the whole content with a
very large size. Since neigh boring RSUs can communicate with each other via high-capability
wired line, a popular large-size content can be divided into several segments and cooperatively
cachedat several RSUs. With the help of AI, content features, such as current popularity,
potential popularity, storage size, and location are easily learnt.

7
 Empowered Edge

The proposed architecture facilitates mobility-aware content delivery. A detailed example of

content delivery is presented in Fig. 1. The content that the yellow smart vehicle requests can
be separated as three segments and distributed cached into three adjacent RSUs. Let us discuss
two separate scenario sabout how the yellow smart vehicle can obtain its requeste dcontent. If
the yellow smart vehicle passes all three RSUs, it can receive each segment directly from the
corresponding RSU. If it enters into the coverage of the first RSU and then stops, it can receive
segment 1from the first RSU but can no to btain the other two segments from the second and
the third RSUs. To meet the request of the yellow smart vehicle, the first RSU calls for segment
2 and segment 3 via wired line to complete content delivery process.

3] V2V Edge caching : V2V communication is proposed with the main advantage of enabling
multiple direct communication between pairs of near-by vehicles. Since state-of-the-artsmart
vehicles are equipped with certain caching resources, the accumulative caching power from a
group of these vehicles issufficient to store reasonable amount of content. The cache-enabled
vehicles can be utilized as mobile caching servers .Therefore, if a specific vehicle requests a
content, it can obtain from a mobile caching server.V2V edge caching has two potential
advantages. Sincevehicles can deliver a certain contents from one place toanother without any
wireless and wired communication, one advantage is that V2V edge caching can save fronthaul
and backhaul bandwidth. The other advantage is that V2V edge caching can allow contents to
reach the place out of the coverage of RSUs and BSs.

3.2 Working
To meet the demands from computationally intensive vehicular applications, some studies
have investigated applying MEC approach in vehicular networks. In , the authors proposed an

8
 Empowered Edge

energy efficient resource allocation for vehicular fog computing centers. In, an MEC-based
architecture was used in urban traffic management in a distributed and adaptive service
manner. In the authors designed a fog vehicular computing framework that integrates
resources from both edge server and remote cloud. In , the authors unveiled underutilized
vehicular computing resources, and put them into use for providing efficient computational
support to MEC servers. To efficiently merge MEC technology in vehicular networks, the
authors in introduced a collaborative task offloading and output transmission mechanism. In ,
the authors designed an MEC service migration scheme that ensures vehicles always connect to
the nearest MEC entities. Although these studies have provided some insights about MEC-
enabled vehicular applications, the effects of vehicular communication on the design of task
offloading strategies have not been thoroughly investigated.

Benefiting from fast commercialization of LTE system, LTE-V has been one of the key
technologies in vehicular networks. Several recent works have focused on analytical models
and implementations of LTE-V. In, direct vehicle communication was utilized to offload data
transmission fromvehicles with poor quality link to infrastructures. Taking into account high
mobility of vehicles, the authors in designeda wireless link formulation mechanism, where
beam widths between vehicular communication pairs were optimized. In the authors discussed
key building blocks of 5G networks in the context of vehicular communications. However, joint
V2I and V2V transmission schemes in a multiple MEC server scenario have not been considered
in the previous studies.Learning is a branch of artificial intelligence, which studiessystems and
acquires knowledge from data. Recently, variouslearning techniques have been deployed for
scheduling task offloading. In , the authors proposed an online learning based workload
offloading scheme in mobile edge computing systems with renewable power supply. In order
to reduce resource consumption in task offloading, the authors in formalized intelligent
offloading metric prediction utilizing a machine learning based approach. Deep Q-learning is a
powerful tool in policy optimization, and has been utilized in various process decisions. For
instance, the authors in designed an integrated resource management scheme for connected
vehicles using a deep reinforcement learning approach. The authors in used deep Q-learning in
scheduling voltage and frequency for real-time systems in embedded devices. In , this learning
approach was adopted in designing a video streaming framework. Moreover, deep Q-learning
also can be used in traffic area. To relieve traffic congestion at highway junctions, the authors
in applied deep Q-learning in traffic simulation study and vehicle pathway optimization.
However, the potential of learning based approaches have not been explored for designing
scheduling algorithms for vehicular edge computing applications. Furthermore, mobile
characteristics of vehicles and reliability of vehicular task offloading have not been considered
in the previous studies. Different from these studies, in this paper we concentrate on task
offloading in an LTE-V network, and propose optimal offloading schemes that jointly schedule
vehicular communication and edge computing through a deep Q-learning approach.

