ArticlePDF Available

Abstract and Figures

5G calls for a network architecture that ensures ultra-responsive and ultra-reliable communication links, in addition to the high degree of flexibility and customization required by different vertical sectors. The novel concept called local 5G networks enables a versatile set of stakeholders to operate 5G networks within their premises with guaranteed quality and reliability to complement Mobile Network Operators' (MNOs) offerings. In this paper, we propose a descriptive architecture for a local 5G operator which provides user specific and location specific services in a spatially confined environment i.e. industrial internet environment. In addition to that, we propose hybrid architecture options where both the local 5G operator and MNO collaboratively contribute to establishing the core network to cater to such communications. The architecture is discussed in terms of network functions and the operational units which entail the core and radio access networks in a smart factory environment which supports Industry 4.0 standards. Moreover, to realize the conceptual design, we provide simulation results for the latency measurements of the proposed architecture options with respect to an Augmented Reality (AR), massive wireless sensor networks and mobile robots use cases. Thereby we discuss the benefits of deploying core network functions locally to cater to specialized user requirements, rather than continuing with the conventional approach where only MNOs can deploy cellular networks.
Content may be subject to copyright.
1
Performance Analysis of Local 5G Operator Architectures for
Industrial Internet
Yushan Siriwardhana, Member, IEEE, Pawani Porambage, Member, IEEE, Mika Ylianttila, Senior Member, IEEE,
and Madhusanka Liyanage, Senior Member, IEEE,
Abstract—5G calls for a network architecture that ensures
ultra-responsive and ultra-reliable communication links, in addi-
tion to the high degree of flexibility and customization required
by different vertical sectors. The novel concept called local 5G
networks enables a versatile set of stakeholders to operate 5G
networks within their premises with guaranteed quality and
reliability to complement Mobile Network Operators’ (MNOs)
offerings. In this paper, we propose a descriptive architecture
for a local 5G operator which provides user specific and location
specific services in a spatially confined environment i.e. industrial
internet environment. In addition to that, we propose hybrid
architecture options where both the local 5G operator and MNO
collaboratively contribute to establishing the core network to
cater to such communications. The architecture is discussed in
terms of network functions and the operational units which
entail the core and radio access networks in a smart factory
environment which supports Industry 4.0 standards. Moreover,
to realize the conceptual design, we provide simulation results for
the latency measurements of the proposed architecture options
with respect to an Augmented Reality (AR), massive wireless
sensor networks and mobile robots use cases. Thereby we discuss
the benefits of deploying core network functions locally to cater
to specialized user requirements, rather than continuing with
the conventional approach where only MNOs can deploy cellular
networks.
Index Terms—Local 5G Networks, Industrial Internet, Indus-
try 4.0, 5G, Augmented Reality, Architecture
I. INTRODUCTION
The landscape of mobile communication service require-
ments is rapidly changing with the proliferation of digitization
technologies. Consequently, in the future, more emphasis
needs to be placed on location specific services in different
vertical sectors such as automotive, health, energy, industry
and media [1]. Hospitals, shopping malls, smart cities, fac-
tories and universities are identified as some of the common
locations, which are heavily benefited by these location spe-
cific services. Location specific requirements stipulate high
demands on reliability, high data rates, low latency, privacy
and security. The key focus of the future 5G wireless systems
is to serve such case specific requirements along with the
provisioning of the traditional mobile broadband services [2].
These case specific and localized requirements are expanding
beyond the current capabilities of the traditional MNOs whose
services are often designed to serve masses. To serve the
Yushan Siriwardhana, Pawani Porambage and Mika Ylinattila were with
the Centre for Wireless Communications, University of Oulu, email: first-
name.lastname@oulu.fi
Madhusanka Liyanage was with the School of Computer Science, Uni-
versity Collage Dublin, Ireland and Centre for Wireless Communica-
tions, University of Oulu, Finland, email: madhusanka@ucd.ie, madhu-
sanka.liyanage@oulu.fi
location specific future communication requirements, the need
for establishing local 5G networks is evident. In speeding up
local service delivery with 5G networks, the present mobile
communication market needs to be opened for local 5G
networks deployed by different stakeholders such as recently
proposed in the micro operator (uO) concept [3].
Unlike the traditional MNOs with wide area coverage,
uOs are local operators who intend to offer case specific
and location specific services through locally deployed 5G
networks [4], [5] . Therefore, the system architecture for
a local 5G operator (L5GO) should be carefully designed
in such a way that it enables efficient and reliable local
service deliveries. The local operator must deliver 5G services
and the system architecture of a L5GO must contain the
network functions defined by the 3rd Generation Partnership
Project (3GPP). Since these L5GOs provide tailored services,
the system architecture and its specific deployment may also
depend on the use case.
A. Our Contribution
Besides the novelty of L5GO concept, very few efforts have
been taken to define the system architecture for a L5GO. A
network architecture for emerging 5G uO, which serves one
use case (i.e. Augmented Reality) in a smart factory environ-
ment has been proposed [6]. This paper extends the work for
two other industrial use cases called massive wireless sensor
networks and mobile robots [7]. The proposed architecture
comprises 5G network functions and the operational units
which entail the core and the access networks to serve the
communications of the use cases. Apart from defining a pure
local architecture, this paper considers different variants of
hybrid architectures where both L5GO and MNO collabora-
tively deploy network functions to establish the core network.
To realize the conceptual design and present the simulation
results, we compare three network deployment models for a
factory, first being served solely by a L5GO, second being
served by a traditional MNO and third being served by an
operator having a hybrid architecture. In a hybrid architecture,
certain core network functions are deployed locally while
other core network functions are deployed by MNO. Based
on the simulation results, we discuss the benefits of locating
the core network functions closer to user locations to serve
specialized user requirements, rather than continuing with the
traditional MNO driven approach. Moreover, we conduct a
testbed implementation to compare and analyze the real world
behavior of a L5GO, hybrid operator and a MNO using the
performance metrics latency and throughput.
2
B. Paper Organization
The remainder of the paper is organized as follows: Sec-
tion II describes the related work on 5G architecture, L5GOs,
industry 4.0 and network slicing. Section III describes the
selected industrial use cases and Section IV explains the pro-
posed architecture for L5GO which serves the industrial use
cases. Section V illustrates the different architectural options
and different algorithms used in deriving those architectural
options. Section VI explains the approach to simulations
and presents the key parameters. Section VII discusses the
simulation results for a typical MNO setting and proposed
local and hybrid architectures. Finally, Section VIII concludes
the paper with the future research directions.
II. BACKGROU ND A ND RE LATE D WORK
A. Future 5G Networks and 5G Network Architecture
5G will be a paradigm shift from present wireless com-
munication technologies and it will also be highly integrative
to provide universal high-rate coverage and a seamless user
experience [8]. Key characteristics of 5G wireless systems
are identified as extremely high data rates, ultra-reliability
and low latency, and massive communication between de-
vices [9]. Three main 5G service classes are as enhanced
Mobile BroadBand (eMBB), Ultra Reliable and Low Latency
Communication (URLLC) and massive Machine Type Com-
munication (mMTC), [10], [11]. Based on the communication
needs of different verticals, future 5G operators must possess
the capability of providing case specific services in addition
to the present generic communications services [1].
3GPP has released the specifications for 5G system architec-
ture [12] and the network architecture of a L5GO should also
include the same Network Functions (NF) defined in generic
5G architecture [3]. Instead of the network elements defined in
Evolved Packet Core (EPC) in 4G systems, Software Defined
Networking (SDN) [13] and Network Function Virtualization
(NFV) are involved in creating Network Functions (NF) in 5G
systems architecture. Network functions can be implemented
on a dedicated hardware or as a software instance on a
dedicated hardware or as a virtualized function instantiated
on an appropriate platform such as a cloud. The concept of
network functions has led operators to add flexibility over the
functionality of the underlying physical infrastructure of the
5G network. 3GPP specifications represent the 5G architecture
in two ways:
Service based representation: Shows how NFs within
the control plane enable other authorized NFs to access
their services, as in Figure 1.
Reference point representation: Shows the point-to-
point interaction existing between two NFs, as depicted
in Figure 2.
B. Local 5G Networks
Currently, local 5G networks are gaining increasing at-
tention because the mobile communication market can be
expanded to new stakeholders and allowing them to deploy
local 5G networks to complement conventional MNOs. These
Fig. 1: Service based representation of 5G architecture [12]
Fig. 2: Reference point representation of 5G architecture [12]
L5GOs such as uOs are expected to provide tailored 5G
services and fulfill case specific and versatile local wireless
communication needs with extremely low latency [3]. A L5GO
can respond rapidly to future communication requirements
than a MNO as it provides case specific, location specific
needs while MNO’s main focus is serving to masses [14].
Moreover, it will be operationally difficult for a MNO to use
existing macro cells to serve these requirements because most
of the traffic will be originated from indoors. L5GOs have a
flexible design in which they can operate a closed network to
serve its own customers, an open network to offer its services
to other MNO’s customers, or a mix of both. Key regulatory
elements and the techno economic aspects related to the uOs
are discussed in [4]. Business model options for local 5G uOs
and the different network deployment options are discussed
in [14].
C. Industrial Internet
Industry 4.0 or smart factory environment is one particular
location which will be highly benefited by the services of
L5GOs. Industry 4.0 refers to the advancement of the present
industries into the next generation [15], [16]. It aims to
interconnect the devices inside the factories, make them smart
by adding more intelligence into the device and ultimately
resulting in improved adaptability, resource efficiency, and
the supply and demand process between factories [17]. In
industrial environments, Machine-to-Machine (M2M) commu-
nication plays a critical role, especially with the deployment
of sensor networks and Automated Guided Vehicles (AGV).
Wireless Sensor Networks (WSN) in current industries are
moving towards industrial wireless networks because of the
low latency, high mobility and high capacity requirements in
3
the future industries [18]. A study report has been released
by 3GPP focusing on typical use cases in Industry 4.0 such
as motion control, mobile robots, augmented reality, massive
wireless sensor networks [7].
D. Network Slicing
Use cases of future 5G communications will demand diverse
and sometimes extreme requirements [19]. Serving the needs
of these diverse requirements using a monolithic network
infrastructure will not be efficient, therefore the need for
flexible and scalable networks to serve these requirements is
a must. Also, having such flexible and scalable networks will
make the introduction of new services much easier. This can
be achieved using network slicing.
Network slicing sub-divides a network into logically iso-
lated sub-networks. These logical networks enable different
types of communications on a common infrastructure [20].
With network slicing, network usage can be optimized by
serving different types of communication via different slices
than serving them with just one network. 3GPP introduces
three network slice management functions for creating and
managing network slices in 5G networks called Communica-
tion Service Management Function (CSMF), Network Slice
Management Function (NSMF) and Network Slice Subnet
Management Function (NSSMF) [21], [22].
CSMF: Responsible for translating communication ser-
vice related requirements to network slice related require-
ments.
NSMF: Derive network slice subnet related requirements
from network slice related requirements and, responsible
for management and orchestration of Network Slice In-
stances (NSI).
NSSMF: Responsible for management and orchestration
of Network Slice Subnet Instances (NSSI).
In addition to the slice management functions, 3GPP intro-
duces Network Slice Selection Function (NSSF) to select the
appropriate network slice for a given communication.
III. USE CA SE S
3GPP Study on communication for automation in vertical
domains [7] describes a number of use cases which will
appear in the factories of the future. To define the network
architecture of the L5GO, we consider three use cases from
the 3GPP study which will frequently be seen in future
industrial environments. The use cases are selected in a way
that they cover all three generic 5G services i.e. eMBB,
URLLC and mMTC. The selected use cases are Augmented
Reality (AR), massive wireless sensor networks and mobile
robots as outlined in Table I.
