Conference PaperPDF Available

Campus-Based Full-Scale and Portable Open-Source 5G SA Networks: Prototyping and Experiments

Authors:
Campus-based Full-Scale and Portable Open-Source
5G SA Networks: Prototyping and Experiments
Kamar Mubasier, Frank Y. Li, Jon Anders S. Øgaard, and Marius-Constantin Vochin
Dept. of Information and Communication Technology, University of Agder (UiA), N-4898 Grimstad, Norway
Telenor Maritime AS, N-4841 Arendal, Norway
Department of Telecommunications, University Politehnica of Bucharest (UPB), 061071 Bucharest, Romania
Email: {kamarm; frank.li}@uia.no; jon.anders.ogaard@telenormaritime.com; marius.vochin@upb.ro
Abstract—In light of the thrilling development of future mobile
communication systems and worldwide 5G deployment, there is
a trend toward open-source-based solutions, which could reduce
costs, provide flexibility, enable innovation, and expand know-
how on 5G RAN and 5G CN operations. In this study, we design
and implement two distinct 5G SA open-source-based testbeds.
The first network architecture, a Portable Network, comprises a
Lenovo laptop used in tandem with a USRP B210. The second
network architecture, a Full-Scale Network, is a campus-based
5G network deployed at a lab facility, consisting of a bare-metal
Dell server and a USRP X300. Based on the implementations
of these two network architectures, extensive experiments have
been performed with a COTS UE connected to the network. Our
results provide insight on the performance of such networks with
respect to network connectivity and computing performance.
Index Terms—Private 5G networks, 5G standalone, open-
source software suites, campus-based testbed.
I. INTRODUCTION
The most recent generation of mobile communication sys-
tems, known as the fifth generation (5G), is currently being
deployed on a global scale and is projected to revolutionize
various vertical industries, supporting novel applications such
as augmented and virtual reality, autonomous driving, and
smart cities [1]. To pave the way for prospective applications
and meet the demand for cutting-edge features, 5G offers three
prominent use cases: enhanced mobile broadband (eMBB),
ultra-reliable and low latency communication (URLLC), and
massive machine-type communication (mMTC).
While some major industrial players and mobile network
operators (MNOs) continue to apply proprietary systems with
closed-source software, an evolving trend toward open-source
software suites is beginning to emerge, opening a door for
more adaptability, innovation, and cost-effective solutions [2].
These solutions, which are typically community-driven and
supported by volunteers, are becoming increasingly popular
for developing future mobile communication systems.
Despite the many benefits brought by open-source 5G
standalone (SA) software suites, the open-source community
is nevertheless facing certain challenges, such as lack of
support for a variety of commercial off-the-shelf (COTS) user
equipment (UE) devices. For instance, certain types of COTS
UEs may support the SA frequency band but are unable
to connect to the network due to device-specific limitations.
These limitations include 1) UE operating on specific public
land mobile networks (PLMNs) and 2) UE being factory
locked to a specific band, which prevents users from actively
searching on other frequency bands. This issue varies from UE
to UE, depending on the UE configurations by manufacturers.
In certain instances, manual user modifications are required to
switch to another frequency band and make other adjustments.
In the meantime, the demand for private networks is rising
quickly in parallel with the 5G deployment and many national
authorities are designating certain spectrum bands to offer
interested stakeholders a means of implementing and experi-
menting with private 5G networks. In Norway, for example,
the Norwegian communications authority (Nkom) has since
2022 started to issue 5G trial licenses in the 3.8–4.2 GHz
bands, facilitating research and fully exploring the potential
of 5G technologies based on the concept of private networks.
Motivated by the observed challenges and opportunities
mentioned above, we design and implement a fully functional
5G private network which is located on the campus of the
University of Agder (UiA), operated based on a license we
obtained from Nkom. More specifically, we implement two
different types of 5G SA networks, namely, a Full-Scale
Network and a Portable Network. The Full-Scale Network,
which is composed of a universal software radio peripheral
(USRP) X300 in conjunction with a Dell R650 bare-metal
server, has been deployed in a lab facility on-campus at UiA.
In addition, a Portable Network, which consists of a USRP
B210 for radio access network (RAN) and a custom Lenovo
laptop for core network (CN) has been implemented.
In both networks, open-source software suites serve as the
basis for our implementations and the prototype is based on
a combined implementation of RAN and CN. With this in
mind, there will be an emphasis on three software components,
srsRAN, OpenAirInterface (OAI), and Open5GS. The perfor-
mance evaluation with respect to metrics such as network con-
nectivity, computing performance, and numerical performance
are conducted through real-world empirical experiments [3].