9
 Empowered Edge

TECHNOLOGIES
1) NETWORK FEATURES VIRTUALIZATION (NFV) AND SOFTWARE-DEFINED NETWORKS(SDN)

NFV enables the abstraction of physical network resources and the flexible sharing of resources
between isolated the objectives of minimizing the delay and preserving the user’s device battery life.
Virtualization technology is a key technology of MEC and realizes the separation of the service layer and
the phys-ical resource layer of edge computing, and can assign task sto various physical resources,
thereby efficiently utilizing re-sources. By integrating NFV into the MEC server, virtualized computing
and storage resources can support the functions of various applications and services and can be applied
to the server for functional programming to support a variety of application services, thereby enhancing
the flexibility ofthe server and reducing the cost function supply [26]. SDN isa new network mode that
was proposed by the CLean Stater esearch group of Stanford University in the United States. It is an
implementation method of NFV. By separating the control surface of network devices from the data
surface and opening the programmability, the logic centralized con trolof distributed network nodes
and mobile devices can berealized . Reference [28] studies SDN in super-dense network task of floading
problems and designs the edge of a cloud or of floads tasks on a local process scheme; the main
calculation and control function is separated from the distributed small unit base station, which is
integrated into the centralized SDUDN in the macro cell base station controller. Based on the decision of
the SD UDN controller, it is decided

Whether the mobile device should perform tasks locally or of fload tasks to the edge cloud for
processing, and the computing resources should be optimally allocated to each task to realize the
objectives of minimizing the delay and preserving the user’s device battery life.

10
 Empowered Edge

2]collaborative Mobile Edge Cloud Computing : Collaborative mobile edge cloud computing combines
the advantages of MEC and MCC, which is of substantial significance for ensuring the full utilization of
MEC and cloud computing resources. While cloud computing may produce long delays during of
floading, it can provide sufficient cloud computing resources. Although an MEC serve rout performs
cloud computing in reducing the communication delay and the energy consumption, with the increasing
useof computation-intensive applications, the limited computing esources of the MEC server cannot
fully satisfy all the stallation requirements. With the increasing number o f computing tasks, the
resource bottleneck problem of the MEC server becomes increasingly prominent. Therefore, cloud
computing and MEC should be highly complementary. Reference [29] proposed a collaborative of
floading schemefor vehicle-to-vehicle networks that is based on mobile edge cloud computing and cloud
computing, and it developed a distributed computing of and resource allocation algorithm for
computational o vehicle-to-vehicle networks. Reference [30] proposes sign framework for edge
computing in wireless broad band access networks that supports smart cities by embedding , viable
virtual network. A suitable resource partitioning is used for each virtual network embedding, and
backing up edge devices by using heuristic policies to deter-mine the number and geographic location is
recommended. [31], et al. designed an iterative heuristic MEC resource allocation algorithm for making
unloading dynamically. In reference [32], et al. Con side red the collaborative calculation of the of ,
combined power and time distribution. They proposed a capture-unload protocol that is based on the
block-time-division mechanism for minimizing the transmission power of wireless access points.

3) Content Distribution In : the context of mass content delivery, a suitable content distribution scheme
can facilitate the avoidance of transmission by the network. In addition, the application of the content
distribution framework in heterogeneous IoV systems can improve the message accuracy and reduce
the communication overhead between vehicles and the infrastructure [37]. Current mobile users have
con substantial amount of the capacity, and the demand vehicle infotainment services is still growing
rapidly. prove the network performance and the user quality of service (QoS), content distribution is
often combined with content caching technology and data technology to further reduce the data access
latency. In reference [38],a content propagation box that is based on edge calculation proposed. First, a
two-stage relay selection algorithm is designed to facilitate edge computing devices in the selective
transmission of content via V2I communication. Then, the vehicle that is selected by the edge computing
device relays the content via V2V communication to the vehicle that is interested in the content during
the trip to the destination .Reference [39] proposed a content distribution framework that utilizes 5G
edge network caching and wireless link timeslot scheduling. The wireless resource allocation and
return link utilization of vehicle-to-roadside-unit communication at each information station are
considered. To maximize the throughput, wireless links are dynamically allocated to vehicles using time
slots. In reference [40], we studied the impacts of the storage cost and the retention time of con-tent
storage on cache optimization in mobile scenarios. In addition, a cache problem in a vehicle network is
modelled ,and its complexity is analyzed. For symmetric cases, an optimal dynamic programming
algorithm with polynomial time complexity is developed. For general cases, a multi-helper caching
algorithm with low complexity and effective retention perception is proposed, which can obtain the
best caching solution.

4.1 Features
Empowered Edge Computing facilitates the operation of end-to-end heterogeneous networking and
distributed cloud platforms, including physical and logical resources and devices, cloud computing
operates on a similar principle as web-based email organizations, allowing users to access all of the

11
 Empowered Edge

features and files of the system without having to keep the bulk of that system on their own computers.
As well, aws helps you migrate with the most robust set of tools and services to simplify and automate
your migration and, the most proven migration methodologies, support services and experienced
people to help you on every step of your migration journey.