TABLE I: Industry 4.0 Use Cases
Use Case eMBB URLLC mMTC
Augmented reality X X ×
Massive wireless sensor networks × × X
Mobile robots ×X X
As with any IoT device, power management of the deivces
is a critical concern for these use cases, especially at the smart
factory environments where massive number of such devices
will be used. L5GO deployments will primarily be with small
cell Ultra-Dense Networks (UDN), hence the user devices are
in close proximity to the base station. This helps to reduce the
transmit power of the devices such as AR devices, sensors, and
mobile robots. Terminal devices such as wearable AR devices
may not be able to perform complex processing within the
device, thus requiring computational offloading. The improved
coverage provided by UDN will support high throughput
transmissions required by the use case for offloading so that
the devices only perform mandatory processing, leading to
extend the battery life of the devices.
A. Augmented Reality
In AR use case, factory workers are supported by AR
devices and these devices are used to identify production
flaws, obtain step by step guidance to carry out pre-defined
tasks, obtain support from the supervisors. In this context, AR
devices should be highly energy efficient and lightweight. This
requires AR devices to carry out minimal processing and more
intensive tasks to be offloaded to a separate image process-
ing server located inside the factory. Typical communication
between AR device and the image processing server of an
Industry 4.0 AR network is depicted in Figure 3.
Fig. 3: AR system model with offloaded processing [7]
AR device takes the images and transmits them to the image
processing server. Server then performs the processing of the
images and sends the augmentations back to AR device to
display. The video transfer from the AR device to the image
processing server needs the 5G system to support higher band-
width. Moreover, 5G system supporting this communication
should be able to provide an End-to-End (E2E) latency of less
than 10 ms for one-way communication with a 99.9% success
of frame delivery [7]. A typical AR network along with the
core and access networks of the served operator is depicted in
Figure 4.
B. Massive wireless sensor networks
Massive wireless sensor networks will be used for monitor-
ing the working environment in future industrial environments.
Sensors can monitor various types of parameters such as
pressure, humidity, temperature, CO2and sound. The main
purposes of having a sensor network is to monitor the environ-
ment, detect malfunctions in the surrounding, take appropriate
actions by a decision-making entity so that the effect from the
malfunction is mitigated. As an example, when an anomaly of
the room temperature in detected, a machine can be triggered
4
Fig. 4: An AR network with terminal devices and the AR
server
to its emergency stop. Placement of the monitoring function is
a key design aspect of sensor networks which directly affects
the latency. In case of less computationally complex sensor
nodes, this functionality can be placed in a central cloud server.
At the same time, the functionality can be placed inside the
factory environment to support low latency requirements. A
typical industrial wireless sensor network along with the core
and access networks of the served operator is depicted in
Figure 5.
Fig. 5: Wireless sensor network with management and con-
trolling servers
In our study, we consider that the sensors are transmitting
the sensed data to a centralized management server, which
decides any controlling action based on the sensed values.
Management server then instructs the controlling entity and the
controlling entity takes the appropriate mitigating action. The
latency required for event based condition monitoring such as
a detecting a high temperature inside the factory premises is
from 50 ms to 1 s. Interval based condition monitoring such
as humidity measurements taken every 1 hour also requires
latency varying from 50 ms to 1 s. The highest priority is
given for condition monitoring for safety, needs the action to
be taken within 5 - 10 ms [7].
C. Mobile robots
In future industrial environments, mobile robots such as
Automated Guided Vehicles (AGV) will be used in numerous
applications and will play an extremely important role. These
robots can be programmed to execute multiple operations
fulfilling number of tasks such as transporting goods and pro-
viding assistance to workers (collaborative robots or cobots).
These robots can sense and react with their environment,
therefore they operate more intelligently than the traditional
machines programmed to travel along the pre-defined paths.
For the proper operation of mobile robots in an industrial
environment, they should be monitored and controlled by a
guidance control system. This will avoid collisions between
robots, assign driving jobs and manage the traffic of mobile
robots. In a smart factory environment, robots can guide
themselves using their own sensors such as cameras and
lasers. In our study, we consider the robots and the guidance
controller are connected to each other using 5G. Figure 6
depicts the connectivity between the robots and the guidance
control server.
Fig. 6: Mobile robots controlled by a guidance control server
Communications of mobile robots can be categorized into
three main cases based on the fact that who is communicating
to whom [7]. They are, communication between mobile robots
and guidance control system, communication between mobile
robots, communication between mobile robots and peripheral
facilities. Very stringent latency is a key requirement for
mobile robots operating in an industrial environment. A cycle
time of 1 ms to 10 ms is required for machine control, 10 ms
to 50 ms for cooperative driving of robots, 10 ms to 100 ms
for video operated remote control and 40 ms to 500 ms for
standard mobile robot operation [7].
IV. PROPOSED ARCHITECTURE
A local 5G network covering the factory could be used
to address the needs of the three use cases. This local 5G
network should comprise architectural components inherited
from generic 5G systems architecture. The concept of locally
deploying the network provides flexibility over the selection
of architectural components and the location where the core
network is hosted. For a low latency requirement, the desirable
5
implementation is to have the core network within the factory
premises itself, but not mandatory.
In our study, we define the architectural components needed
in the core network to serve each use case separately and then
derive the final architecture which can simultaneously serve
all three use cases.
A. Network Architecture for AR communication
Generally, AR use case requires the 5G system facilitate the
following three steps of communications.
Registering the AR devices into the network
Establishing data session between the AR device and
image processing server
Data transfer between AR device and image processing
server
Architectural components needed for completing above
steps can be identified based on the message transfer between
each element in 5G system including AR device, Next Gen-
eration NodeB (gNB) and core network functions.
a) Device registration: We define the registration proce-
dure for AR device based on 3GPP specifications [23]. Figure
7 illustrates the message sequence between the entities in
the architecture. AR device initiates the registration process
by sending registration request to gNB. gNB forwards the
request to AMF. After that, AMF and AR device exchange the
identity request and response messages. In the next step, AMF
contacts AUSF for the device authentication. AUSF facilitated
the authentication after contacting UDM and retrieving the
authentication data. Once the authentication data is received
from UDM, AUSF sends the authentication response to AUSF.
Identity request/response messages are transmitted between
AMF and the AR device again. After the identity verification,
AMF then works with PCF for the policy association for the
AR device. Once the policy association is successful, AMF
sends an update to SMF informing the session context. AMF
also sends the registration accept message to the AR device
and the device then sends the registration complete message
to AMF concluding the registration process.
b) Session establishment: After completing the registra-
tion process, AR device has to establish a data session with
the image processing server to enable continuous data transfer.
We define the Protocol Data Unit (PDU) session establishment
procedure between AR device and the image processing server
based on 3GPP specifications [23]. Figure 8 illustrates the
message sequence required for PDU session establishment
process.
Here, AR device initiates the process by sending PDU
session establishment request to AMF via gNB. AMF then
sends a request for a new session creation to SMF. In the next
step, SMF registers with UDM, subsequently UDM stores data
related to the session. After that SMF sends the Response to
AMF. Then, PDU session authentication/authorization process
occurs by exchanging messages between AR device, gNB,
AMF, SMF, UPF and Server. Once this step is completed,
SMF works with PCF for policy association for the ses-
sion. Then UPF and SMF exchange the session establish-
ment/modification request and the respective response. Mes-
sage transfer from SMF to AMF allows AMF to know which
Fig. 7: Message sequence chart for AR device registration
procedure
access towards the AR device to use. AMF then sends the
PDU session ID information to gNB so that gNB can work
with AR device for the gNB specific resource setup. After
that gNB sends the acknowledgement for the PDU session
request to AMF. Based on that, AMF sends request regarding
PDU session update to SMF. SMF then requests UPF for
session modification. Once SMF received the response from
UPF, SMF finally sends the response for PDU Session update
to AMF completing the PDU session establishment process.
c) Data transfer: After successful completion of above
steps, AR device can send a continuous data stream to the
server and retrieve the augmentations sent by the server, so
that the device can overlay them on the camera view. Entities
involved in this data transfer process are AR device, gNB,
UPF and the server. Based on the above steps, 5G network
functions needed to serve the AR use case can be identified
to derive uO architecture depicted in Figure 9.
Network functions that are not used in the architecture
are Network Exposure Function (NEF) which handles the
masking of network and user sensitive information to external
Application Function’s (AF) according to the network policy,
Network Repository Function (NRF) and AF.
B. Network architecture for massive wireless sensor networks
The following steps of communications should be supported
by 5G network operator to serve the massive wireless sensor
networks use case.
Registering the sensors, actuators and servers in to the
5G network
6
Fig. 8: Message sequence chart for session establishment
between AR device and server
Fig. 9: Architectural components of L5GO to serve the AR
use case
In case of anomaly, establishing data connectivity from
the sensor devices to the alarm management server
Transfer the data related to anomaly from sensor to the
management server
Session establishment between the management server
and the controlling instance
Data transfer from management server to the controlling
server
Session establishment between the controlling server and
the actuator
Taking appropriate action on the actuator by transferring
action related data to the actuator
We define the message sequence for the sensor network
communication scenario using 3GPP specifications [23]. Fig-
ure 10 illustrates the message sequence diagram for a typi-
cal sensor network communication. Registration process uses
the comprehensive message sequence illustrated in Figure 7,
where the terminal devices in this case being the sensors,
machines or servers. Then the sensor detects and abnormal
behaviour and communicates it to the alarm management
server. For this to happen, the sensor requires a PDU session
to be established with the alarm management server. This
happens according to the message sequence illustrated in
Figure 8, while sensor being the initiator in this case. Once
the PDU session is established, sensor transfers the data to
the alarm management server. The server then analyses the
severity of the anomaly and establishes a PDU connection with
the controlling server, again using a similar message sequence
illustrated in Figure 8. Once this is done, alarm management
server successfully transfers the data to the controlling server.
Controlling server decides which action to take on which
machine by processing the received data. Then the controlling
server establishes a PDU session with the relevant machine.
Next, controlling server transfers the data to the relevant
machine completing the communication process. Relevant
action will be executed on the machine after this process. The
action could be an emergency stop of a machine, decreasing or
increasing temperature of an air conditioner, sound an alarm.
Fig. 10: Message sequence chart for emergency action based
on sensor detection
Using the message sequence depicted in Figure 10, we
designed the architecture for the L5GO catering to the needs
of sensor network communication. For the device registration
process AN, AMF, PCF, AUSF and UDM network functions
are needed. For the PDU session establishment process AN,
AMF, UPF, SMF, PCF and UDM network functions are
needed. For the data transfer process AN and UPF network
functions are used. Therefore, the architecture should comprise
all the above mentioned network functions to support the
communication. Hence, we can use the network architecture
depicted in Figure 11 to support the sensor network com-
munication, which is the similar to the one we used for AR
7
communication.
Fig. 11: Architectural components of L5GO to serve the
massive wireless sensor networks use case
C. Network architecture for Mobile robots
With the increased usage of mobile robots in future fac-
tories, the robots themselves should have the ability to take
decisions on their own by analyzing the data they receive from
the sensors in the environment, the data they receive from
the other robots and the data they receive from the guidance
control system. Following communication steps will be seen
in Industry 4.0 mobile robots use case.