The rest of this paper is structured as follows. After an
overview of the open-source software suites and UE connec-
tivity tools in Sec. II, the design and implementation of the
two networks are presented in Sec. III. The details regarding
the performance assessment metrics, and the performed exper-
iments based on these two implementations are presented in
Sec. IV. Finally, the paper is concluded in Sec. V.
II. SO FT WARE SU IT ES A ND 5G CONNECTIVITY TOOLS
This section summarizes noteworthy open-source imple-
mentations of 5G RAN and CN and a list of tools available
for UE 5G network connectivity experiments.
A. Open-Source RAN and CN Platform Overview
srsRAN Project is an open-source project written in the
C++ programming language and it provides a complete 5G
RAN solution with an open RAN (ORAN)-native architecture.
The software radio systems (SRS) team recently launched a
new codebase in March 2023 under a new repository [4]. This
new release (version 23.3) is separate from its predecessor,
which focused mainly on 4G and 5G non-standalone (NSA).
The srsRAN project is characterized by several key features,
including the provision of both frequency division duplexing
(FDD) and time division duplexing (TDD) modes, the ability
to handle up to 100 MHz bandwidth in single-input single-
output (SISO), and the capability to function with 15/30 kHz
sub-carrier spacing (SCS) (i.e., numerology 0 and 1 in 5G new
radio (NR)). The upcoming release is expected to introduce
several new features, supporting 256-quadrature amplitude
modulation (QAM) in downlink, ORAN 7.2 front-haul, and
Release 17 non-terrestrial network (NTN) features. Currently
this new project supports only USRP devices, but it will
be expanded to accommodate BladeRF and other types of
software-defined radios (SDRs) soon.
OAI is an open-source project written in the C and C++
programming language and it provides 5G RAN and 5G
CN (5GC) implementations, developed by the French-initiated
non-profit OAI Software Alliance. The goal of OAI is to de-
velop a 3rd generation partnership project (3GPP) compatible
5G gNodeB (gNB) RAN stack and 5G CN. Several essential
characteristics of the 5G gNB contain but are not restricted to
its ability to support SCS of 15 to 30 kHz for frequency range 1
and 120 kHz SCS for frequency range 2, as well as bandwidth
of 10, 20, 40, 80, and 100 MHz [5]. Furthermore, the OAI 5GC
supports three deployment modes, namely, minimalist 5GC,
basic 5GC, and Slicing 5GC.
Magma is an open-source project written in the Go pro-
gramming language and it is a 3GPP-compliant project that
supports 4G, 5G NSA, and 5G SA. Magma comprises three
major components: the Access Gateway, the Orchestrator, and
the Federation Gateway. While the access gateway is the
central component that offers network services and policy
enforcement, the orchestrator is a cloud service that is respon-
sible to monitor network performance. The federation gateway
connects the core network to magma by utilizing standard
3GPP interfaces to existing MNO components [6].
Open5GS is an open-source 5GC and evolved packet core
(EPC) implementation written in C and C++. The project
was established in 2017 by S. Lee. Open5GS supports, but
not limited to, the following features: Release-17 compliant,
Internet protocol (IP) version 6 (IPv6) support, multiple packet
data unit (PDU) sessions, voice over long term evolution
(VoLTE) with home subscriber server (HSS)-Cx interface, and
voice over NR (VoNR). However, this project has a few known
(a) Network Architecture I: Portable Network.
(b) Network Architecture II: Full-Scale Network.
Fig. 1: An overview of two 5G network architectures.
limitations, such as no roaming, no emergency calls, and no
inter-working with EPC [7].
B. Tools for Connecting COTS UE to 5G Networks
Many users in the 5G open-source community are expe-
riencing difficulty when trying to connect a COTS UE to a
5G network. This subsection presents an overview of three
functioning applications on the COTS UE that can be utilized
for shifting the preferred network selection mode to 5G NR.
Nettfart is a free and open-source-based speed test appli-
cation provided by Nkom. It works on all platforms including
all web browsers and mobile apps for Android and iPhone
operating system (iOS). Even though its main functionality
is speed testing, it provides valuable additional functions like
imposing the COTS UE to the 5G band.
Network Signal Guru is an Android-based voice and
data service testing tool. The primary feature is that it can
be used to lock the UE to a certain 5G NR band, and it
also supports extra functions such as examining the signaling
procedure between the UE and the network. In order to use
this application and its many features, the COTS UE must be
rooted, and this application requires a monthly fee.