4.1.1 Digital Development

By contrast, edge computing is a decentralized computing platform that brings cloud computing
capabilities near IoT devices, that is, the network edge, deliver the best online experiences possible with
your edge cloud platform, built to strengthen your current teams and technology. Also, the product
features powerful edge computing capabilities, to reduce cloud-end computing resources, and realize
data optimization, real-time response, agile connection and model analysis on the IoT edge, further
advancing the development of digital networking in the AI era.

4.1.2 Paramount Data

High demand for low latency services and local data processing has given rise for edge computing, as an
extension of the cloud, the recently emerging edge computing migrates enormous computing and
storage resources to the network edge, which forms an edge layer that is close to IoT end devices. More
than that, securely transferring only actionable data and exceptions to the cloud becomes paramount to
business success when capturing and immediately sifting information at the network edge.

12
 Empowered Edge

5. APPLICATIONS chapter-5
5.1 Application 1
5.1. OFFLOADING DECISION
In offloading decision-making, data transfer between dependent tasks is typically considered. Mobile
terminal computing offloading methods mainly include local computing, offloading to the MEC server
for execution and offloading to the cloud server for execution. Many studies have been conducted on
offloading decision-making, such as studies on whether to offload, the quantity and location of of
floading ,service type, user perfection, access technology, network traffic, device performance, and edge
node property . Theo f floading method is mainly based on the resource size, the calculation and return
time and the power consumption of the calculation. The main influencing factors are the delay and the
energy consumption. To minimize the cost in terms of communication and computing resources, the
author of reference proposed a task diversion mechanism in the edge computing network of vehicles
under the condition of high mobility of the vehicles.

2] Computational Resource Allocation The: objective of computing resource allocation is to mini-

mize the cost of task processing so that resources can befully and reasonably utilized. It consists of two
processes: task assignment, namely, the assignment of tasks that can be executed in parallel to specified
resources, and resource allocation. The execution order of tasks is determined accord-ing to the pre-
established resource allocation strategy. In the MEC scenario, computing resource allocation is also
used to improve the overall system performance and to reducethe overall execution time and resource
consumption. Com-puting resources are often considered in conjunction with of floading decisions,
which can be divided into single- Accessrithm is proposed for handling the joint task of loading
andallocating computing and communication resources to mini-mize the energy consumption, the
calculated weighted totalcost, and the maximum latency among all users. In reference[35], an algorithm
is designed for making joint selection decisions and calculating resources and the of fload rate. Acom
prehensive task processing delay is used to develop thesystem utilities, which considers both the
transmission and computation times. This scheme substantially improves theperformance of load
balancing and maximizes the systemavailability.

3] mobility Management of Computational Of floading : Mobility management of computing

resource offloading isof substantial significance to the integrity of the user com-puting of floading
process. Due to the mobility of the user,it is inevitable that the user will be disconnected from the base
station. Mobile cellular networks ensure the continuityand quality of service by switching among base
stations. For scenarios with low user mobility, during the process ofoffloading the application to the
MEC server, the power ofthe current base station can be adjusted adaptively to ensureuninterrupted
service. If the user switches to a new servicebase station, virtual machine migration of the compute
nodeis used to solve the problem. In reference [24], a joint taskallocation, subchannel allocation and
power allocation problem is formulated. Aiming at maximizing the total of floadingrate, a hybrid
computing shunt management framework forreal-time traffic in 5G networks is proposed. A joint power
control and channel allocation scheme is designed based onnon-orthogonal multiple access and mobile
edge computing.MEC can reduce the computing limitations and extend the service life of mobile devices;
however, it will lead to thedense distribution of MEC servers. Although MEC server sare close to the
mobile users, they face user-related challenges, which will affect the computing shunt. Reference[36]

13
 Empowered Edge

focuses on the joint computing of offloading and multi-task user correlation, and it studies the
scheduling proble min distributed MEC systems with densely distributed MEC servers. To reduce the
energy consumption or improve the performance, an efficient algorithm for calculating of floadingis
proposed by considering the distribution of the computational resource and the transmitted power. A
comparison of computing offloading in MEC is presented .