Registering the environment sensors, mobile robots and
server to the 5G network
Establishing data connectivity between elements such as
mobile robots, guidance control server and sensors
Robots’ regular operation which includes data transfer
between different elements
Perform handovers due to the mobility of robots
Regular operations of mobile robots require registration
in the 5G network and become the network elements of
5G, establishing PDU session between the relevant devices,
data transfer between the devices and seamless mobility. We
already identified the network functions needed for registration
and session establishment under previous two use cases in
Figure 7 and 8. Hence, in this case also, it is mandatory to
have AN, AMF, PCF, AUSF, UDM, UPF and SMF.
We define the message sequence for seamless mobility
based on the handover process in 3GPP specifications [23]
as depicted in Figure 12. The handover process execution is
initiated by the source gNB forwarding the data to the target
gNB. Target gNB then sends path switch request to AMF.
Once AMF receives the path switch request, AMF sends a
session update request to SMF and then SMF sends the session
modification request to UPF. UPF then sends the session
modification response to back to SMF. In the meantime, UPF
sends the end marker packets to source gNB and it also sends
the downlink packets to mobile robot via target gNB. After
that SMF sends the PDU session update response to AMF,
and AMF then acknowledges the path switch request to target
gNB. Then the target gNB sends release resources message to
source gNB completing the handover process. Therefore the
mandatory NFs to cater to handover process are AN, AMF,
SMF and UPF. Hence, we can use the network architecture
depicted in Figure 13 to support the communication of mobile
robots.
Fig. 12: Message sequence chart for handover procedure of
robots
Fig. 13: Architectural components of L5GO to serve the
mobile robots use case
D. Consolidated Architecture
When L5GO uses three network slices to cater to the three
use cases, it must create network slices before any actual
communication happens over a selected slice. To simultane-
ously cater to AR, massive wireless sensor network and mobile
robots use cases, the L5GO needs to create NSIs before the
communication begins. To create the network slices, three net-
work slice management functions CSMF, NSMF and NSSMF
should also be there in the architecture of L5GO. These
three network functions allow the communication service
requirements to be translated to network slice requirements,
and then those network slice requirements to be translated to
network slice subnet requirements and ultimately the network
slice subnets can be created. These network slice subnets are
then used to create the network slice instances.
Apart from that, the best fitting slice must be selected
8
Fig. 14: Proposed architecture of L5GO with three network
slices
before the communication begins. This is done by NSSF,
which is an obligatory element in the architecture of L5GO
which supports communication over multiple network slices.
Combining these concepts, final architectural components to
support AR, massive wireless sensor network and mobile
robots use cases can be derived. Figure 14 depicts the derived
architectural components for L5GO.
V. ARCHITECTURAL OPTIONS
In our architecture, we consider that the factory owns all the
end devices such as AR devices, sensors, robots and servers.
All of them are located within factory premises. It is assumed
that the servers are located in a different cell site within the
factory premises away from other end devices. We consider
several architectural options for the 5G network operator(s)
who serve the communication needs of the use cases.
A. Pure local architecture
First one is the pure local architecture where all NFs are
deployed by the L5GO inside the factory premises. This
includes both the core network which includes all NFs and the
access network. This is direct use of the proposed architecture
in Section IV. The architecture comprising the core network of
L5GO, access network, backhaul connectivity and the terminal
devices is depicted in Figure 15a.
B. MNO architecture
Secondly, we consider MNO architecture where all core
NFs are deployed by MNO and they are located outside the
factory premises at a given distance. The backhaul connects
the factory location and the core network location of the MNO.
This is depicted in Figure 15b.
C. Hybrid architecture
Finally, we consider hybrid architecture options where the
5G communication network is collaboratively deployed by
both L5GO and MNO as depicted in Figure 15c. Both L5GO
and MNO serve the factory communications using their core
networks. A set of NFs are provided by L5GO and the
remaining NFs belong to MNO. We consider that L5GO is
limited in resources and can have only a limited number of
NFs deployed at a given time instant. Assuming L5GO is the
operator with the highest priority, the first MNFs needed in
the communication are deployed by L5GO which is located
on factory premises and all the remaining NFs are deployed
by MNO. Here, Mis the number of NFs deployed at L5GO.
Therefore, the communication may also need the involvement
of a NFs deployed by MNO based on the value of M. It is
evident that the hybrid architecture approaches MNO archi-
tecture when M= 0 and it approaches pure local architecture
when M=Mmax. If 0 <M< M max , then it becomes a hybrid
architecture and in that case, messages between core network
functions may also utilize the backbone network to travel from
L5GO to MNO and vice versa, incurring a higher latency.
We consider different design approaches for these hybrid
architecture deployments based on the idea of which net-
work functions are provided by which operator. One of our
architecture design approaches are based on NF placement
algorithms proposed in [24], which is First Come First Served
(FCFS) based allocation by L5GO. Secondly, we consider
Most common NFs at L5GO (MCNF) and our third approach
is Operator Policy Based Placement (OPBP) of NFs. Finally,
we consider a Predictive Placement (PP) Algorithm similar to
the one proposed in [25].
1) First Come First Served algorithm (FCFS): In this
design approach, outlined in Algorithm 1, NFs are deployed
(a) Pure L5GO (b) MNO (c) Hybrid
Fig. 15: Architectural Options
9
at L5GO using the basis of when they are first used. If a
certain communication happens first, then the NFs needed for
the communication are deployed first. Based on the value of
M, number of NFs deployed at L5GO is decided and the
remaining NFs are deployed at MNO. The communication
procedures involve the device registration procedure (reg),
session establishment procedure (ses), data transfer (dat) and
handover procedure (hnd). Each procedure contains multiple
message transfers and each message has to be considered
in identifying the NFs needed for communications. As an
example, the device registration usually happens first, therefore
NFs needed for message transfers in registration procedure
will be deployed first.
Algorithm 1: FCFS NF Placement
Result: Order of NFs to be deployed at L5GO
Procedure List = {reg, ses, dat, hnd};
for Proc in Procedure List do
for each Message in Proc do
Identify NF needed for the Message transfer;
if NF is already identified then
Discard NF and move to next Message;
end
end
end
2) Most Common NFs at L5GO (MCNF): In the second
design approach outlined in Algorithm 2 we place the most
common NFs to all use cases in L5GO core network while
placing the remaining NFs in MNO core network, irrespective
of the amount traffic flow. This is decided after considering
the message sequence charts for all use cases. As an example
for M= 1, the most common NF is SMF and it is deployed
by L5GO and other NFs are deployed by MNO. When the
number Mincreases, more NF will be deployed by L5GO.
Algorithm 2: MCNF Placement
Result: Order of NFs to be deployed at L5GO
Procedure List = {reg, ses, dat, hnd};
for Proc in Procedure List do
for each Message in Proc do
Calculate Instance Count against each NF;
end
end
Arrange Instance Count in Descending order;
Define Order of NFs;
3) Operator Policy Based Placement (OPBP): In the third
algorithm, we consider the placement of NFs based solely
on network operator policies. This means that the decision
on where to deploy each network function is taken by the
operator based on the external facts. For example, MNO
may decide that certain core network functions should not
be deployed outside MNO premises due to security reasons.
Another example could be to deploy most heavily loaded
NFs locally to reduce congestion at the infrastructure links
and the MNO core network. Deployment cost of NFs at
different places would also be another factor. In the policy
based architecture, we take that the first NF to be deployed at
L5GO is UPF because it is the most used NF while AMF and
SMF are deployed as the last option because MNO prefers
for a centralized control functions. Then we take UDM, PCF
and AUSF based on their usage. Hence in our policy based
algorithm the order of NFs would be UPF, UDM, PCF, AUSF,
AMF and SMF.
4) Predictive Placement algorithm (PP): In the predictive
placement depicted in Algorithm 3, we consider a given period
and for that period we identify the past usage of the NFs.
The period could be hours, days or even a longer duration.
Based on the history data, allocate NFs to L5GO in such
a way that minimum latency is achieved with the minimum
number of NFs at L5GO. This algorithm considers the usage
of a particular NF as well as the frequent interactions between
NFs to come up with an optimized result. This way, most of
the traffic towards the core network is handled by the L5GO
ensuring less traffic transfer to the core network of MNO.
The allocation is re-defined after every period and the present
usage records can be used in the algorithm to take decisions
for future days.
Algorithm 3: Predictive Placement of NFs
Result: Order of NFs to be deployed at L5GO
NF List = {UPF, AMF, SMF, UDM, PCF, AUSF};
M= 1, Mmax = NF count in NF List;
for M < M max do
for each NF in NF List do
Calculate avg. Latency using past Data;
Find NF with Latencymin ;
end
Identify Mth NF to Deploy using Latencymin;
Remove NF with Latencymin from NF List;
end
VI. NUMERICAL ANA LYSIS
We consider three deployment models in our simulations
which uses the architectures explained in Section V. In the
first model, the factory is served by a local 5G network having
pure local architecture. Communication is facilitated by a 5G
network deployed inside factory premises including the core
network. This setup is depicted in Figure 16.
In the second model, we assume that the entire factory is
covered by a 5G network deployed by an MNO. We consider
that the MNO is simultaneously serving a total of Nsuch
factories having similar use cases. Each factory having similar
network setup and similar requirements as seen in Figure 17.
Core network of MNO is located outside the factory. Figure
18 illustrates the MNO based model, serving for the use cases
of a given factory.
Without loss of generality, we have assumed that L5GO’s
processing power is 1/Nof the possessing power of an MNO.
It validates the fact that MNO’s resources are equally divided
among NL5GOs serving each factory.
10
Fig. 16: Deployment model for pure local operator serving the
factory use cases
Fig. 17: MNO’s service for Nfactories having the three use
case
The third model represents the hybrid architecture where
both L5GO and MNO simultaneously contribute to form the
architecture, which is depicted in Figure 19.
For all pure local, MNO and hybrid models, we simulate
the entire factory operation comprising of all three use cases
AR, massive wireless sensor networks and mobile robots. We
assume all three use cases are simultaneously occurring. All
the simulations are carried out using MATLAB [26] and the
latency results are averaged over 100 iterations. Simulations
are carried out for a single day period of factory operation.
To model a real factory operation, we assumed the frequency
of occurrence of each use case during a simulation period. If
the use cases are served by L5GO, then there is no external
traffic from other factories on the core NFs but in the case of
MNO, the NFs are loaded with the traffic from other N1
factories. In the hybrid architecture, only the NFs located at
MNO are subjected to external traffic. Since we have a slice
based architecture for each use case, the traffic of one use case
is not interfered by the traffic of other two use cases.
The average latency of the system comprising all use cases
can be expressed as follows.
Lavg =Ltot
Ptot
(1)
Fig. 18: Deployment model for MNO serving the factory use
cases
Fig. 19: Hybrid deployment model to serve factory use cases
where Ltot is the aggregated latency of all packets and Ptot
is the count of all data packets transferred during a day. For
a factory having AR, massive wireless sensor networks and
mobile robots use cases, Ltot can be expressed as
Ltot =LAR +LS+LR(2)
where LAR is the aggregated latency of all AR packets, LSis
the aggregated latency of all sensor network packets and LR
is the aggregated latency of all packets from robots during a
day. Ptot can be expressed as
Ptot =PAR +PS+PR(3)
where PAR is AR data packet count, PSis sensor data packet
count and PRis data packet count from robots, during a day.
Simulations for each use case are described below along
with the respective formulas for latency. For the AR use case,
we derive all the formulas and extend those formulas to other
two use cases.
