Beyond 5G Private Networks: the 5G CONNI Perspective

Full text

Beyond 5G Private Networks: the 5G CONNI
Perspective
Emilio Calvanese Strinati∗, Thomas Haustein †, Mickael Maman ∗, Wilhelm Keusgen †, Sven Wittig †,
Mathis Schmieder †, Sergio Barbarossa ‡, Mattia Merluzzi ‡, Henrik Klessig §, Fabio Giust ¶,
Daniele Ronzani ¶, Shuo-Peng Liangk, Jack Shi-Jie Luok, Cheng-Yi Chien∗∗, Jiun-Cheng Huang∗∗,
Jen-Sheng Huang †† Tzu-Ya Wang‡‡,
∗CEA-Leti, France †Fraunhofer HHI, Germany ‡Sapienza University of Rome, Italy §Robert Bosch GmbH, Germany
¶Athonet, Italy kITRI, Taiwan ∗∗Chunghwa Telecom, Taiwan ††Alpha Networks Inc., Taiwan ‡‡III, Taiwan
Abstract—Future Smart Factories will leverage Industry 4.0
and 5G technology to increase both flexibility and efficiency of
manufacturing processes, thus ensuring global competitiveness of
industrial manufacturing. 5G technologies such as network slic-
ing may accommodate industrial applications in public networks,
while Private 5G Networks, operating locally and being highly
optimized towards specific applications, may require disruptive
technologies to meet the specific and challenging industrial use
cases requirements. The 5G CONNI Europe-Taiwan project
investigates innovative solutions for Private 5G and beyond
Network. This includes the definition of new architectures and
measurements tools as well as, the development of innovative
technologies and their enabling components in the context of
mobile edge assisted URLLC. Building on the premise of Private
5G Networks, the 5G CONNI project will offer an unprecedented
integrated end-to-end 5G testing network for testing specifically
industrial applications in accordance with updated 5G standard-
ization specifications. The 5G CONNI project will validate its
innovative and technologically advanced solutions and compo-
nents with a real-field cross-continental end-to-end industrial
Private 5G Network demonstration between two interconnected
industrial manufacturing sites in Taiwan and in Europe.
I. INTRODUCTION
The fifth generation of mobile communication networks
(5G) expands the scope of communication wireless networks
beyond individual human end users towards an integrated
communication system, which also provides wireless connec-
tivity to new vertical applications driven by industries such
as manufacturing, automotive, health or agriculture. Sparked
by this promise, the last years have seen a surge in research
and development activities targeting 5G technologies around
the globe. With 3GPP Release 15 finished in mid-2018, the
first commercial deployments of 5G networks will become
operational within the year. While 4G was largely associated
to operators, because of the spectrum available and the cost of
the equipment, 5G opens the new market of private networks
and, the already kicked-off research on 6G [1] anticipate its
fundamental role for future evolution of 5G networks. Indeed,
if industries can set up a private network to dynamically claim
reserved and specified frequency bands then they could use
them to provide 5G services in limited areas. A 5G network
split will allow operators to create virtual private networks for
their industrial customers, using the computing and storage
capacity of their own servers in the edge cloud.
The 5G CONNI project embraces the challenge of designing
and trial effective solutions for future beyond 5G private
networks, focusing on three key challenging innovation areas:
This work was supported by the European Union H2020 / Taiwan Project
5G CONNI under grant n.861459
1) Design, Planning, Operation and Maintenance (OAM)
of Private 5G and beyond Networks: Future design of private
networks requires the definition of novel network architectures
enabling effective capital expenditure (CAPEX) and opera-
tional expenditure (OPEX) through functional splits within
radio access and core network, allowing for different opera-
tional models by shifting network functions between privately
and publicly operated cloud environments. Moreover, new
tools for dynamic monitoring of performance are needed. To
this end, 5G CONNI investigates novel solutions for com-
munication environment, available Mobile Edge Computing
(MEC) resources and online monitoring of QoS associated
slice performance. The goal is to define for the specific
challenging context of private networks, tools for cost effective
radio and computing network deployment (including sensors
and gNBs) and online monitoring of key quality of service
indicators such as coverage, throughput, latency and reliability.
2) Technology Verification: The Smart Factory test-bed
realization will follow a hybrid approach consisting of re-
cent state of the art 5G equipment, 5G system components
purpose-built within the project and innovative research on
new early technologies. Based on current 3GPP standards, key
components of the 5G system such as industrial Customer
Premise Equipment (CPE) and base station equipment (gNB)
and core network functions will be developed and integrated
into selected use cases at both test-bed facilities.