5.2. DDPG- EMPOWERED VEHICULAR EDGE COMPUTING AND

CACHING SCHEME
PROBLEM FORMULATION :- The joint vehicular edge computing and caching problemcan be
formulated as an optimization problem to maximizesystem utility. In this problem, system utility
consists of com-puting utility, caching utility, and cost. Specifically, computingutility is equal to ∆tij ·BCi,
where ∆tij is the saving timewhen edge server jfulfills computation task ibefore itsdeadline, and BCiis
the price that smart vehicle ipays toedge server j. If edge server jcompletes the task before itsdeadline,
the saving time is positive and the corresponding computing utility is also positive. Otherwise, both
saving timeand computing utility are negative. This is reasonable becausefor a delay-sensitive task, it
may result in a certain loss ifit is not completed before its deadline. Caching utility isequal to ρi·Tij ·BSi,
where Tij is the caching duration thatedge server jcaches content i,BSiis the price to requestthe content,
and ρiis the popularity of content i. The cost isthe energy consumptions on communication, computing,
andcaching, respectively.Exploiting the proposed architecture, the information interms of computing
and caching capabilities, the requirementsof different vehicular applications, content popularity,
mobilityparameters, can be collected and sent to the agent. Here, weconsider three mobility
parameters, i.e., location, speed, anddirection. Location can be utilized to determine the nearestbase
station for a specific vehicle. Speed and direction areformulated as the sojourn time that a specific
vehicle will stay in the coverage of its nearest base station.After collecting the above information, the
agent designsan action to perform resource allocation. There are three keyelements in the deep
reinforcement learning, namely state,action, and reward

1) State: The state in deep reinforcement learning is aspace to reflect the situation of vehicular
environment. Thestate consists of four components S= (Di, Fj, Gj, Bj).Diis the state of smart vehicle i,
which includes the size ofits computation task, the size of the requested content, therequired
computation resources, the required cache resources,deadline, popularity of the requested content, and
mobilityparameters. Fj,Gj, and Bjare the available computation re-sources, available caching resources,
and available bandwidthof edge server j, respectively.

2) Action: The objective of an agent is to map the space ofstates to the space of actions. In this system,
the action consistsof three parts: fij , gij , and bij , which are the amount ofcomputation resources,
caching resources, and bandwidth thatthe BS or the RSU allocates to smart vehicle i, respectively. 5Note
that fij ,gij , and bij are continuous values here.

3) Reward: Based on current state and action, the agentobtains a reward from the environment. Since
reward functionis related to the objective function, in this scenario, systemutility can be regarded as the
reward function.

DDPG-empowered Vehicular Edge Computing andCaching Scheme

14
 Empowered Edge

Since the action space of the joint vehicular edge computing and caching problem contains plenty of
continuous variables,we propose a DDPG-based vehicular edge computing and caching resource
allocation, as shown in Fig. 2.Three components constitute the functioning of the DDPG algorithm (i.e.,
primary network, target network and replay memory.)The primary network consists of two deep
neural networks, namely an actor network and a critic network. The actornetwork is used to explore
the policy, which specifies the current policy by deterministically mapping state to a specificaction. The
critic network estimates the performance and provides the critic value which helps the actor to learn
thegradient of the policy. Thus, the input of the primary networkis the current state from vehicular
environment, the trainingstate and the training action from replay memory. The outputof the primary
network is the action which is adopted by theBS, RSUs and smart vehicles.The target network can be
defined as an old version of theprimary network, which is used to generate the target valuefor training
Critic-P.

The replay memory stores experience tuples which include current state,the selected action ,reward
and next state includes a target actor network includecurrent state, the selected action, reward, and
next state. The stored experience tuples can be randomly sampled for train-ing primary network and
target network. Randomly sampledexperience tuples aim to reduce the effects of data correlation.In
DDPG, the vehicular environment consists of a BS,RSUs and vehicles, as shown in Fig. 2. The agent is at
theBS to design action and it will send the designed action toRSUs and vehicles. Based on the action, the
BS, RSUs, andvehicles cooperatively execute edge computing and caching.Then, vehicles give feedback
to reward the received services.The detailed processes of DDPG works as follows.First, the state of
vehicular environment is sent to the Actor-P of the primary network and replay memory.Second, based
on current state and experience tuples, theagent uses primary network and target network to
determinenext action. The action includes three continuous values (i.e.,computation resources, caching
resources, and bandwidth). Inthis step, Actor-T and Critic-T in target network update targetpolicy value
θµ0and target Q value θQ0based on experiencetuples from replay memory, respectively. Then, Critic-T
intarget network sets yi=ri+γQ0(si+1 , µ0(si+1|θµ0)|θQ0)andtransmits it to Critic-P in primary
network. After receivingyi, Critic-P in primary network updates primary Q value θQby solving an
optimization problem which aims to minimizethe loss function Ls(θQ)=(yi−Q(si, ai|θQ))2. Based
onθQand experience tuples, Actor-P in primary network makespolicy gradient and then generates next
action.Third, the BS, RSUs, and smart vehicles, based on thedetermined action, execute edge computing
and caching. Afterexecuting resource allocation, each state of the BS, RSUs, andsmart vehicles is
transited from one to another, respectively.Then, the environment feedbacks an immediate system
rewardto the agent, based on the new state. If current resourceallocation satisfies all constraints of the
proposed joint edgecomputing and caching problem, and the system utility ofcurrent resource
allocation is greater than the existing maximalsystem utility, the immediate system reward is updated
ascurrent system utility and the environment updates its statebased current resource allocation. If
current resource allocationsatisfies all constraints of the proposed problem but the systemutility of
current resource allocation is smaller than the existingmaximal system utility, it indicates that DDPG
does not gener-ate a better solution such that the immedated based on the existingbest resource
allocation policy.