1) Augmented Reality: For AR use case, we assume the
properties of AR device deciding the parameters of the video
stream. We take the resolution of AR device screen as 400
x 240 with a frame rate of 20 frames per second. We take
that the AR device captures 8-bit color video. We assume that
the video transmission is uncompressed due to the processing
power limitations at the AR device. We take the average
duration of video is 10 s for the simulation purpose, even
though it may be longer in reality. The device registers with
11
the network as soon as it turned on each day and we assume
this happens only once a day. Whenever there is a need to
use the AR device by a factory worker, it establishes a data
connection with the server and then transfers data. Number of
such data transfers are decided based on the frequency of AR
sessions per day.
Mathematical expression for the aggregated AR latency
LAR will be
LAR =Lreg +Nses .(Lses +Ldat)(4)
where Lreg is the aggregated latency of registration process,
Lses is the aggregated latency of one session establishment and
Ldat is the aggregated latency of one data transfer process.
Nses is the average number of AR sessions per day. Lreg,
Lses and Ldat are latency components belong to AR com-
munications. Those latency components include the latency of
5G access network, gNB, access router, backhaul, core router
and NF processing delay. Therefore Lreg for L5GO can be
expressed as follows.
Lreg lo =k1.Tacc +k2.Tg nb +k3.Trouter acc+
k4.Tback lo +k5.Trouter lo +k6.TNF lo
(5)
where Tacc is the mean delay from AR device to AN, Tgnb
is the mean delay at gNB due to processing and queuing,
Trouter acc is the mean queuing delay at the access router,
Tback lo is the delay of the backhaul network from access
router to core router, Trouter lo, is the mean delay at the
L5GO’s core router and TNF lo is the mean NF processing
delay of L5GO. k1,k2,k3,k4,k5,k6are the the number
of times where the registration message passes through each
respective element.
Average queuing delay of elements such as routers can be
expressed with their packet arrival rate and service rate. We
model the router queues as M/M/1 queues and the average
delay at any router Trouter xx can be expressed as
Trouter xx =1
µλ(6)
where λbeing the arrival rate while µbeing the service rate.
Since we compare the performance of the pure local, MNO
and hybrid architectures, first three terms of equation (5) are
common and can be considered as a constant kreg . Therefore
equation (5) can be re-written as
Lreg lo =kr eg +k4.Tback lo +k5.Trouter lo +k6.TNF lo (7)
Similarly, we can write the expressions for Lses lo and
Ldat lo and finally, the aggregated latency LAR for L5GO can
be expressed as
LAR lo =aAR +a1.Tback lo +a2.Trouter lo +a3.TNF lo (8)
where aAR,a1,a2,a3are constants applicable for M=
0. These constants are taken to represent the entire AR
communication in one day, therefore Nses is also integrated
into these constants.
For MNO, equation (8) can be used with minor modifica-
tions as in (9).
LAR mno =aAR +a1.Tback mno+
a2.Trouter mno +a3.TNF mno
(9)
For hybrid architecture, both L5GO and MNO cores are
involved and the latency of AR use case can be expressed as
LAR hybrid =aAR +a11 .Tback lo +a12 .Tback mno +
a21 .Trouter lo +a22 .Trouter mno+
a31 .TNF lo +a32 .TNF mno
(10)
where a11,a12 ,a21,a22 ,a31,a32 are being the number of
times where different messages are processed at respective
element. When M= 0, equation (10) simplifies to equation
(9) and when M=Mmax equation (10) simplifies to equation
(8).
aAR is constant for AR use case for all architectures while
aij values depend two factors. 1. the number of NFs hosted at
L5GO (M) and 2. which NFs are hosted for that M. Therefore,
deriving a direct relationship between LAR hybrid and Mis
not possible and the latency results should be taken using the
simulations for each M.
We can express Tback mno using the distance to core net-
work Dback mno as follows.
Tback mno =a .Dback mno (11)
where ais a constant represents per km latency over fiber
backhaul channel.
To achieve the given E2E AR latency defined in Section
III-A, for a known M, we can identify Dback mno in a hybrid
architecture using equation (10) and (11).
2) Massive wireless sensor networks: For sensor network
communications, we assume that the sensors are working
24 x 7 and device registration to the network is already
completed. Whenever there is a need to transfer sensor data
and take appropriate action, the message flow occurs according
to Figure 10. It includes three session establishments and three
data transfers between different entities.
Similar to the AR use case, we can derive expressions for
LS lo,LS mno LS hy brid as follows.
LS lo =bS+b1.Tback lo +b2.Trouter lo +b3.TNF lo (12)
LS mno =bS+b1.Tback mno +b2.Trouter mno +
b3.TNF mno
(13)
LS hy brid =bS+b11 .Tback lo +b12 .Tback mno+
b21 .Trouter lo +b22 .Trouter mno+
b31 .TNF lo +b32 .TNF mno
(14)
where bSis constant for massive wireless sensor network use
case for all architectures while bij values have the similar
interpretation as in AR use case. For a given M, to identify
Dback mno satisfying a given latency in any event of sensor
networks use case, equation (11) can be used with equation
(14), similar to the way it is used in AR use case.
12
3) Mobile robots: For the mobile robot simulations, we
consider the devices are switched on at the beginning of each
day making device registration process mandatory for each
robot. Session establishments are data transfer processes occur
based on the frequency of robot usage. Handovers due to the
mobility of robots are also simulated with a given frequency.
Similar to the AR and sensor network use cases, we can
express LR lo,LR mno LR hybrid for mobile robots use case
as follows. Here, latency due to handover in also included in
these formulas.
LR lo =cR+c1.Tback lo +c2.Trouter lo +c3.TNF lo
(15)
LR mno =cR+c1.Tback mno +c2.Trouter mno+
c3.TNF mno
(16)
LR hybrid =cR+c11 .Tback lo +c12 .Tback mno +
c21 .Trouter lo +c22 .Trouter mno+
c31 .TNF lo +c32 .TNF mno
(17)
where cRis constant for mobile robots use case for all
architectures while cij values have the similar interpretation
as in previous use cases.
Table II outlines the general simulation parameters. Latency
of Access Network (AN) is based on the 3GPP study on next
generation access technologies [27] and we assume that access
networks of both MNO and L5GO have similar properties. We
take backhaul as a fiber connection and the latency parameters
are selected based on a study of 5G backhaul challenges [28].
We assume that the gNB supports packet forwarding at 1 Gbps
which provides a service time of 0.5 s. Similarly, we consider
the access router has 1 Gbps interfaces while core routers
having 10 Gpbs interfaces, providing 0.05 ms of service time.
This approach is motivated by [29]. For the service time of
network functions, we assume that the operations require a
certain number of CPU cycles [30] and the computing capacity
at each network function is designed to handle the loads
proportionally. Therefore, we take them as 1 ms for L5GO
and 0.1 ms for MNO. The same concept is used for taking the
processing delay of servers.For each experiment, we measure
the E2E latency of the communications.
VII. SIMULATION RESULTS ANALYS IS
In this section, we present the simulation results for pure
local deployment, MNO deployment and hybrid architecture
deployments. First we consider each use case separately and
analyze the latency. This is possible because each use case
is served by a different network slice. Under the hybrid
architecture, the performance of four algorithms are discussed
while varying Mfrom 0 to Mmax. We keep the distance to core
network constant at 250 km and identify the best algorithm to
be used to deploy NFs. We also consider an average latency
for a factory comprising all three use cases. Secondly, we use
the best algorithm selected at the first step, vary the distance to
core network and identify at which core network distance each
TABLE II: General simulation parameters
Parameter Value
Parameters common to all use cases
Number of factories served by MNO (N) 10
Simulation duration for factory (per day) 8 working hours
Distance to L5GO core network 500 m
Distance to MNO core network 250 km
Latency between terminal devices and AN 0.5 ms [27]
Latency between AN and core network 0.05 ms/km [28]
Mean service time of gNB 0.5 ms [29]
Mean service time of access router 0.5 ms [29]
Mean service time of L5GO core router 0.05 ms [29]
Mean service time of MNO core router 0.05 ms [29]
Mean service time of NF at L5GO 1 ms [30]
Mean service time of NF at MNO 0.1 ms [30]
Use case 1 - AR
Mean video duration 10 seconds
Size of UDP packet from AR device 64 kB
Frequency of AR sessions every 30 minutes
AR server processing delay 1 ms [30]
Number of AR devices operating in parallel 3 [7]
Use case 2 - Sensor networks
Data packets needed for sensor data 10
Frequency of sensor data transfer every 10 minutes
Alarm server processing delay 1 ms [30]
Action server processing delay 1 ms [30]
Processing delay of the machine 1 ms [30]
Number of sensors operating in parallel 20 [7]
Use case 3 - Mobile robots
Data packets needed for robot data 16
Frequency of robot data transfer every 5 minutes
Frequency of robot handover every 30 minutes
Processing delay of the robot 1 ms [30]
Number of robots operating in parallel 10 [7]
event of the use cases can be satisfied. Finally, we formulate
a relationship between the number of NFs needed to deploy
locally with respect to the MNO core network distance so that
the required latencies are satisfied.
A. Analysis for a fixed MNO distance
1) Augmented Reality: Simulation result for the average
latency of pure local architecture is 4.77 ms and for MNO
based architecture, it is 29.29 ms. This means that the pure uO
architecture can satisfy the required one way E2E latency of 10
ms [7]. For the hybrid architectures, the average latency will
vary based on which NFs are deployed at the L5GO. Different
algorithms will provide different set of NFs to be deployed at
L5GO, therefore the average latency of the architecture derived
from different algorithms may vary even for the same M.
a) FCFS algorithm: To identify the NF which is needed
first, it is fair to assume that the device registrations oc-
cur before the session establishments and data transfers in
daily factory operations. Therefore, under this algorithm, NFs
needed for device registration will be deployed at the L5GO’s
core network first. Table III outlines the NFs which will be
implemented at L5GO under this algorithm for each M.
The average latency of AR packets under FCFS algorithm
is depicted in Figure 20 with 95% confidence intervals. La-
tency of pure local architecture and MNO architecture is also
depicted in Figure 20 for reference. It is clearly observed that
when M= 0, latency of hybrid architecture approaches MNO
13
TABLE III: NFs deployed at L5GO under FCFS algorithm
Number of NFs (M) NFs
0 –
1 AMF
2 AMF, AUSF
3 AMF, AUSF, UDM
4 AMF, AUSF, UDM, PCF
5 AMF, AUSF, UDM, PCF, SMF
6 AMF, AUSF, UDM, PCF, SMF, UPF
0123456
Number of Network Functions at L5GO
0
5
10
15
20
25
30
35
End-to-End Latency
MNO
L5GO
FCFS
Fig. 20: Average AR packet latency under FCFS algorithm
latency while M=Mmax, hybrid architecture takes the latency
of pure local architecture.
The key observation is, for certain Mvalues, the latency
is even higher than MNO latency. Reason for this can be
explained as follows. If two NFs have number of message
transfers between them, then locating one of those functions at
L5GO and the other function at MNO will increase the latency
because every time when these NFs wants to communicate,
the backhaul connection is used. In the message sequence
diagrams, we observed that there are number of interactions
between AMF and SMF. Since AMF is the first NF to be
deployed at L5GO, until SMF is deployed at M= 5, latency
is higher than MNO latency. Second observation is that there
is no significant latency reduction until UPF is deployed at
L5GO. This is due to the fact that, UPF is the most utilized
NF in AR communications which is used for data transfers
(user plane data) and it has comparatively less interactions with
other NFs during the data transfer. Overall, the architecture
derived from FCFS algorithm does not provide better latency
results until M= 6.
b) MCNF algorithm: In the factory communications,
certain NFs are used more frequently than the others. MCNF
algorithm utilizes this fact and deploys most common NFs at
L5GO. Most common NFs are decided after considering all
three use case communications. Table IV outlines the order
which the NFs are assigned under each M.