3) Enabling Technologies for Industrial Applications: 5G
CONNI will develop four advanced components: Radio Net-
work Technical Enabler includes 5G RAN components (e.g.,
CPE and gNBs), which comply with 3GPP standards and some
enhancements providing deterministic Ultra Reliable and Low
Latency Communication (URLLC) protocols with a flexible
tradeoff between reliability and latency, supporting multi-
connectivity and reliable with intermittent channels. Core Net-
work Technical Enabler includes a 5G Core (5GC) prototype
and a NFV-like lightweight orchestration framework to handle
Virtual Network Function (VNF) placement and lifecycle
management. Mobile Edge Cloud Enabler includes a 5G
MEC Cloud based on 3GPP standard and a programmable
traffic steering function, based on 3GPP enablers for edge
computing. Finally , Industrial Application Enabler includes
Computer Numerical Control (CNC) diagnostic in MEC, a
joint optimization of enabling technologies (radio, core, MEC),
an AI assisted orchestration to jointly optimize the deployment
of VNF and the allocation of cloud resources and a dynamic
resource (e.g., virtual machines) allocation for computation
offloading from mobile inspectors to mobile edge hosts.
The paper is organized as follows. Section II analyses
the potential use-cases for private networks and Section III

gives an overview of the key enabling technologies of 5G
CONNI project. Section IV discusses about the regulatory
and standardization perspective for private networks. Section
V presents the selected Proof of Concepts (PoC) with their
respective requirements and Section VI concludes the paper.
II. ANA LYSIS ON POTE NT IA L USE CA SE S FO R PRI VATE
NET WORKS
This section investigates six potential use cases which are
relevant in the 5G CONNI context of private 5G networks
along with descriptions of techno-economic benefits.
A. Process Diagnostics by CNC and Sensing Data Collection
For factories with small batch production, high quality and
high precision workpiece process stability is crucial and can be
achieved by process diagnosis systems. Such systems utilize
sensor and CNC data to implement tool condition monitor-
ing, chatter detection, spindle health check, and alarming.
Typically, multiple sensors with very high sampling rates
and data rates often exceeding 200 Mbps per machine are
required for high speed, high precision machining tasks. Data
pre-processing, such as feature extraction, necessitates high-
performance computers, which however are costly and less
flexibly deployable across the whole shop floor. 5G along
with edge computing brings new opportunities for large scale
process diagnostic and condition monitoring for multiple ma-
chines across the process chain. Analysis systems deployed in
the edge data center as virtualized services can dynamically
be allocated for specific machines or production tasks while
being interconnected with the online monitoring system at the
machine via the 5G network. Hence, data collection systems
can be flexibly deployed reducing time and cost for diagnosis
system setup and redesigning machine and cable layouts.
Moreover, machine learning or deep learning models establish
process models with higher accuracy in the edge data center.
B. Process Diagnostics using Augmented/Virtual Reality
Conventional machining process planning relies on time-
consuming Trial-and-Error (T&E) activities to determine
force-related parameters (e.g.,feed rate, spindle speed, depth
and width of cut). In order to reduce the cost for T&E, addi-
tional diagnosis models are used to detect vibration, collision,
acoustic emission, temperature, or energy consumption – input
that goes beyond that of traditional computer-aided manu-
facturing software to generate tool paths along a workpiece
geometry. Virtual Reality (VR) or Augmented Reality (AR)
can help process engineers reveal full productivity of machines
by superimposing graphical objects (e.g., 3D models, charts,
vector fields and text messages) on top of a video image
playback on head-mounted displays. Here, high data rate along
with low latency properties of 5G can be utilized, while high
resolution and high frame rate video streams are rendered
remotely in the edge cloud and are wirelessly transferred to
the lightweight VR/AR device facilitating high mobility on
the shop floor. Recognized objects (e.g., machines, spindles
and workpiece) are then linked with the digital twins to query
corresponding 3D models, sensing data, and condition values,
and synthesized as virtual objects in the AR/VR scenes. For
example, tool paths can be plotted and color-coded according
to features, such as the vibration level, and superimposed
on the real machine image or the machine 3D model. This
enables process engineers observe machining conditions in a
more intuitive way, shorten the T&E process planning time
and interact with various digital twins at the same time in a
focused and hands-free manner.