5.3. RESEARCH CHALLENGE AND OPEN ISSUES

In the previous sections, we reviewed the architecture andrelated technologies of MEC, AI, and AI-based
vehicle edgenetworks in IoV. In addition, we analyzed the previous re-search from three aspects.

15
 Empowered Edge

However, the future IoV still faceschallenges. In this section, we will discuss several possibleresearch
challenges and propose several promising researchdirections.

A. SECURITY AND PRIVACY

In recent years, security and privacy issues in IoV havereceived extensive attention. Mobile vehicles
collect infor-mation via V2V communication between vehicles and viaV2I communication from vehicles
to roadside infrastructure.Due to the high mobility of vehicles, communication isoften interrupted,
thereby resulting in frequent failures ofcommunication links. In addition, hackers’ security attackson
communication channels and sensor tampering will leadto severe privacy invasion. In addressing these
security andprivacy issues, challenges in the solution of identity privacy. data privacy and location
privacy issues will be encountered.Potential solutions include communication authentication,MEC and
access control of cloud computing servers [70].Reference [71] proposes an architecture of edge
auxiliarynetwork connecting vehicles. To solve the problem of loca-tion privacy, a location-based
differential privacy protectionservice framework is proposed for ensuring location privacywithin the
coverage of the edge nodes. Li et al. [72] proposedan online double auction scheme for k-anonymous
locationprivacy protection, which could solve the problems of op-timal charging scheduling for electric
vehicles and locationprivacy protection for owners of electric vehicles. Chen etal. [73] designed a data
trading method for the Internetof vehicles that is based on block chain. An iterative dualauction
mechanism is used to protect the privacy of bothparties in data transaction, to reduce the data
transmissioncost and to improve the system stability.

B GREEN ENERGY SAVING

Green energy saving has a profound impact on the construc-tion of a green IoV. Automobile exhaust
emission is themain factor that affects the human environment and the airenvironment. To alleviate the
current environmental pollutionscenario while adapting to the highly dynamic traffic envi-ronment, it is
highly important to use RSUs to communicatewith nearby vehicles to realize efficient task scheduling
tosatisfy vehicle communication requirements. Energy savingin RSU scheduling and RSU energy
collection are essentialfor solving the problem of energy consumption. The imple-mentation of a wind
or solar RSU in an energy-constrainedvehicle environment can increase the network capacity
andpromote energy recovery. In addition, the minimum numberof active RSUs can be set to maintain
the network operationand connectivity [74]. To minimize the total energy con-sumption of RSUs under
the delay constraint, reference [75]constructed an MEC-based IoV energy-saving schedulingframework
for balancing the computing tasks among RSUs.A heuristic algorithm is designed that considers the
taskscheduling among MEC servers and the energy consumptionof the RSU downlink. In addition,
electric vehicles, which arepowered by electric engines instead of internal combustionengines, which
are powered by fossil fuels, can effectivelyreduce the carbon footprint and play an important role
inrealizing efficient energy management .

C. HIGH MOBILITY

Mobility is an important feature of vehicle networks. Withthe rapid increase of the road traffic density,
high speedand frequent vehicle movements are the main factors thatrender the network topology
dynamic. The high mobility ofintelligent vehicles not only adds considerable complexityin co-optimizing
the allocation of computing and cache re-sources but also hinders the provision of stable and
reliablewireless communication [77]. First, the data transmissiondistance is constantly changing due to
vehicle movements.Therefore, the data rate and the effective duration of channel transmission in V2X
communication will also be affected.Second, the changes in the vehicle speed and direction overtime

16
 Empowered Edge

will lead to frequent handovers between edge servers.Active communication management is

transferred from oneRSU or BS to another RSU or BS. The duration for which avehicle remains within
the coverage area of the RSU or BSalso varies. Due to the widespread use of various Global Po-sitioning
System (GPS) devices and mobile Internet in dailylife, vehicle trajectory data can be easily obtained on a
largescale [78]. Therefore, addressing the high mobility of IoV bypredicting the vehicle movement
direction and studying datarouting distribution protocols is a feasible solution .