The average latency of AR packets is depicted in Figure 21
along with the MNO and L5GO latency. Significant reduction
in latency is observed when UPF is deployed at L5GO for M
= 3, the same observation with UPF under FCFS algorithm.
c) Operator Policy Based Placement: Allocation of NFs
for operator policy based placement under each Mis listed in
Table V. The basis for this was described in Section V-C3.
TABLE IV: NFs deployed at L5GO under MCNF algorithm
Number of NFs (M) NFs
0 –
1 SMF
2 SMF, AMF
3 SMF, AMF, UPF
4 SMF, AMF, UPF, UDM
5 SMF, AMF, UPF, UDM, PCF
6 SMF, AMF, UPF, UDM, PCF, AUSF
0123456
Number of Network Functions at L5GO
0
5
10
15
20
25
30
35
End-to-End Latency
MNO
L5GO
MCNF
Fig. 21: Average AR packet latency under MCNF algorithm
Average AR latency under OPBP is depicted in Figure 22.
As expected, a significant improvement in latency is achieved
at M= 1 by deploying UPF by L5GO. After that, latency has
no specific pattern because the main focus of this algorithm
is not to achieve minimum latency, but to follow the policy.
0123456
Number of Network Functions at L5GO
0
5
10
15
20
25
30
35
End-to-End Latency
MNO
L5GO
OPBP
Fig. 22: Average AR packet latency under OPBP
d) Predictive Placement algorithm: Predictive placement
algorithm utilizes available information on NF usage of the
factory use cases and tries to minimize the latency with
minimum M. Simulation results of FCFS, MCNF and OPBP
algorithms showed that deploying UPF at L5GO contributes
in significant reductions of latency. Next two NFs should be
AMF and SMF because they are the next most used NFs in
these communications. The rest of the functions will be UDM,
PCF and AUSF respectively. Table VI shows the order of NFs
14
TABLE V: NFs deployed at L5GO under OPBP
Number of NFs (M) NFs
0 –
1 UPF
2 UPF, UDM
3 UPF, UDM, PCF
4 UPF, UDM, PCF, AUSF
5 UPF, UDM, PCF, AUSF, AMF
6 UPF, UDM, PCF, AUSF, AMF, SMF
deployed at L5GO with respect to M. Generic algorithm used
in obtaining the order of NFs is explained in Section V-C4.
TABLE VI: NFs deployed at L5GO under PP algorithm
Number of NFs (M) NFs
0 –
1 UPF
2 UPF, AMF
3 UPF, AMF, SMF
4 UPF, AMF, SMF, UDM
5 UPF, AMF, SMF, UDM, PCF
6 UPF, AMF, SMF, UDM, PCF, AUSF
Average latency results of AR use case under predictive
algorithm is depicted in Figure 23. As expected, a significant
improvement in latency is achieved at M= 1 by deploying
UPF at L5GO. The next improvement is observed when
both AMF and SMF are at L5GO when M= 3. Finally, it
approaches pure local architecture latency when M=Mmax.
0123456
Number of Network Functions at L5GO
0
5
10
15
20
25
30
35
End-to-End Latency
MNO
L5GO
PP
Fig. 23: Average AR packet latency under PP algorithm
Latency performance of hybrid architectures obtained using
all the algorithms for AR use case is shown in Figure 24.
Since AR use case has significant data transfers in user plane,
latency is heavily reduced once UPF is deployed locally. It
is clear that the predictive placement algorithm derives the
architecture which provides the best performance for AR use
case with respect to latency.
2) Massive Wireless Sensor Networks: Simulation results
for the massive wireless sensor networks use case under all
algorithms are depicted in Figure 25 with 95% confidence
intervals. As we can see, FCFS algorithm provides higher
latency than MNO until M= 4. In the sensor networks use
case, there are multiple session establishments and SMF is
heavily involved in these session establishments with terminal
Fig. 24: Average packet latency of AR use case
devices. Under FCFS algorithm, SMF is deployed locally at
M= 5, therefore there is a significant deduction in latency
is observed for M= 5. Under MCNF algorithm, even though
SMF gets deployed locally first, it makes AMF and SMF being
deployed in two locations. This results of an increase in latency
due to the communications between AMF and SMF. Therefore
latency is higher than MNO with M= 1. Since the most used
NFs (SMF, AMF and UPF) are deployed locally at M= 3
onwards, latency gets significantly lower. In general,predictive
algorithm provides better latency with minimum Min this use
case also, because deploying UPF locally at M= 1 will help
low latency data transfers, similar to the behavior of MCNF
from M= 3 onwards.
There is a significant difference in latency curves between
AR use case and sensor networks use case. In AR use case,
UPF is heavily used because of the high data transfers. There-
fore deploying UPF locally contributes to a drastic decrease in
latency as the data stream does not use the backhaul network to
MNO. However, for sensor networks, a data transfer happens
only after a session establishment process. There is no specific
NF which has a significantly higher usage than the other.
Therefore, drastic reductions are not observed.
0123456
Number of Network Functions at L5GO
0
20
40
60
80
100
120
140
160
180
200
End-to-End Latency
MNO
L5GO
FCFS
MCNF
OPBP
PP
Fig. 25: Average packet latency of sensor networks use case
15
3) Mobile Robots: For the mobile robots use case, sim-
ulation results for the architectures derived from all the
algorithms are depicted in Figure 26 with 95% confidence
intervals. Results show that the FCFS algorithm provides the
worst performance and policy based placement gives slightly
better performance than FCFS. Mobile robot communications
include a number of handovers. Because of this additional
handover procedure, AMF and SMF are utilized heavily than
the other use cases. Under FCFS, AMF is deployed locally
at M= 5, therefore latency starts to reduce at M= 5. under
policy based algorithm, both AMF and AMF are deployed
last therefore it does not provide low latency performance
until M= 6. MCNF algorithm provides comparatively better
performance for M= 3 and beyond. Under MCNF algorithm
AMF, SMF and UPF are deployed when M= 3 and most of
the message transfers happens between L5GO’s core network
and the factory. A slightly better performance is seen when
predictive algorithm is used. In general, mobile robots use
case needs AMF, SMF and UPF to be deployed locally to
achieve better latency performance.
Behavior of latency for mobile robots use case differs from
both AR and sensor networks use cases because of the way
each NF is used for the communication. A heavy usage of
UPF is not seen in mobile robots use case like in AR use
case. Also, deploying one NF locally does not result in better
performance as seen in sensor networks use case. However,
when the number of NFs deployed locally is three, far better
performance is seen under MCNF and PP algorithms.
0 1 2 3 4 5 6
Number of Network Functions at L5GO
0
10
20
30
40
50
60
70
End-to-End Latency
MNO
L5GO
FCFS
MCNF
OPBP
PP
Fig. 26: Average packet latency of mobile robots use case
4) Average for all three use cases: Considering a factory
environment having all three use cases, the average packet
latency can be calculated using equation (1) and the results
are illustrated in Figure 27 with 95% confidence intervals.
Considering the latency results of each use case and the
average of all three use cases depicted in Figure 27, in general
we can take the architecture derived from predictive algorithm
is the best architecture to serve the factory. For the remaining
analysis, we consider only the architectures derived using
predictive placement algorithm.
0 1 2 3 4 5 6
Number of Network Functions at L5GO
0
10
20
30
40
50
60
End-to-End Latency
MNO
L5GO
FCFS
MCNF
OPBP
PP
Fig. 27: Average packet latency combining all three use cases
B. Latency with respect to MNO distance
1) Augmented Reality: So, far we have considered a fixed
250 km as the distance to MNO core network. However in
reality, the core network distance (d) could differ. Therefore,
we take several values for dand observe how the deployment
of NFs in a hybrid architecture affects the latency performance
of these industry 4.0 use cases. We keep dat 50 km, 100
km, 250 km and 500 km, and observe how many network
functions that a hybrid operator should deploy locally to cater
to the requirements. We consider only the predictive placement
algorithm for this analysis as it is the most efficient algorithm
to achieve low latency.
Latency requirements of each use case were highlighted
in Section III and we use those values for the analysis. As
an example E2E latency of AR packets should be 10 ms.
Results of the analysis for AR use case is depicted in Figure
28 alongside with the required AR latency.
0123456
Number of Network Functions at L5GO
0
10
20
30
40
50
60
End-to-End Latency
AR E2E Latency
L5GO
Hybrid with MNO at 500 km
Hybrid with MNO at 250 km
Hybrid with MNO at 100 km
Hybrid with MNO at 50 km
MNO at 500 km
MNO at 250 km
MNO at 100 km
MNO at 50 km
Fig. 28: Average packet latency of AR use case under PP
algorithm
We can see that if the core network of MNO is at 50 km,
MNO itself can satisfy the required AR E2E latency (10 ms).
16
However, if the core network is at 100 km, MNO cannot satisfy
the requirement but a hybrid operator with UPF deployed
locally would be able to serve. Since AR communication
utilizes UPF heavily than other NFs due to its high data
transfer needs, deploying UPF locally would be enough to
satisfy the required AR E2E latency as seen in d= 250 km
and d= 500 km curves.
2) Massive wireless sensor networks: The same analysis
was carried out for massive wireless sensor networks use case
and the results are depicted in Figure 29. Latency requirement
of sensor network use case was highlighted in Section III-B.
For the event based monitoring the requirement is between
50 ms and 1 s, therefore we consider the mean 475 ms. For
condition monitoring for safety, the requirement is between 5
ms and 10 ms therefore we take it as 7.5 ms, again using the
mean.
0123456
Number of Network Functions at L5GO
0
50
100
150
200
250
300
350
400
450
500
End-to-End Latency
Safety
Interval/Event Based
L5GO
Hybrid with MNO at 500 km
Hybrid with MNO at 250 km
Hybrid with MNO at 100 km
Hybrid with MNO at 50 km
MNO at 500 km
MNO at 250 km
MNO at 100 km
MNO at 50 km
Fig. 29: Average packet latency of massive wireless sensor
networks use case under PP algorithm
It is seen from the results that for interval and event based
requirements can be catered by MNO whereas the latency
requirement of condition monitoring for safety cannot be
catered even by L5GO. This is because of the number of
session establishments and data transfers involved in the sensor
networks communications. One solution for further reduce the
latency is to keep the sessions established at all times between
the terminal devices such as sensors and servers. However,
additional concerns like energy consumption will arise with
continuous session establishments. Therefore, Keeping the
sessions established only for critical sensors is a possible
solution.
3) Mobile robots: We take the required latencies for each
event in mobile robots based on Section III-C, 30 ms for
cooperative driving of robots, 55 ms for video operated remote
control.
Figure 30 depicts the latency performance of each event
under mobile robots use case. For cooperative driving, when
dis 100 km or less, MNO can satisfy the latency requirement,
but when the core network is at 150 km it needs at least 1 NF
deployed locally to meet the latency requirements. When dis
200 km and beyond, it needs at least 3 NFs. Similar analysis
0 1 2 3 4 5 6
Number of Network Functions at L5GO
0
20
40
60
80
100
120
End-to-End Latency
Cooperative Driving
Remote Control
L5GO
Hybrid with MNO at 500 km
Hybrid with MNO at 250 km
Hybrid with MNO at 100 km
Hybrid with MNO at 50 km
MNO at 500 km
MNO at 250 km
MNO at 100 km
MNO at 50 km
Fig. 30: Average packet latency of mobile robots use case
under PP algorithm
is possible for remote control of robots. Results reveal that
when dis high, it needs more NFs deployed locally to satisfy
the required latency.