C. Robot Platform with Edge Intelligence and Control
Stationary robot platforms perform a variety of different
tasks, which range from assembly, inspection, and packaging,
to more complex tasks such as collaboratively carrying and
manipulating objects, mostly at designated locations. More
compact, semi-mobile robot platforms are used to carry out
tasks on less heavy workpieces. Especially for processing
smaller batches, they can already be used with a flexibility
up to a certain extent, but the necessity of steadily increasing
plant productivity, enhanced flexibility on the shop floor and
higher cost efficiency calls for further automation in factories.
Offloading of robot control and intelligence to the cloud,
instead of having dedicated Industrial PCs or programmable
logic controllers, is a promising approach to further increase
the flexibility and versatility of the production system, to lower
the cost of robotics equipment and to improve the scalability
through softwarization and pooling of computing resources. A
5G link can interconnect the robot with the backend to fulfill
the timeliness and reliability requirements imposed by the
exchange of motion control (e.g. target values of joint positions
or velocities) and feedback messages (e.g. joint angles and
torques). Here, multiple robots can collaborate in a flexible
manner because sensory, control and task data can directly be
exchanged through appropriate interfaces between the virtu-
alized control functions. In addition, complementary sensors,
such as industrial cameras with workpiece inspection tasks
offloaded to the edge cloud, can further enhance the level of
cloud-based intelligence. Finally, higher production efficiency
of factory personnel can also be achieved by centrally man-
aging, troubleshooting, monitoring and programming robots.
D. Digital Enhanced Cordless (DECT) Phones Replacement
It is common to find workers using traditional Push-To-Talk
(PTT) radio devices in manufacturing sites, production plants
and warehouses, and office staff. Nevertheless, the bandwidth
of PTT devices limits their use to voice calls only and mostly
bound to DECT phones and the WiFi network. A private
5G network integrated with the corporate ICT system can
provide a single platform to realize voice over 5G integrated
with mission critical services that enable provisioning of QoS-
guaranteed voice calls, group calls, video calls, PTT, instant
messaging, and data transfer. Such services are important to
coordinate activities of technicians on the shop floor, to share
images and recordings in real time or to send emergency alerts.
In addition, the same device can be used to retrieve/upload data
from the local cloud, e.g., to send a barcode scan of an item in
the warehouse to an inventory to fetch the item’s data sheet.
Furthermore, 5G offers better integration with the application
domain through an extended set of APIs, such as the unified
data management facilitating the exposure of network and
subscriber data to applications. For example, providing SIM
credentials of an employee in combination with her position
using the localization capabilities of the 5G network may serve
to authorize her to access the company’s facilities.
E. Shop Floor Asset Tracking
Conventional asset tracking on the shop floor relies on
check-ins and check-outs of materials between production
stations so that the production management system can locate
specific parts, workpiece, assembly, fixture, and tools on
the shop floor. Nevertheless, the time delay between check-
ins and check-outs may cause problems because of missing
path information of the material between them. Although the
latest Automated Guided Vehicles (AGVs) are equipped with
sensors, such as ultrasonic sensor and LiDar for autonomous

navigation, AGV dispatching systems can only track materials
handled by AGVs and AGVs can only navigate through a pre-
defined map with fixed shop floor layout. Using a 5G private
network, sensors (e.g., cameras, LiDar, ultrasonic, infrared)
can be wirelessly and flexibly connected around the factory
so that also other material flow systems (e.g., storage and
conveyor systems) can track, plan and optimize material flow.
F. Cloud-Based Controller for CNC
To meet the requirements of small batch production and
massive customization, configuration of machine tools or
production stations needs to be flexible. The architecture
of conventional CNC systems for machine tools are fixed.
Hence, once the configuration and number of axes is de-
cided, the software architecture and control block diagram is
fixed. However, in order to meet the massive customization
scenario, the number of moving axes and combination of
hybrid manufacturing processes (e.g. additive and machining
processes) must be able to be modified with very low cost to
quickly respond to many small batch or one piece contracts.
Although the industrial control over 5G network or cyber-
physical control has been described in [2] and communications
for automation in vertical domains, such as factory of the
future has been described in [3], design, implementation and
deployment of a cyber-physical controller is still needed to
define actual specification for various scenarios (e.g., flexible
machining machine consisting of many spindles and moving
axes to perform sequential operations of a specific part, or
a specialized production station consisting of an additive 3D
printing machine and subtractive milling or turning machines).