D INTELLIGENT COMPUTATION

With many edge nodes deployed in 5G networks, edge com-puting has the advantage of reducing the
traffic load and thebackhaul pressure, but edge devices still face the challengeof real-time processing.
Edge cognitive computing has be-come a new paradigm. By analyzing and interpreting theavailable data
and information in cyberspace, the intelligenceof machines can be increased for the prediction and
genera-tion of new information, thereby providing more intelligentcognitive services. Reference [80]
proposed an architectureof edge cognitive computing by combining edge computingand cognitive
computing. Considering the elastic distributionof cognitive computing services and the mobility of users
,a dynamic cognitive service migration mechanism that isbased on edge cognitive computing is
designed. It integratesthe communication, computing, storage and application on the edge network,
improves the user experience and realizesrational resource allocation and cognitive information .

5.4.edGe IntellIGence And blockchAIn empowered5G beyond IIot

networks
Figure 1 shows the framework of our proposed intelligent and secure 5G beyond IIoT networks,

which leverages edge intelligence and block-chain in edge service scheduling and resource transaction
management. The framework con-sists of three planes. The industrial application plane, which is at the
bottom of the framework, illustrates typical applications with intensive edge resource requirements.
With the aid of edge resources distributively deployed in close prox-imity tothe production site,
warehouse, logistics road, and manager offi ce, these applications can be executed such that their
performance require-ments can be met. In order to effi ciently sched-ule heterogeneous edge resources
while off ering a secure and trustworthy digital platform edge service transactions, we deploy
blockchain and AI modules in the edge intelligent network plane. The core network plane located at the
top of the framework is a fundamental part of the IIoT network, seamlessly integrating IIoT and remote
cloud. This plane consists of high-per-formance computing servers and large-capaci-ty storage space.
When edge resources are not adequate for serving the industrial applications, the servers in the core
network can provide aux-iliary service, especially for delay-tolerant appli-cations.

5.5. edGe IntellIGence for 5G beyond IIot

By integrating environment cognition, data analytics, machine-to-machine communication, and robotic
automation, IIoT holds great potential to enhance manufacturing effi ciency and to enable powerful
industrial applications [8]. These appli-cations have intensive resource demands and low latency
constraints. For instance, in automatic manufacturing, to make robotics operate effi cient-ly, high-
performance computing and communica-tion resources are required for real-time sensory data
processing and low-latency control signaling, respectively. In addition, the implementation of smart
warehousing demands caching resource to store a large scale of warehousing information. Itis a critical
challenge to perform these powerful applications on resource constrained industrial devices.In 5G

17
 Empowered Edge

beyond, both infrastructure and devices will be connected ubiquitously and reliably with the help of
seamless communication coverage. This brings high agility in resource deployment and facilitates
pushing service to the proximity of end subscribers. By off loading tasks directly to edge servers, the
resource burden of the devices can be alleviated, while application processing

latency is reduced.To meet diverse resource demands and diff er-ent performance requirements of the
generated applications, edge resources need to be effi ciently managed. Edge intelligence, which enables
edge serving nodes with distributed AI capability, is a promising paradigm for resource scheduling. By
integrating AI function into edge networks, edge intelligent servers have full insight of working
environments in terms of the resource correlation between heterogeneous types and the feasibility of
cooperation with adjacent nodes. In addition, accurate prediction of resource states and appli-cation
demands can be obtained with the help of AI algorithms in the edge, which contributes to improving
resource scheduling efficiency. More-over, as edge services typically utilize resources approximate to
subscribers, edge resource sched-uling does not need to know the information of the entire IIoT
network. Transmitting locally learned serving information directly to distribut-ed AI processing entities
in such an edge intelli-gence approach is more efficient than obtaining resource scheduling strategies
from remote cen-tral AI nodes. Thus, inspired by edge intelligence, diverse servers at the edge
intelligent network plane can recognize serving context, understand industrial application demands,
incorporate het-erogeneous resources, and offer flexible commu-efficient manner.Blockchain for 5G
Beyond IIoTIn edge intelligence empowered IIoT networks, resource-hungry industrial applications can
be offloaded to or aided by edge servers for better performance and lower latency. In the serving
process, some sensitive information, such as prod-uct parameters and business decisions, may be
passed to the edge servers for processing. A pro-tection mechanism is required to minimize the risk of
sending data to fake or untrusted edge nodes [9]. Furthermore, during the service, resource
transactions are performed between subscriber devices and edge service providers. In order to
guarantee the authenticity and fairness of transac-tions, key information about the transaction, such as
resource type, quantity, and price, must be recorded. In addition, applications always need to be
accomplished with the aid of heterogeneous resources at multiple edge nodes. For instance, a
computing task offloading first uses communi-