Figure 31 summarizes the results of the analysis by rep-
resenting how many NFs are needed locally for each MNO
distance so that it can cater to each event of the use cases.
In this case, we vary dfrom 50 km to 500 km in 50 km
intervals to obtain better results. In general, when the distance
is increasing more NFs needs to be deployed locally and it
depends on the nature of the use case communications.
50 100 150 200 250 300 350 400 450 500
Distance to MNO Core Network
0
1
2
3
4
5
6
Number of NFs needed to Deploy Locally
Incident/Interval based Monitoring
Augmeted Reality
Corporative Driving of Robots
Remote Control of Robots
Fig. 31: Number of NFs needed with respect to d
VIII. CONCLUSIONS
The novel concept of local 5G operator enables a versa-
tile set of stakeholders to operate 5G networks within their
premises with a guaranteed quality and reliability to comple-
ment traditional Mobile Network Operator (MNO) offerings.
In this paper, we analyzed the feasibility and performance
advantage of using local 5G operators instead of a traditional
MNO in a smart factory environment which supports industry
4.0 standards.
17
We proposed two architecture options called pure local
architecture and hybrid architecture based on 5G standards
which is customized for factory environment. The architectures
were discussed in terms of 5G core network functions and the
number of core network functions that are deployed locally
instead of deploying them at MNO core network. To realize
the conceptual design, we conducted several experiments for
an Augmented Reality (AR), massive wireless sensor networks
and mobile robots use cases which will be heavily used in
future factories.
The experiments revealed that a local 5G network estab-
lished within the factory premises can provide low end-to-end
latency for each use case compared to an MNO provided 5G
network, where the core network is located outside the fac-
tory premises. The end-to-end latency of the communication
exhibits a significant increase over the distance between the
core network and the factory. In a hybrid architecture where
both local 5G operator and MNO collaboratively establish the
core network, the decision on which network functions to be
deployed locally depends on the nature of the use case(s).
The most efficient way to achieve low latency with minimal
network functions deployed locally is to analyze the past
usage data and decide the network functions, as done in the
predictive placement algorithm. When the collaborating MNO
is located close to the factory environment, a given latency can
be achieved with few network functions. When the distance
to MNO core network is increasing, more network functions
needs to be deployed locally to achieve the same latency. In
a pure local 5G operator served factory, the data stream stays
within the factory premises because the core network is located
inside a confined environment. This ensures improved secure
communication between terminal devices and the servers.
In future, we consider more network function placement
algorithms to derive better hybrid architecture options. More-
over, we consider the analysis of power profile and manage-
ment of IoT devices under the use cases served by different
architectures.
ACK NOW LE DG EM EN T
This work is supported by Business Finland in uO5G and
MOSSAF projects, Academy of Finland in 6Genesis Flagship
(grant no. 318927) and 5GEAR (Grant No. 319669) projects,
and European Union in RESPONSE 5G (Grant No: 789658)
project.
REFERENCES
[1] S. E. Elayoubi, M. Fallgren, P. Spapis, G. Zimmermann, D. Mart´
ın-
Sacrist´
an, C. Yang, S. Jeux, P. Agyapong, L. Campoy, Y. Qi et al.,
“5G Service Requirements and Operational Use Cases: Analysis and
METIS II Vision,” in 2016 European Conference on Networks and
Communications (EuCNC). IEEE, 2016, pp. 158–162.
[2] European Commission, “Strategic Spectrum Roadmap towards 5G for
Europe: RSPG Second Opinion on 5G Networks,” Radio Spectrum
Policy Group (RSPG), RSPG18-005,2018.
[3] M. Matinmikko, M. Latva-Aho, P. Ahokangas, S. Yrj ¨
ol¨
a, and
T. Koivum¨
aki, “Micro Operators to boost Local Service Delivery in 5G,
Wireless Personal Communications, vol. 95, no. 1, pp. 69–82, 2017.
[4] M. Matinmikko-Blue and M. Latva-aho, “Micro Operators Accelerating
5G Deployment,” in Industrial and Information Systems (ICIIS), 2017
IEEE International Conference on. IEEE, 2017, pp. 1–5.
[5] A. Prasad, Z. Li, S. Holtmanns, and M. A. Uusitalo, “5G Micro-Operator
Networks-A key enabler for New Verticals and Markets,” in 2017 25th
Telecommunication Forum (TELFOR). IEEE, 2017, pp. 1–4.
[6] Y. Siriwardhana, P. Porambage, M. Liyanage, J. S. Walia,
M. Matinmikko-Blue, and M. Ylianttila, “Micro-operator driven
Local 5G Network Architecture for Industrial Internet,” arXiv preprint
arXiv:1811.04299, 2018.
[7] 3GPP, “Study on Communication for Automation in Ver-
tical Domains (CAV),” Technical Report, Dec 2018. [On-
line]. Available: https://portal.3gpp.org/desktopmodules/Specifications/
SpecificationDetails.aspx?specificationId=3187
[8] J. G. Andrews, S. Buzzi, W. Choi, S. V. Hanly, A. Lozano, A. C. Soong,
and J. C. Zhang, “What will 5G be?” IEEE Journal on selected areas
in communications, vol. 32, no. 6, pp. 1065–1082, 2014.
[9] M. Shafi, A. F. Molisch, P. J. Smith, T. Haustein, P. Zhu, P. Silva,
F. Tufvesson, A. Benjebbour, and G. Wunder, “5G: A Tutorial Overview
of Standards, Trials, Challenges, Deployment, and Practice,IEEE
Journal on Selected Areas in Communications, vol. 35, no. 6, pp. 1201–
1221, 2017.
[10] M. Series, “IMT Vision–Framework and Overall Objectives of the Future
Development of IMT for 2020 and Beyond,Recommendation ITU, pp.
2083–0, 2015.
[11] M. Liyanage, I. Ahmad, A. B. Abro, A. Gurtov, and M. Ylianttila, A
Comprehensive Guide to 5G Security. John Wiley & Sons, 2018.
[12] 3GPP, “System Architecture for the 5G Sys-
tem,” Technical Specification, June 2019. [On-
line]. Available: https://portal.3gpp.org/desktopmodules/Specifications/
SpecificationDetails.aspx?specificationId=3144
[13] S. Sezer, S. Scott-Hayward, P. K. Chouhan, B. Fraser, D. Lake,
J. Finnegan, N. Viljoen, M. Miller, and N. Rao, “Are we Ready
for SDN? Implementation Challenges for Software-Defined Networks,”
IEEE Communications Magazine, vol. 51, no. 7, pp. 36–43, 2013.
[14] P. Ahokangas, M. Matinmikko-Blue, S. Yrj¨
ol¨
a, V. Sepp¨
anen, H.
H¨
amm¨
ainen, R. Jurva, and M. Latva-aho, “Business Models for Local
5G Micro Operators,” IEEE International Symposium on Dynamic
Spectrum Access Networks, 22-25 October 2018, Soul, South Korea.
[15] M. Wollschlaeger, T. Sauter, and J. Jasperneite, “The future of Industrial
Communication: Automation Networks in the Era of the Internet of
Things and Industry 4.0,” IEEE Industrial Electronics Magazine, vol. 11,
no. 1, pp. 17–27, 2017.
[16] H. Lasi, P. Fettke, H. Kemper, T. Feld, and M. Hoffmann, “Industry
4.0.” Business & Information Systems Engineering, vol. 6, no. 4, pp.
239–242, 2014.
[17] A. Varghese and D. Tandur, “Wireless Requirements and Challenges in
Industry 4.0,” in Contemporary Computing and Informatics (IC3I), 2014
International Conference on. IEEE, 2014, pp. 634–638.
[18] X. Li, D. Li, J. Wan, A. V. Vasilakos, C.-F. Lai, and S. Wang, “A Review
of Industrial Wireless Networks in the Context of Industry 4.0,Wireless
Networks, vol. 23, no. 1, pp. 23–41, 2017.
[19] NGMN Alliance, “Description of Network Slicing Concept,” NGMN 5G
P, vol. 1, 2016.
[20] H. Zhang, N. Liu, X. Chu, K. Long, A.-H. Aghvami, and V. C. Leung,
“Network Slicing based 5G and Future Mobile Networks: Mobility, Re-
source Management, and Challenges,” IEEE Communications Magazine,
vol. 55, no. 8, pp. 138–145, 2017.
[21] 3GPP, “Study on Management and Orchestration of Network
Slicing for Next Generation Network,” Technical Report, June
2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/
Specifications/SpecificationDetails.aspx?specificationId=3091
[22] C.-Y. Chang, N. Nikaein, O. Arouk, K. Katsalis, A. Ksentini, T. Turletti,
and K. Samdanis, “Slice Orchestration for Multi-Service Disaggregated
Ultra-Dense RANs,” IEEE Communications Magazine, vol. 56, no. 8,
pp. 70–77, 2018.
[23] 3GPP, “Procedures for the 5G System,” Technical Specification,
June 2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/
Specifications/SpecificationDetails.aspx?specificationId=3145
[24] S. Clayman, E. Maini, A. Galis, A. Manzalini, and N. Mazzocca,
“The Dynamic Placement of Virtual Network Functions,” in 2014 IEEE
network operations and management symposium (NOMS). IEEE, 2014,
pp. 1–9.
[25] B. Malet and P. Pietzuch, “Resource Allocation across Multiple Cloud
Data Centres,” in Proceedings of the 8th International Workshop on
Middleware for Grids, Clouds and e-Science. ACM, 2010, p. 5.
[26] “MATLAB.” [Online]. Available: https://www.mathworks.com/
[27] 3GPP, “Study on Scenarios and Requirements for Next
Generation Access Technologies,” Technical Report, September
18
2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/
Specifications/SpecificationDetails.aspx?specificationId=2996
[28] M. Jaber, M. A. Imran, R. Tafazolli, and A. Tukmanov, “5G Backhaul
Challenges and Emerging Research Directions: A Survey,IEEE Access,
vol. 4, pp. 1743–1766, 2016.
[29] R. Ramaswamy, N. Weng, and T. Wolf, “Characterizing Network Pro-
cessing Delay,” in IEEE Global Telecommunications Conference, 2004.
GLOBECOM’04., vol. 3. IEEE, pp. 1629–1634.
[30] J. Ren, Y. He, G. Huang, G. Yu, Y. Cai, and Z. Zhang, “An Edge-
computing based Architecture for Mobile Augmented Reality,” IEEE
Network, vol. 33, no. 4, pp. 162–169, 2019.
Yushan Siriwardhana is currently a Doctoral
degree student in the Centre for Wireless Communi-
cations, University of Oulu, Finland. He received the
bachelor’s degree in electronics and telecommunica-
tion engineering from the University of Moratuwa,
Sri Lanka, in 2009 and the master’s degree in
wireless communication engineering from the Uni-
versity of Oulu, Finland, in 2019. He has over 7
years experience in telecommunication industry. His
research interests include 5G, Local 5G and future
networks, Security in IoT, MEC.
Pawani Porambage Pawani Porambage is a post-
doctoral researcher at the Centre for Wireless Com-
munications, University of Oulu, Finland. She re-
ceived her doctoral degree on Communications En-
gineering from University of Oulu, Finland. She
obtained MSc and BSc degrees respectively from
University of Nice Sophia-Anipolis, France (2012)
and University of Moratuwa, Sri Lanka (2010). Her
main research interests include lightweight security
protocols, blockchain, security and privacy on IoT
and MEC, Network Slicing and Wireless Sensor
Networks.