III. KEY ENABLING TECHNOLOGIES
A. MEC
MEC is a standardized architecture related to edge com-
puting, which brings high bandwidth, low latency, improved
security and enhanced network congestion management, by
allowing dynamic application/service deployment and cloud
capabilities at the edge of the network. Privately-owned or
a public one, the MEC technology is meant for on-premise
enterprise deployments. It provides dedicated services in small
closer areas for each enterprise and thus opens new business
opportunity. In 5G CONNI project, MEC deployment sits in
proximity of the 5G RAN equipment to process the data as
close as possible to where they are generated and consumed.
It must provide the ability to handle high-bandwidth and ultra-
low latency data that matches 5G enhanced Mobile BroadBand
services (eMBB) and URLLC scenarios. MEC also provides
cloud platform to onboard the applications, steer traffic route
and manage industrial applications. In 5G networks, MEC
functionalities are deployed behind the User Plane Function
(UPF) [4]. Then, within the factory, the edge data centers will
enable real-time analytics on the data generated by several
industrial sensors, with low latency requirements, to perform
process diagnosis, anomaly detection, predictive maintenance.
One of the goals for the smart factory is to enable machine
learning/AI at the edge [5]. It comes with several challenges
for network resource management, with the need of exploring
the best trade-off between energy, latency, reliability, and
learning accuracy, in the dynamic and complex environments
of the factory facilities. In particular, sensor networks are
characterized by battery limited nodes with poor computa-
tional capacities. Thus an important aspect is to dynamically
orchestrate radio and computation resources to upload these
data to the edge servers, enabling machine learning and AI
algorithms with the least possible energy consumption, target
accuracy constraints and low end-to-end delays [6].
B. Network Slicing and Orchestration
Network slicing is a framework to divide programmable and
divisible physical network resources into many virtual network
slices having independent network resources to each other.
In the 5G scenario, a slice spans across resources from the
RAN to the 5GC, and each slice can contain multiple network
services or applications required by users. Network slices
are built according to specific requirements, and are thence
specialized with different characteristics (e.g., for URLLC or
eMBB slices). The network slicing orchestration, mentioned
in [7] and [8], elaborates it around three key points: i)
to create virtual network instances on the physical network
resources to use infrastructure resources, ii) to map network
services or applications to virtual network instances for pro-
viding service chain to users, and iii) to manage the basic
operation of virtual network instances, such as maintaining
communication between services and the network slicing and
managing the lifecycle of virtual network instances. Network
slicing is important primarily for large-scale operators, which
can offer dedicated communication services to verticals with
fulfillment of stringent requirements as an operator-managed
private network. Nevertheless, in the context of enterprise-
managed private networks, 5G CONNI will exploit the flexi-
bility and manageability offered by network slicing to differ-
entiate services within the organization, in order to meet the
communication requirement in different environments (e.g.,
manufacturing lines, offices, packaging and logistics).
C. Multi-connectivity for unreliable high capacity link
While mmWave communications provide high data rate to
enable eMBB, they are strongly affected by blockage due to
obstacles absorption. This is amplified by the directionality
of the communication, used to compensate the strong path
loss typical of high frequencies. Thus, mmWave communica-
tions are intermittent and unreliable, which is not compatible
with URLLC required in the smart factory. Thus, multi-
connectivity is a promising strategy to benefit from the high
data rate by maintaining the target reliability, by establishing
the connection with multiple gNBs, to reduce the probability
of outages [9]. Multi-connectivity can refer to multi-Radio
Access Technologies (RAT), in which frequency diversity is
exploited by using backup reliable links at lower frequencies,
or to single RAT multi-link strategies, in which the connection
with multiple mmWave gNBs is established. While the first
solution is more reliable due to lower frequency stable links,
the latter allows to maintain high data rate connection even
if one or more links are lost. In [10], block erasure channel
coding is investigated as an enabler of mmWave reliable
multi-link communications. Even in this case, the challenge
is to devise dynamic resource allocation strategies able to
guarantee a certain target reliability of the wireless connection,
possibly with low end-to-end delays and reduced energy
consumption. Another aspect pertaining diversity, but related
to computational aspects, is the fact that machine learning and
AI algorithms could benefit from multiple running instances in
different servers or different cores of the same server, so that
better learning accuracy is achieved, with the cost of a higher
energy consumption and computation resource utilization.