5G Switch is an application available for Android-based
devices, and its primary function is the ability to force UE
switching to the 5G mode. Although this tool is free of charge
and provides the same network force functionality as the
Nettfart tool, it contains many advertisements.
III. SYS TE M DESIGN AND IMPLEMENTATIONS
This section presents an overview of the two prototype net-
works, from architecture, system design, to implementations.
A. System Architecture
The following two network architectures are defined:
Architecture 1 Portable Network: Hardware:
See Tab. I. Software: OAI for 5G RAN (OAI
RAN:2023 w21) and OAI 5GC for 5GC (OAI 5GC:
v1.5.1).
Architecture 2 Full-Scale Network: Hardware: See
Tab. I. Software: srsRAN for RAN (v23.3 (initial re-
lease)) and Open5GS for 5GC (v2.6.4).
TABLE I: Hardware specifications for 5G SA networks.
Specifications Portable Network Full-Scale Network
Host Machine Lenovo Legion 5 Dell R650 Server
CPU Intel i7 10750H Intel Xeon Gold 5315Y
RAM 16 GB 64 GB
Storage 1 TB 4.8 TB
SDR USRP B210 USRP X300
Antenna ANT-5GM2S1-SMA ANT-5GM2S1-SMA
Fig. 2: Portable network: End-to-end connectivity.
B. Design and Implementation Overview
1) Portable Network: This is a lightweight and cost-
effective network that can be transported with relative ease.
A Lenovo Legion 5 Laptop is connected to USRP B210
over a universal serial bus (USB) 3.0 link. This USB 3.0
interface may impose sampling rate and bandwidth restrictions
on USRP B210. Due to the USB protocol overhead and other
hardware factors, the maximum achievable bandwidth may be
less than the analog to digital converter (ADC) processing
capacity of 61.44 MS/s quadrature, which is specified by the
USRP B210 specification [8].
Furthermore, the RAN and CN software suites are both
installed inside the same laptop. The open-source software
suites for both RAN and CN are based on OAI with basic
network functions. The 5GC has access to the Internet through
a wireless fidelity (WiFi) interface. It is worth mentioning
that this particular setup serves as a basic prototype aimed
at showcasing a portable network. In a commercial setting,
the laptop can be applied to certain configurations that require
minimal equipment, such as search and rescue applications.
2) Full-Scale Network: It refers to a more complex, costly,
and stationary setup that is deployed and mounted at an on-
campus 5G lab facility. A Dell R650 bare-metal server is
connected with a USRP X300 using a 10-gigabit ethernet
(GbE) connection. Both the RAN and the CN software suites
are installed on the bare-metal server. The 5GC is connected
to the campus back-haul via a fiber connection. The benefits
of a Full-Scale Network include powerful processing power,
flexibility, fewer stability issues, and the ability to connect to
it from a remote location through a secure socket shell (SSH)
connection.
3) Global connectivity and UE: Both architectures con-
tain a link between the CN and the global Internet through
typical IP connections via the operational campus network.
The configuration of the prototype that has been implemented
consists of a single COTS UE (Oneplus 10 Pro), and two SDRs
(namely USRP B210 and Ettus USRP X300, with one SDR
designated for each implementation). A schematic illustration
Fig. 3: Full-scale network: End-to-end connectivity.
Fig. 4: Illustrates the portable setup including all the components.
of this prototype of these two architectures is depicted in Fig. 2
and Fig. 3, respectively.
C. Implementation 1: OAI 5G RAN and OAI 5GC
Fig. 4 illustrates the implemented entire testbed for the
Portable Network which includes the COTS UE, the SDR,
and the custom laptop.
One main reason of selecting USRP B210 as the main
SDR for this architecture is that B210 is intended for light-
weight and low-cost implementations in comparison with other
USRP models. It has a frequency range of 70 MHz to 6
GHz and employs AD9361, which is a high-performance radio
frequency (RF) transceiver capable of streaming up to 56 MHz
bandwidth. It should be noted that this USRP device can be
connected to a host machine through a USB 3.0 connection,
which poses a minor limitation in terms of the supported
sampling rate, as previously mentioned in Sec. III-B
The laptop in the portable network is running OAI 5G RAN
and OAI 5G CN. The OAI 5G RAN configuration file is tuned
to 5G NR band N77, 106 physical resource blocks (PRBs),
channel bandwidth of 40 MHz, and SCS of 30 kHz.