Mika Ylianttila (M. Sc, Dr.Sc, eMBA) is a full-
time associate professor (tenure track) at the Cen-
tre for Wireless Communications (CWC), at the
Faculty of Information Technology and Electrical
Engineering (ITEE), University of Oulu, Finland.
He is leading a research team and is the direc-
tor of communications engineering doctoral degree
program. Previously he was the director of Center
for Internet Excellence (2012-2015), vice director
of MediaTeam Oulu research group (2009-2011),
and professor (pro tem) in computer science and
engineering, and director of information networks study programme (2005-
2010). He received his doctoral degree on Communications Engineering at the
University of Oulu in 2005. He has coauthored more than 150 international
peer-reviewed articles. His research interests include edge computing, network
security, network virtualization and software-defined networking. He is a
Senior Member of IEEE, and Editor in Wireless Networks journal.
Madhusanka Liyanage is currently an Ad Astra
Fellow/Assistant Professor at School of Computer
Science, University College Dublin, Ireland. He
is also an adjunct professor at the University of
Oulu, Finland. He received his B.Sc. degree (First
Class Honours) in electronics and telecommunica-
tion engineering from the University of Moratuwa,
Moratuwa, Sri Lanka, in 2009, the M.Eng. degree
from the Asian Institute of Technology, Bangkok,
Thailand, in 2011, the M.Sc. degree from the Uni-
versity of Nice Sophia Antipolis, Nice, France, in
2011, and the Ph.D. degree in communication engineering from the University
of Oulu, Oulu, Finland, in 2016. He was also a recipient of prestigious Marie
Skłodowska-Curie Actions Individual Fellowship during 2018-2020. From
2011 to 2012, he worked a Research Scientist at the I3S Laboratory and
Inria, Sophia Antipolis, France. He has been a Visiting Research Fellow at
the Department of Computer Science, University of Oxford, Data61, CSIRO,
Sydney, Australia, the Infolabs21, Lancaster University, U.K., and Computer
Science and Engineering, The University of New South Wales during 2015-
2018.
He has co-authored over 85 publications including two edited books with
Wiley and one patent. He is the demo co-chair of WCNC 2019, publicity chair
of ISWCS 2019 and poster chair of 6G Summit 2020. He served as a Technical
program Committee Members at EAI M3Apps 2016, 5GU 2017, EUCNC
2017, EUCNC 2018, 5GWF 2018, MASS 2018, MCWN 2018, WCNC
2019, EUCNC 2019, EUCNC 2020, MASS 2020. ICBC 2021 conferences
and Technical program co-chair in SecureEdge workshop at IEEE CIT2017
conference and Blockchain for IoT workshop at IEEE Globecom 2018, IEEE
ICC 2020 and IEEE 5GWF 2020. He has also served as the session chair in
a number of other conferences including IEEE WCNC 2013, CROWNCOM
2014, 5GU 2014, IEEE CIT 2017, IEEE PIMRC 2017, 5GWF 2018, Bobynet
2018, Globecom 2018, WCNC 2019, ICC 2020. Moreover, He has received
two best Paper Awards in the areas of SDMN security (at NGMAST 2015)
and 5G Security (at IEEE CSCN 2017). Additionally, he has been awarded
three research grants and 22 other prestigious awards/scholarships during his
research career.
Dr. Liyanage has worked for more than twelve EU, international and
national projects in ICT domain. He held responsibilities as a leader of
work packages in several national and EU projects. Currently, he is the
Finnish national coordinator for EU COST Action CA15127 on resilient
communication services. In addition, he is/was serving as a management
committee member for four other EU COST action projects namely EU COST
Action IC1301, IC1303, CA15107 and CA16226. Liyanage has over three
years’ experience in research project management, research group leadership,
research project proposal preparation, project progress documentation and
graduate student co-supervision/mentoring, skills. In 2015, 2016 and 2017, he
won the Best Researcher Award at the Centre for Wireless Communications,
University of Oulu for his excellent contribution in project management and
dissemination activities. Additionally, two of the research projects (MEVICO
and SIGMONA projects) received the CELTIC Excellence Award in 2013 and
2017 respectively.
Dr. Liyanage’s research interests are 5G, SDN, IoT, Blockchain, MEC,
mobile and virtual network security. http://madhusanka.com
... In (Siriwardhana et al., 2019) and (Siriwardhana, Porambage, Ylianttila, & Liyanage, 2020) the advantages in performance are presented for low latency use cases for a private network by deploying core network functions locally in factory premises compared to a MNO with a remote core network. The latency measurements are based on different architecture options for the use cases of AR, massive wireless sensor networks and mobile robots. ...
Chapter
Smart Factory and 5th Generation Mobile Communication Systems gained much scientific and business attention during the years before the COVID-19 pandemic, as part of the so-called 4th Industrial Revolution (Industry 4.0). Even though there are second thoughts about whether the pandemic crisis will slow up Industry's 4.0 implementation, there has been an acceleration in the use of intelligent and reliable communications solutions in all business aspects, leading to the need for deepening our understanding about digital transformation and digitalization strategies. The current chapter aims to describe the concept of Smart Factory as a key factor of the 4th Industrial Revolution and to deliver its most important factors for a successful implementation. Moreover, smart factory is interrelated with 5th generation of mobile communication systems (5G), which seems to offer the capabilities needed for the advanced industrial digitalization.
... However, nearly all emerging communication techniques should be implemented on the telecommunication equipment manufactured by different vendors [17,18]. In general, vendors provide large amounts of parameter configurations to fit for different scenarios. ...
Article
Full-text available
With the rapid growth of users and sustained network demands powered by different industries, the quality of service (QoS) of the cellular network is affected by network traffic and computing loads. The current solutions of QoS improvement in academia focus on the fundamental algorithms within the physical and medium access control (MAC) layer. However, traffic features of various scenarios extracted from field data are rarely addressed for practical network configuration refinement. In this paper, we identify significant indicators of high traffic load cells according to the field data provided by telecommunication operators. Then, we propose the analysis flow of high traffic load cells with basic principles of network configuration refinement for QoS improvement. To demonstrate the proposed analysis flow and the refinement principles, we consider three typical scenarios of high traffic load cells, including high population density, emergency, and high-speed mobility. For each scenario, we discuss traffic features with field data. The corresponding performance evaluation demonstrates that the proposed principle can significantly enhance the network performance and user experience in terms of access success rate, downlink data rate, and number of high traffic load cells.
... Further advances in 5G developments will make it possible to seamlessly link embedded sensors to any object and quickly accelerate IoT adoption by many sectors. 5G will play an enormous role in enabling smart towns and intelligent homes, enhancing monitoring of assets and logistics, reinventing procedures for industrial production and maintenance and altering other IoT-ready businesses (Humayun et al., 2020;Siriwardhana et al., 2020;Rodriguez et al., 2021aRodriguez et al., , 2021b. The research objectives of this paper are as under: ...
Article
Purpose-The development of new communication technology such as 5G is now a solid choice for Industry 4.0. 5G in the fifth generation replaces the presently used mobile telecommunications networks. In every sector of modern life, this technology is designed to resolve the issue of the enormous rise in daily gadgets. 5G allows rapid data transfer and significantly improves the Internet of Things (IoT), which contains billions of devices. The purpose of this paper is to study the potential applications of 5G for Industry 4.0. Design/methodology/approach-Relevant research publications from Scopus, Web of Science and Google Scholar were identified to accomplish the objectives of this paper. 5G for Industry 4.0 enables crucial communications such as wireless management of machines and robotics to unleash Industry 4.0's full potential, including the considerable proliferation of IoT devices in 5G. This technology enhances quality control, increasing inspections that rely on real-time analysis for rapid and early detection. Findings-In Industry 4.0, 5G technology provides high speed and flexibility connectivity which helps to enhance the entire manufacturing system. This paper briefs about 5G and different network technologies used in 5G. Advancements, associated features and specialties, and significant enablers of 5G for Industry 4.0 are discussed. Finally, the paper identifies and discusses eighteen 5G applications for Industry 4.0. 5 G-enabled robots are used in manufacturing to carry out extensive work to enhance connectivity. Originality/value-5G is the next-generation communication technology, allowing numerous examples of inventive usage, including Industry 4.0. In line with its long-term vision of digitisation, 5G benefit the whole value chain, including consumers and businesses.
... First, recent studies [38]- [40] have paid attention to UADP based Time Sensitive Networking (TSN) oriented for achieving real-time industrial communication. Second, the OPC UA PubSub based on 5G will be a key research topic in the future [41]. In particular, AMQP PubSub can provide strong security and ultra-reliable features when combining IIoT with 5G with the cloud domain side [42]. ...
Article
Full-text available
Recently, the manufacturing industry has revolutionized with ’smart manufacturing’ and ’Industry 4.0’, and with the development of the Industrial Internet of Things (IIoT), increasing number of physical objects can access network services. However, critical issues such as real-time communication, interoperability, and interlinking of internet and industrial devices are yet to be addressed. These issues arise from the problem of interconnectivity between smart manufacturing systems and non-semantic industrial equipment. Therefore, we referred to ISO/IEC 30141, a standard IoT reference architect document, to solve the interconnection problem and bridge the gap between the Internet and industry. Furthermore, we proposed a four-domain integrated architecture that divides the manufacturing system into user, cloud service, sensing controlling, and device domains. The proposed architecture is validated by constructing the applicable industrial pick-and-place testbed for the manufacturing system. Thus, this testbed can serve as guideline to examine the state of the art of the OPC UA PubSub based protocol in addition to several other types of protocol approaches.
... However, wired connections are often used in VR and AR today, which could potentially be hazardous in some manufacturing scenes. Private 5G provides new opportunities to improve the experience of AR andVR [87]. Relying on 5G modems, AR/VR devices are able to host some on-device processing and distribute heavier computing to edge computers located in the on-premise network. ...
Article
Full-text available
A private fifth generation (5G) network is a dedicated 5G network with enhanced communication characteristics, unified connectivity, optimized services, and customized security within a specific area. By subsuming the advantages of both public and non-public 5G networks, private 5G networks have found their applications across industry, business, utilities, and the public sector. As a promising accelerator for Industry 4.0, the concept of a private 5G network has recently attracted significant research attention from industry and academia. This article is one of the first attempts to provide a comprehensive view on the research in private 5G network. Specifically, this paper first provides an overview of the concept and architecture of private 5G networks. It then discusses implementation issues and key enabling technologies for private 5G networks, followed by their more appealing use cases and existing real-life demonstrations. Finally, it examines some research challenges and future directions regarding private 5G networks.
... They present the local microlicensing model from regulatory perspective to complement the existing individual authorization and general authorization models by combining their benefits. Siriwardhana et al. [13] focuses on localized 5G networks and local operators that operate such networks within their premises with guaranteed quality and reliability to complement MNOs offerings. The authors investigate specific network architectures (and the related network function deployments), with a focus on two use cases: massive wireless sensor networks and mobile robots. ...