D. Core Network
The 5G Core Network (5GC) has been standardized in
3GPP specifications in order to enable cloud native imple-
mentations and deployments [11]. The key innovations are i)

a clear separation between control plane and user plane, with
the replacement of mobile core gateways by the UPF which
can be deployed as one or multiple instances and controlled
by the session management function, ii) the service-based
architecture of the control plane, inspired by modern software
architecture paradigms used in web applications and in the
cloud, iii) the exposure of network information and services to
other external services, let them belong to the network operator
or to a trusted third party. The 5G CONNI project embraces
the points above and proposes a mobile 5GC split into two
main parts with different nature: the on-premise part consists
of a network of UPFs, which forwards traffic between the RAN
and the applications and the cloud part which consists of the
control plane and OAM functions. The UPFs are installed on
on-premise commercial hardware designed to optimize traffic
throughput. Their operations are driven by the control plane
functions sitting in the cloud, which hosts also the configu-
ration, provisioning and monitoring facilities. This way, the
cloud part accommodates all the functions and workloads that
both network and system administrators require to operate the
overall solution, whereas the on-premises functions are limited
to traffic forwarding. This approach not only enhances system
operations and maintenance, but also simplifies integration
with the application ecosystem, which leverages traditional
and edge cloud computing. For smart factories and Industry
4.0 use cases, the Time sensitive networking (TSN) is one of
the key requirements on industrial communication technology.
TSN can provide deterministic services over IEEE standard
802.3 Ethernet wired networks, and can be applied in many
verticals. 5G CONNI project will seamlessly integrates the 5G
core system as a bridge to TSN system.
E. OAM and Dynamic KPIs Monitoring
A 5G end-to-end system including 5G RAN, MEC, 5GC
and related application will be deployed into factories to
provide demonstration and service for specific industrial ap-
plication in 5G CONNI project. The target is not only to
build up the 5G enabled communication infrastructure but
also to make sure the operation and maintenance of specific
industrial application will meet the KPI. To fulfill this, an
OAM and KPI monitoring system will be designed into the
system as well. Its implementation includes the management
of Fault, Configuration, Account, Performance and Security.
In 5G CONNI project, except the account management, other
management requirement is going to build up in this 5G end
to end system. The specific KPI will be implemented with
two requirements. One is from 3GPP specification. The other
is from the use cases proposed in the project (e.g., end-to-
end latency, service bit rate, time synchronization and secure
remote access). With OAM and related KPI implemented in
the 5G end to end system, users could monitor, configure the
system and improve the efficiency of operation accordingly.
IV. REG UL ATORY & STANDARDIZATION PER SP EC TI VE O N
5G A ND BE YON D PRI VATE NE TW OR KS
Private networks constitute a paradigm shift for design
and operation of mobile radio networks which originate from
standards created for global deployments of public mobile
network infrastructure to connect mobile phones everywhere
where people live, travel and work. 5G setting the foundation
for purpose targeted private networks in particular in an
industrial context addressing non-functional and functional
requirements and KPIs as well as spectrum and regulation,
thus providing the framework for non-public deployments of
5G technology in Mission Critical Communication (MCC) in
factory environments.
A. Spectrum allocation models for private networks
The electromagnetic spectrum is, for most parts, regu-
lated by governments and harmonization across regions and
continents was key for global success of standardized radio
technologies. Some parts of the spectrum are allocated to
general purposes, e.g. Industrial, Scientific, Medical (ISM),
and are available unlicensed under strict usage rules. Other
parts of the spectrum are licensed, which means that only
the license holder can deploy and operate radio equipment
and services in this particular spectrum. Spectrum that is
designated for terrestrial mobile telecommunication services,
needed to operate 4G, 5G and beyond, is usually divided
into sub-bands auctioned off to mobile network operators
(MNO) at high prices for a state or country wide license. Such
mechanism became an obstacle for availability of designated
spectrum to be licensed to professional users for localized
private networks which had so far to share unlicensed spectrum
potentially with other users and equipment operated in the
same spectrum and the same location. This puts feasibility
limits for MCC because it is impossible to guarantee quality of
service or latency. As a consequence locally available licensed
spectrum becomes a prerequisite for the success of Industry
4.0 while unlicensed spectrum provides additional capacity for
non-MCC supplementary services. Fortunately, governments
have started the process of opening allocated spectrum as
licensed shared spectrum or dynamic spectrum sharing for
localized use in specific bands to enable the deployment and
operation of private 5G networks.