cation resources for file transfer and then utilizes computing resources for processing. Thus, there is a
correlation between transactions in an edge ser-vice process. An efficient and automatic arrange-ment
of consecutive edge resource transactions is therefore necessary. As 5G beyond technique brings
enhanced communication capacity as well as ubiquitous edge services, a large scale of diverse resource
transactions can be performed all cross the net-works. Traditional central control approaches are
usually prone to high overhead and delay for transferring volumes of information across geog-raphy.
Moreover, in edge intelligence empow-ered IIoT, the implementation of automatic edge service relies on
a reliable and tamper-resistant transaction management mechanism. Howev-er, a central control node
may be vulnerable to malicious attacks. Thus, designing efficient, flexi-ble, and secure transaction
management for 5G beyond IIoT emerges as a critical challenge. To address the challenge while meeting
the above requirements, we incorporate blockchain into the IIoT networks. Blockchain, which is a key
technology in the 5G beyond era, acts as a tam-per-resistant distributed ledger sharing and storing data
among a large amount of nodes, and helps keep the security and privacy of the IIoT edge network
ecosystem [10]. Specifically, the distribut-ed ledger technology is of particular relevance to the
distributed topology of IIoT networks, which can contribute to accelerating transactions .

5.6. State-of-the-art AI Approaches

18
 Empowered Edge

Convex optimization [2], [4], [5], [7], [9] and game theory[1], [8], [10] have been extensively used to
address the edgecomputing and caching problems. Nevertheless, these existingoptimization methods
may suffer from the following issues:1) Some key factors, such as wireless channel condition
andcontent-popularity are assumed to be known. In reality, theyare time-varying and difficult to obtain.

2) High mobility ofsmart vehicles results in dynamic communication topologyand complicated

correlation between vehicular communicationpairs. As a result, efficient and reliable data transmission
invehicular networks becomes even more challenging;

3) Mostof the proposed algorithms are optimal or close-to-optimalonly for a snapshot of the system
without considering the long-term effect of the current decision on resource allocation.AI is a promising
approach to facilitate in-depth featurediscovery such that the uncertain input fields for edge com-puting
and caching problems can be obtained in advance.In addition, AI is also a promising tool to tackle
complexoptimization problems, such as resource allocation [12], [13].Deep reinforcement learning is a
branch of AI research wherean agent learns by interacting with environment. There arethree common
deep reinforcement learning algorithms: Q-learning, Deep Q Network (DQN), and DDPG:

Q-learning: Q-learning is a classical model-free and off-policy deep reinforcement learning algorithm
based on bell-man equation. It iteratively approximates a Q-function Q(s, a),where s, a are the state and
action, respectively. The agent ofQ-learning needs to compute the Q-function of each state-action pair
and stores its result, named Q-value, into a uniqueQ-table Q-learning is a simple incremental algorithm
de-veloped from dynamic programming. But it is not an idealalgorithm for the problem with a high-
dimensional observationspace, since the possible states of a system may be more thanten thousands
making it difficult to store all Q values of statesinto a Q-table.

DQN: DQN is a kind of deep reinforcement learningalgorithm which uses deep neural networks instead
of Q-function to explore actions [14]. DQN has two major factors tomake it powerful, namely experience
replay and fixed Q-target.Experience replay is a method to destroy the dependence ofcollected
experience, allowing the network to learn more frompast experience and fixed Q-target. DQN is a
powerful tool thatcan learn optimal policies with high-dimensional observationspaces. However,
applying DQN to a continuous domain is stilla challenge since it relies on finding the action that
maximizesaction-value function.

DDPG: DDPG is an actor-critic and model-free algorithmthat can learn policies in high-dimensional,
continuous actionspaces [15]. DDPG combines the actor-critic approach withDQN. The actor network is
used to explore policy and thecritic network is used to estimate the performance of the pro-posed
policy. To improve the learning performance, techniquessuch as experience replay and batch
normalization of DQN areemployed. Compared to the other two algorithms, the mostsignificant feature
of DDPG is that it can make decisions orallocations in a continuous action space while Q-learning and
DQN only can support the problem with a discrete actionspace

.B. Design Challenges

One of the key architecture principles is to utilize AI tointelligently monitor vehicular network and
manage resources.However, because of mobility, handovers between smart vehi-cles and RSUs are
quite frequent which make edge computingand caching resource management even more
challenging.Since a computation task or content needs to be dividedinto several segments to support