Article
Full-text available
Private networks will play a key role in 5G and beyond to enable smart factories with the required better deployment, operation and flexible usage of available resource and infrastructure. 5G private networks will offer a lean and agile solution to effectively deploy and operate services with stringent and heterogeneous constraints in terms of reliability, latency, re-configurability and re-deployment of resources as well as issues related to governance and ownership of 5G components, and elements. In this paper, we present a novel approach to operator models, specifically targeting 5G and beyond private networks. We apply the proposed operator models to different network architecture options and to a selection of relevant use cases offering mixed private–public network operator governance and ownership. Moreover, several key enabling technologies have been identified for 5G private networks. Before the deployment, stakeholders should consider spectrum allocation and on-site channel measurements in order to fully understand the propagation characteristic of a given environment and to set up end-to-end system parameters. During the deployment, a monitoring tools will support to validate the deployment and to make sure that the end-to-end system meet the target KPI. Finally, some optimization can be made individually for service placement, network slicing and orchestration or jointly at radio access, multi-access edge computing or core network level.
Article
Network slicing has been widely agreed as a promising technique to accommodate diverse services for the Industrial Internet of Things (IIoT). Smart transportation, smart energy, and smart factory/manufacturing are the three key services to form the backbone of IIoT. Network slicing management is of paramount importance in the face of IIoT services with diversified requirements. It is important to have a comprehensive survey on intelligent network slicing management to provide guidance for future research in this field. In this paper, we provide a thorough investigation and analysis of network slicing management in its general use cases as well as specific IIoT services including smart transportation, smart energy and smart factory, and highlight the advantages and drawbacks across many existing works/surveys and this current survey in terms of a set of important criteria. In addition, we present an architecture for intelligent network slicing management for IIoT focusing on the above three IIoT services. For each service, we provide a detailed analysis of the application requirements and network slicing architecture, as well as the associated enabling technologies. Further, we present a deep understanding of network slicing orchestration and management for each service, in terms of orchestration architecture, AI-assisted management and operation, edge computing empowered network slicing, reliability, and security. For the presented architecture for intelligent network slicing management and its application in each IIoT service, we identify the corresponding key challenges and open issues that can guide future research. To facilitate the understanding of the implementation, we provide a case study of the intelligent network slicing management for integrated smart transportation, smart energy, and smart factory. Some lessons learnt include: 1) For smart transportation, it is necessary to explicitly identify service function chains (SFCs) for specific applications along with the orchestration of underlying VNFs/PNFs for supporting such SFCs; 2) For smart energy, it is crucial to guarantee both ultra-low latency and extremely high reliability; 3) For smart factory, resource management across heterogeneous network domains is of paramount importance. We hope that this survey is useful for both researchers and engineers on the innovation and deployment of intelligent network slicing management for IIoT.
Article
Purpose The development of new communication technology such as 5G is now a solid choice for Industry 4.0. 5G in the fifth generation replaces the presently used mobile telecommunications networks. In every sector of modern life, this technology is designed to resolve the issue of the enormous rise in daily gadgets. 5G allows rapid data transfer and significantly improves the Internet of Things (IoT), which contains billions of devices. The purpose of this paper is to study the potential applications of 5G for Industry 4.0. Design/methodology/approach Relevant research publications from Scopus, Web of Science and Google Scholar were identified to accomplish the objectives of this paper. 5G for Industry 4.0 enables crucial communications such as wireless management of machines and robotics to unleash Industry 4.0’s full potential, including the considerable proliferation of IoT devices in 5G. This technology enhances quality control, increasing inspections that rely on real-time analysis for rapid and early detection. Findings In Industry 4.0, 5G technology provides high speed and flexibility connectivity which helps to enhance the entire manufacturing system. This paper briefs about 5G and different network technologies used in 5G. Advancements, associated features and specialties, and significant enablers of 5G for Industry 4.0 are discussed. Finally, the paper identifies and discusses eighteen 5G applications for Industry 4.0. 5 G-enabled robots are used in manufacturing to carry out extensive work to enhance connectivity. Originality/value 5G is the next-generation communication technology, allowing numerous examples of inventive usage, including Industry 4.0. In line with its long-term vision of digitisation, 5G benefit the whole value chain, including consumers and businesses.
Conference Paper
Full-text available
In addition to the high degree of flexibility and customization required by different vertical sectors, 5G calls for a network architecture that ensures ultra-responsive and ultra-reliable communication links. The novel concept called micro-operator (uO) enables a versatile set of stakeholders to operate local 5G networks within their premises with a guaranteed quality and reliability to complement mobile network operators' (MNOs) offerings. In this paper, we propose a descriptive architecture for emerging 5G uOs which provides user specific and location specific services in a spatially confined environment. The architecture is discussed in terms of network functions and the operational units which entail the core and radio access networks in a smart factory environment which supports industry 4.0 standards. Moreover, in order to realize the conceptual design, we provide simulation results for the latency measurements of the proposed uO architecture with respect to an augmented reality use case in industrial internet. Thereby we discuss the benefits of having uO driven local 5G networks for specialized user requirements, rather than continuing with the conventional approach where only MNOs can deploy cellular networks.
Article
Full-text available
Ultra-dense networks (UDNs) are a natural deployment evolution for handling the tremendous traffic increase related to the emerging 5G services, especially in urban environments. However, the associated infrastructure cost may become prohibitive. The evolving paradigm of network slicing can tackle such a challenge while optimizing the network resource usage, enabling multi-tenancy and facilitating resource sharing and efficient service-oriented communications. Indeed, network slicing in UDN deployments can offer the desired degree of customization in not only vanilla radio access network (RAN) designs, but also in the case of disaggregated multi-service RANs. In this article, we devise a novel multi-service RAN environment, RAN runtime, capable of supporting slice orchestration procedures and enabling flexible customization of slices as per tenant needs. Each network slice can exploit a number of services, which can be either dedicated or shared between multiple slices over a common RAN. The novel architecture we present concentrates on the orchestration and management systems. It interacts with the RAN modules, through the RAN runtime, via a number of new interfaces enabling a customized dedicated orchestration logic for each slice. We present results for a disaggregated UDN deployment where the RAN runtime is used to support slice-based multi-service chain creation and chain placement, with an auto-scaling mechanism to increase the performance.
Conference Paper
Full-text available
The demand for wireless information exchange in the form of data and multimedia content has been growing exponentially over the past few years, and the trends are expected to persist for the future. Due to the continued growth in demand and support required for new use cases and features, currently the fifth generation (5G) of mobile networks are being developed. We present the concept of micro-operators and discuss the use of cloud-based deployments to enable new connectivity and service provisioning paradigms. The use of such networks for supporting new verticals, related challenges and impacts on network security and the regulatory support from new spectrum regimes are also briefly discussed. Performance issues and optimization approaches are briefly outlines to enable a fast and efficient deployment of 5G networks which are suitable for new verticals and markets, providing tailored services and applications.
Article
Full-text available
Future digital society depends heavily on timely availability of high quality wireless connectivity the offering of which today is dominated by mobile network operators (MNOs). Future 5G systems aim at connecting billions of devices to serve versatile location and case specific needs of vertical sectors in parallel with the provisioning of traditional mobile broadband services. As the majority of mobile traffic originates from indoors, cost-efficient and fast deployment of new indoor small cell networks is fundamental, which calls for new developments in regulation and technology to enable new business. This paper proposes the concept of micro operators (uO) for local service delivery in 5G to build indoor small cell communication infrastructure and offer context related services and content. Key elements of the new micro operator concept are introduced including regulation-related factors of local spectrum access rights, and technology-related factors of flexible network implementation. Several business opportunities are identified for the uO concept including the provisioning of hosted local connectivity to all MNOs in specific locations, operation of secure networks for vertical sector specific use, and offering of locally tailored content and services.
Article
In order to mitigate the long processing delay and high energy consumption of mobile augmented reality (AR) applications, mobile edge computing (MEC) has been recently proposed and is envisioned as a promising means to deliver better Quality of Experience (QoE) for AR consumers. In this article, we first present a comprehensive AR overview, including the indispensable components of general AR applications, fashionable AR devices, and several existing techniques for overcoming the thorny latency and energy consumption problems. Then we propose a novel hierarchical computation architecture by inserting an edge layer between the conventional user layer and cloud layer. Based on the proposed architecture, we further develop an innovative operation mechanism to improve the performance of mobile AR applications. Three key technologies are also discussed to further assist the proposed AR architecture. Simulation results are finally provided to verify that our proposals can significantly improve latency and energy performance as compared to existing baseline schemes.
Book
The first comprehensive guide to the design and implementation of security in 5G wireless networks and devices. Security models for 3G and 4G networks based on Universal SIM cards worked very well. But they are not fully applicable to the unique security requirements of 5G networks. 5G will face additional challenges due to increased user privacy concerns, new trust and service models and requirements to support IoT and mission-critical applications. While multiple books already exist on 5G, this is the first to focus exclusively on security for the emerging 5G ecosystem. 5G networks are not only expected to be faster, but provide a backbone for many new services, such as IoT and the Industrial Internet. Those services will provide connectivity for everything from autonomous cars and UAVs to remote health monitoring through body-attached sensors, smart logistics through item tracking to remote diagnostics and preventive maintenance of equipment. Most services will be integrated with Cloud computing and novel concepts, such as mobile edge computing, which will require smooth and transparent communications between user devices, data centers and operator networks. Featuring contributions from an international team of experts at the forefront of 5G system design and security, this book: Provides priceless insights into the current and future threats to mobile networks and mechanisms to protect it. Covers critical lifecycle functions and stages of 5G security and how to build an effective security architecture for 5G based mobile networks. Addresses mobile network security based on network-centricity, device-centricity, information-centricity and people-centricity views. Explores security considerations for all relative stakeholders of mobile networks, including mobile network operators, mobile network virtual operators, mobile users, wireless users, Internet-of things, and cybersecurity experts. Providing a comprehensive guide to state-of-the-art in 5G security theory and practice, A Comprehensive Guide to 5G Security is an important working resource for researchers, engineers and business professionals working on 5G development and deployment.
Article
There is considerable pressure to define the key requirements of 5G, develop 5G standards, and perform technology trials as quickly as possible. Normally, these activities are best done in series but there is a desire to complete these tasks in parallel so that commercial deployments of 5G can begin by 2020. 5G will not be an incremental improvement over its predecessors; it aims to be a revolutionary leap forward in terms of data rates, latency, massive connectivity, network reliability and energy efficiency. These capabilities are targeted at realising high speed connectivity, the internet of things, augmented virtual reality, the tactile internet, etc. The requirements of 5G are expected to be met by utilising large bandwidths available in mm-wave bands, increasing spatial degrees of freedom via large antenna arrays and 3D MIMO, network densification and new waveforms that provide scalability and flexibility to meet the varying demands of 5G services. Unlike the one size fits all 4G core network, the 5G core network must be flexible and adaptable and is expected to simultaneously provide optimised support for the diverse 5G use case categories. In this paper, we provide an overview of 5G research, standardization trials and deployment challenges. Due to the enormous scope of 5G systems, it is necessary to provide some direction in a tutorial article and in this overview the focus is largely user-centric, rather than devicecentric. In addition to surveying the state-of-play in the area, we identify leading technologies, evaluating their strengths and weaknesses, and outline the key challenges ahead, with research test-beds delivering promising performance but pre-commercial trials lagging behind the desired 5G targets.
Article
With the introduction of the Internet of Things (IoT) and cyberphysical system (CPS) concepts in industrial application scenarios, industrial automation is undergoing a tremendous change. This is made possible in part by recent advances in technology that allow interconnection on a wider and more fine-grained scale. The purpose of this article is to review technological trends and the impact they may have on industrial communication. We will review the impact of IoT and CPSs on industrial automation from an industry 4.0 perspective, give a survey of the current state of work on Ethernet time-sensitive networking (TSN), and shed light on the role of fifth-generation (5G) telecom networks in automation. Moreover, we will point out the need for harmonization beyond networking.