1) Licensed shared operation: Several countries, including
Germany, UK, and Taiwan, have started the process of allo-
cating parts of the 5G spectrum for local private use instead
of for nationwide coverage. Non- internet service providers
can apply for a license for up to 100 MHz of spectrum in
the range of 3.7 to 4.9 GHz, depending on the country. For a
small (yearly) fee, companies can then use those frequencies
exclusively on their premises to deploy a private network.
2) Dynamic Spectrum Access: In US, the 3.5 GHz fre-
quency band was recently opened up for commercial use by
the US Federal Communications Commission (FCC). This
band is known as the Citizen Broadband Radio Service
(CBRS) and does not require spectrum licenses. Access and
operation is governed by a dynamic spectrum access system,
but the users are required to take care not to interfere with
others already using nearby bands.
B. Standardization perspective
Beside the regulatory aspects, standardization is another key
for the success of 5G private networks. 5G CONNI devotes
to a number of targeted, interrelated and closely coordinated
standardization activities, mainly on 3GPP. These activities
are derived from the architecture aspects as well as the
research pillars. In the 3GPP RAN#80 plenary [12], several
Rel-16 Study Items (SIs) and Work Items (WIs) were agreed
addressing non-public networks (NPN) issues closely related
to the research in 5G CONNI, examples are given below:
1) Management of Non-Public Networks [13]: The ongoing
Rel-17 WI defines management requirements for and roles in
NPN and specifies deployment scenarios, including those in
factories. Special attention is given to provisioning and expo-
sure of management functions, services and data. These topics
are well aligned with the 5G CONNI work plan and partners
will actively follow and contribute to the standardization.
2) Study on enhanced support of NPN [14]: The main fo-
cus of this ongoing Rel-17 SI is the credential and subscription
handling in NPN. Onboarding and provisioning procedures

will be defined and entities handling the subscription will be
specified. Furthermore, the study aims to investigate enhance-
ments to the service requirements for audio-visual content.
These topics are very relevant to 5G CONNI and contributions
in the 3GPP are planned.
3) Study on channel modeling for indoor industrial sce-
narios [15]: This Rel-16 SI aims to extend the 3GPP TR
38.901 channel model with Indoor-Industrial scenarios. Before
system deployment, the propagation characteristics have to be
fully understood and parameterized for system level simula-
tions. The aspect of interference coordination is especially im-
portant for private networks. Even though a first version of the
channel model has already been released, further investigations
of the indoor industrial channel [16], [17] were necessary.
5G CONNI plans further extensive measurements in industrial
environments both at 3.7 and 28 GHz to be contributed to
standardizing to enhance the channel model.
Several other ongoing SIs and WIs regarding unlicensed ac-
cess, dynamic spectrum sharing, edge applications and flexible
local area data networks are related to NPN, highlighting the
relevance in standardization.
V. 5G AND BEY ON D PRI VATE NE TWORK TARGET PROOF
OF CO NC EP T FO R TARGET USE CAS ES
Three of the use cases outlined above are selected for im-
plementation at two trial sites in Europe and Taiwan: Process
diagnostics by CNC and sensing data collection (UC-1), Pro-
cess diagnostics using AR/VR (UC-2) and Robot platform with
edge intelligence and control (UC-3). This section describes
some implementation aspects and challenges relevant for the
demonstration implementations and studies. The analysis of
non-functional and functional requirements are summarized
in Tables I and II, respectively.
TABLE I: Non-functional Requirements
Non-functional
requirement
UC-1 UC-2 UC-3
Service bitrate 208 Mbps per
machine
Up to 1 Gbps
per device
0.2–1.6 Mbps
(control), 100s
Mbps (video)
Communication
area
Some 1000 m²
(2644 m² for
demo setup)
Some 1000 m²
(2644 m²)
100 m x 100
mx15m
Connection
density
10s per shop
floor (11)
Up to 10 per
shop floor (6)
1 to few tens
per shop floor
Area traffic
capacity
86.5 Mbps per
100 m²
227 Mbps per
100 m²
100s Mbps
per 100 m²
UE speed stationary <3 km/h <2 km/h
Positioning acc. n/a <1 m (horiz.) n/a
Positioning lat. n/a <15 ms n/a
Motion-to-photon
latency
n/a <50 ms n/a
End-to-end
latency
n/a <10 ms 1 ms – 7 ms
Transfer interval n/a n/a 5 ms – 20 ms
Transmission time n/a n/a 1.4 ms – 7 ms
Survival time n/a n/a 20 ms
Message size n/a n/a 200 bytes
Video latency n/a n/a <N times
transf. int.