19
 Empowered Edge

computation offloading oredge caching, how to divide each one of them to improvesystem utility while
satisfying diverse requirements of vehiclesshould be taken into policy design. Furthermore,
interferencemanagement also should be incorporated for a wireless com-munication scenario.Since lots
of uncertain and time-variant states need to bewell predicted, AI algorithms should store plenty of
historicaldata for accurate prediction. This behavior will consumea certain amount of caching
resources. Since the cachingresources are limited and mainly used to popular contentcaching, there
exists a trade-off between prediction accuracyand content caching.The proposed AI-empowered
vehicular network utilizesAI approaches to perform in-deep cognition and intelligentorchestration for
vehicular edge computing and caching. How-ever, AI has mostly been studied in the area of computer
visionand data processing, while little attention has been paid to thevehicular network and MEC system.
In order to realize theproposed architecture, there are still some challenges to beaddressed. One of the
key architecture principles is to utilize AI tointelligently monitor vehicular network and manage
resources.However, because of mobility, handovers between smart vehi-cles and RSUs are quite
frequent which make edge computingand caching resource management even more challenging.Since a
computation task or content needs to be dividedinto several segments to support computation
offloading oredge caching, how to divide each one of them to improvesystem utility while satisfying
diverse requirements of vehiclesshould be taken into policy design. Furthermore,
interferencemanagement also should be incorporated for a wireless com-munication scenario.Since lots
of uncertain and time-variant states need to bewell predicted, AI algorithms should store plenty of
historicaldata for accurate prediction. This behavior will consumea certain amount of caching
resources. Since the cachingresources are limited and mainly used to popular contentcaching, there
exists a trade-off between prediction accuracyand content caching.

20
 Empowered Edge

6. ADVANTAGES AND DISADVANTAGES chapter-6

6.2 Advantages
 It offers high speed, reduced latency better reliability which allows for quicker data processing
and content delivery.
 It offers better security by distributing processing, storage, and applications across a wide range
of devices and data centers, which makes it difficult for any single disruption to take down the
network.
 It offers a far less expensive route to scalability and versatility, allowing companies to expand
their computing capacity through a combination IoT devices and edge data centers.

6.2 Disadvantages
 Security challenges in edge computing is high due to huge amount of data.
 It only analyse the data.
 Cost of edge computing is very high.
 It requires advanced infrastructure.
 It requires more storage capacity.

21
 Empowered Edge

7. CONCLUSION AND FUTURE SCOPE chapter-7

7.1 conclusion
In this study, two key technologies, namely, MEC and AI,were analyzed by focusing on the development
of intelligent IoV and the previous research on combining the two technologies. First, the
communication mode and architecture of the traditional Internet of vehicles were introduced, along
with the advantages of the emerging 5G network. in addition, MEC, FC and MCC were compared by
studying the development history of MEC. The advantages of MEC were analyzed, the MEC, FC and MCC
were compared by studying the development history of MEC. The advantages MEC were analyzed, the
key technologies of MEC were evaluated, and several key technologies for calculating the in MEC were
studied. Then, the differences and IoV were discussed, with a focus on the characteristics
andapplication status of DRL in dynamic vehicle networks. Then, combining the advantages of MEC and
AI technologies, theprevious research on the application of AI to vehicle edge networks was analyzed.
Finally, the possible future research by directions ofIoV were discussed.

7.2 Future Scope

Shifting data processing to the edge of the network can help companies take advantage of the growing
number of IoT edge devices, improve network speeds, and enhance customer experiences. The scalable
nature of edge computing also makes it an ideal solution for fast-growing, agile companies, especially if
they’re already making use of colocation data centers and cloud infrastructure.

By harnessing the power of edge computing, companies can optimize their networks to provide
flexible and reliable service that bolsters their brand and keeps customers happy.

Edge computing offers several advantages over traditional forms of network architecture and will
surely continue to play an important role for companies going forward. With more and more internet-
connected devices hitting the market, innovative organizations have likely only scratched the surface of
what’s possible with edge computing.

22
 Empowered Edge

REFERENCES

1) Author’s Name, “Article Title”, Complete URL of Web Page.

2) Author’s Name, “Title of the Book”, Publication, Edition, Year of Printing.

3) Artificial_Intelligence-
Empowered_Edge_of_Vehicles_Architecture_Enabling_Technologies_and_Applications.p
df
4) “Architecture of computer”, http://computer.org/cspress/instruct. html

5) Artificial Intelligence Empowered Edge Computing and Caching for Internet of Vehicles

6) Edge computing - Wikipedia

7) http://dx.doi.org/10.1109/JIOT.2019.2903191

8) https://theartofservice.com/empowered-edge-computing-what-are-the-features-of-
networks-that-support-effective-access-to-cloud-computing-services.html

9) https://www.geeksforgeeks.org/advantage-and-disadvantage-of-edge-computing

10)https://www.vxchnge.com/blog/the-5-best-benefits-of-edge-computing

23
 Empowered Edge

11)

12)“Architecture of computer”, http://computer.org/cspress/instruct. html

13)Artificial Intelligence Empowered Edge Computing and Caching for Internet of

Vehicles

14)Edge computing - Wikipedia

15)http://dx.doi.org/10.1109/JIOT.2019.2903191

16)https://theartofservice.com/empowered-edge-computing-what-are-the-
features-of-networks-that-support-effective-access-to-cloud-computing-
services.html
17)https://www.geeksforgeeks.org/advantage-and-disadvantage-of-edge-
computing

18)https://www.vxchnge.com/blog/the-5-best-benefits-of-edge-computing