A. Overall PoC- and Site-Specific Challenges
The two production facilities are each equipped with their
own RAN (CPEs / gNB) and MEC capacities. The user data
plane is terminated locally with appropriate network functions
being present in the local edge cloud. As productive real-world
industrial manufacturing may be distributed over multiple
facilities and large geographical distances, an approach for
interconnecting sites via wide area networks is considered as
TABLE II: Functional Requirements
Functional requirement UC-1 UC-2 UC-3
Mobility management X (X)
Energy efficiency X
End-to-end QoS X X X
Network capability exposure X X
Priority, QoS and policy control X
Time synchronization X X
Localization service X
Context-aware network X
Real-time end-to-end QoS monitoring X X X
5G LAN-type service support (X)
Proximity services (X)
Secure remote access X X (X)
Edge computing integration X X X
shown in Fig. 1. Here, the control plane of the network resides
at a central location (e.g., within a public cloud environment),
allowing mobility of terminal equipment, assets and data.
Fig. 1: Testbed architecture of 5G CONNI.
The manufacturing shop floor is an environment typically
characterized by high operational, security and safety stan-
dards. In this regard, the target PoC will consider some
important restrictions, regulations and challenges, which can
be summarized as follows: (1) Planning, design and roll-out
of the 5G network without disturbing the ongoing production
tasks, (2) 5G system integration without disturbing the existing
networking system, (3) Ensuring safety for factory personnel,
(4) Considering special propagation characteristics due to the
shop floor environment, (5) Enterprise IT security regulations,
e.g. to securely provide a service provider access to a corporate
network, and (6) Factory IT security, in particular, the concept
of security zones, as introduced in ISA/IEC 62443 [18].
B. Process Diagnostics by CNC and Sensing Data Collection
In this use case, various sensors will be attached on the
spindle and workpiece with sampling rates between 10 and 400
kHz to collect all necessary physical quantities. As shown in
Fig. 2, collected data will be aggregated into an online monitor
system and transferred via 5G to the analysis system, which is
deployed in the edge data center to conduct process diagnosis.
The updated model parameters and threshold values derived
from the analysis system are transferred back to the online
monitor system to monitor the milling process in real-time.
Fig. 2: Implementation of process diagnostics by CNC and
sensing data collection.
Firstly, one target cell (i.e. CNC lathe, a robot and a con-
veyor) will be investigated. This equipment will be connected

to an industrial PC running the online monitor system. The col-
lected data will be uploaded to the analysis system in an edge
data center using the 5G CPE and gNB. A detailed analysis of
the requirements for this use case will be conducted. Lastly,
the setup will be extended to 11 CNC cells distributed across
the entire shop floor and a network emulator will introduce
different impairments (bandwidth limitations, latency, and
packet loss) to evaluate the impact on the applications.
C. Process Diagnostics using Augmented/Virtual Reality
Fig. 3 illustrates the architecture of the AR/VR use case.
The 5G network will be deployed on the shop floor to connect
equipment (e.g., CNC, robot, conveyor and AGV) with a CPE
to link with the gNB and the edge cloud. The data collec-
tion is achieved by a secured open platform communications
unified architecture client server pair. The control agent is a
software module located in the edge data center. The agent,
implemented by the Unity3D software, will act as a bridge
between user and 3D scene. In this setup, the user device will
be a tablet PC or head-mounted display with a camera module.
Fig. 3: Implementation of AR/VR for process diagnostics.
Using the camera module, the relative position and orienta-
tion between the end user and the equipment can be identified
and transformed into corresponding viewport and navigation
commands to the 3D model. Sensing data can be associated
with corresponding 3D components to show process status
information. For example, vibration levels can be shown on
the 3D model of the spindle as color-coded contour. Machine
users will use the plotted tool paths and other 3D objects,
such as charts or text messages, to optimize the machining
parameters. Performance between users using the conventional
T&E method and users with AR/VR devices will be compared
in terms of total process planning time, cycle time for the
workpiece before and after optimization.
D. Robot Platforms with Edge Intelligence and Control
Fig. 4 depicts the planned PoC for the robot platform with
offloaded control and intelligence. The robot platform consists
of a robot arm with a gripper hand and an industrial camera
system mounted onto the arm or the robot’s stand, whose task
is to record footage to inspect workpiece quality using an
edge cloud-based video analytics function. Since the motion
control functions are offloaded to a nearby edge cloud, a design
consideration is the split of the control function into an outer
and an inner control loop, where the latter will have a min-
imal footprint. Robot arm path planning and motion control
adjustments are directly made in the edge cloud based on the
output of the video analytics function by exchanging values
between the offloaded control and intelligence functions.
Fig. 4: Implementation of robot platform into factory IT.
5G CONNI will validate communications system robustness
(meeting the requirements of Tables I and II for different
video bitrates and transfer intervals), traffic isolation (two user
data traffic streams, control and OAM planes), safety, service
continuity from an application perspective (for which network
capability exposure and context-aware network support is
needed), integration of the 5G network into the factory IT
infrastructure, and compliance with ISA/IEC 62443 [18].
VI. CONCLUSIONS
This paper presents the ambition of the 5G CONNI joint
Taiwan-Europe collaborative H2020 project that aims at de-
ploying a 5G end-to-end testing system to validate the inclu-
sion into 5G Private Networks of 5G key enabling technologies
such as 5G RAN, MEC, 5GC. To this end, we provide first
the 5G CONNI vision on a selection of challenging use cases
relevant to 5G Private Networks. Secondly, we detail on the
identified necessary enhancement of enabling technologies for
industrial applications. Lastly, we give a brief overview of
relevant standardization activities mainly led by 3GPP.
REFERENCES
[1] E. Calvanese Strinati et al., ”6G: The Next Frontier: From Holographic
Messaging to Artificial Intelligence Using Subterahertz and Visible Light
Communication,” in IEEE Vehicular Technology Magazine, Sept. 2019.
[2] 3GPP, TS 22.104 V17.2.0, ”Service Requirements for cyber-physical
control applications in vertical domains; Stage 1 (Release 17),” 2019.
[3] 3GPP, TR 22.804 V16.2.0, ”Study on Communication for Automation in
Vertical Domains (Release 16),“ 2018-12.
[4] ETSI white paper, ”MEC in 5G networks”, 2018
[5] J. Park, S. Samarakoon, M. Bennis, M. Debbah ” Wireless Network
Intelligence at the Edge ”, Proceedings of the IEEE, 2019
[6] M. Merluzzi, P. Di Lorenzo, S. Barbarossa ”Dynamic Resource Allo-
cation for Wireless Edge Machine Learning with Latency And Accuracy
Guarantees ”, in Proc. 2020 IEEE International Conference on Acoustics,
Speech and Signal Processing (ICASSP)
[7] 3GPP, TR 28.801 V15.1.0, ”Study on management and orchestration of
network slicing for next generation network (Release 15)”, 2018-01.
[8] R. Buyya, S.N. Srirama, ”Fog and Edge Computing: Principles and
Paradigms”, 2019-01
[9] S. Barbarossa, E. Ceci, M. Merluzzi and E. Calvanese-Strinati, ”Enabling
effective Mobile Edge Computing using millimeterwave links”, in Proc.
of 2017 IEEE International Conference on Communications Workshops.
[10] N. di Pietro, M. Merluzzi, E. Calvanese-Strinati and S. Barbarossa,
”Resilient design of 5G mobile-edge computing over intermittent mmWave
links”, 2019
[11] 3GPP, TS 23.501 V16.4.0, ”System architecture for the 5G System
(5GS)” 2020-03.
[12] 3GPP, RP-181491, ”Draft report of RAN #80 held in La Jolla, USA,
10.06.-13.06.2018”, 2018-09
[13] 3GPP, SP-200189, ”New WID on Management of non-public networks”,
2020-03
[14] 3GPP, SP-200094, ”Revised SID: Study on enhanced support of Non-
Public Networks”, 2020-03
[15] 3GPP, RP-182138, ”New SI proposal: Study on Channel Modeling for
Indoor Industrial Scenarios”, 2018-09
[16] M. Schmieder et al., ”Directional Wideband Channel Measurements at
28 GHz in an Industrial Environment”, IEEE Globecom 2019
[17] M. Schmieder et al., ”Measurement and Characterization of an Indoor
Industrial Environment at 3.7 and 28 GHz”, EuCAP 2020
[18] International Electrotechnical Commission, ”IEC 62443: Industrial Net-
work and System Security“.