D. Implementation 2: srsRAN Project and Open5GS
Fig. 5 depicts the entire network setup for the Full-Scale
Network, including USRP X300 and the bare-metal server.
This campus-based 5G network is deployed at the lab facility,
as shown on the right-hand sub-figure of Fig. 5. It is worth
mentioning that the effect generated by USRP X300 is rela-
tively low and complies with the requirements designated by
the Nkom, and it does not affect the commercial operations
of any of the nearby MNOs (Telenor, Telia, and Ice) that are
depicted in the same figure.
Fig. 5: Illustrates the USRP X300 placed on top of the cabinet and
the Dell R650 bare-metal server placed inside the server rack cabinet.
The location of the test facility and its proximity to other commercial
base stations is marked on the right-hand sub-figure.
The Dell R650 server is mounted in a server cabinet as
a rack server. This server may be accessed remotely from
any location through an SSH connection, making it adaptable
for conducting experiments. However, note that in order to
perform experiments with the COTS UE, physical presence at
the lab facility is required.
As mentioned earlier, the Dell R650 server is running both
the 5G RAN (srsRAN) and CN (Open5GS) software suites.
The srsRAN configuration is assigned to the frequency band
of N77, with channel bandwidth of 50 MHz, and SCS of 30
kHz. With such a channel bandwidth and SCS configuration,
it provides a link capacity with 133 PRBs.
The RAN component of this network, USRP X300, is a
high-performance SDR manufactured by Ettus Research and
owned by National Instruments (NI). The SDR can transmit
and receive radio signals up to 6 GHz, supporting baseband
bandwidth up to 160 MHz. This SDR is operated a full duplex
mode, meaning it can transmit and receive simultaneously. The
hardware design consists of two extended daughter-board slots
with the possibility of dual 10 GigE or even 1 GigE high-speed
ports, and the adopted field programmable gate array (FPGA)
in the X300 is the Kintex-7 FPGA [9].
IV. EXPERIMENTS AND NUMERICAL RES ULTS
The performance of the prototype networks has been as-
sessed through comprehensive experiments based on the two
implementations, with the aim of verifying the full operability
of the network in terms of connectivity and performance.
A. Experimemtal Scenarios and Assessment Metrics
Two distinct test scenarios have been identified based on the
two implemented prototypes. While the first scenario focuses
on connectivity and computing resource utilization, the second
scenario concentrates on numerical performance. To perform
our experiments, the following assessment metrics are defined.
Network connectivity is assessed by demonstrating that
a UE is able to connect to the mobile network. For this
test to be deemed successful, the factors involving the UE
Fig. 6: Successful PDU session establishment: Full-scale network.
attach procedure should pass through. Upon successful
attachment to the network, the UE should be given an
IP address and assigned a unique identifier known as the
globally unique temporary identifier (GUTI).
Computing capacity utilization is performed under the
condition that the RAN and CN software suites are
operating in the background and the UE is actively
connected to the network. The metrics of interest are the
central processing unit (CPU) and memory usage over
time, quantified in percentage units.
Numerical performance is evaluated based on four
key metrics, namely, link capacity, latency, jitter, and
packet loss. The purpose of this set of experiments is
to document the performance of the prototype network
and identify any notable occurrences that can potentially
affect the performance of the network.
B. Connectivity and Computing: Portable versus Full-Scale
1) Network connectivity: The connectivity test is demon-
strated with respect to two key aspects, namely, UE attach
registration, and connection information from the 5GC. Note
that these two factors are interrelated. If the attach operation
is successful, the connection information from both the 5GC
and the UE will show that they are connected.
As illustrated in the screenshots shown in Fig. 7a) and
Fig. 7b) for the Portable Network, the UE attach procedure,
including the authentication, security mode, registration accep-
tance, registration completion, and other phases, demonstrates
that the UE has been successfully connected to the network.
For the Full-Scale Network, Fig. 6 demonstrates a successful
PDU session establishment that marks the final stage of the UE
attach procedure, thereby signifying the successful connection
of the UE to the 5G mobile network.
Furthermore, the connection information from the 5GC
is shown in Fig. 8. In this figure, the gNB information
shows status: connected, meaning that the gNB is successfully
connected to the 5GC, and the PLMN is 999,70, which is
identical to the UE PLMN code. The UE information is shown
at the bottom of the Fig. 8, with the 5G mobility management
(5GMM) status as 5GMM-REGISTERED, indicating that the
(a) (b)
Fig. 7: (a) Initial authentication and security mode setup; (b) Additional phases in the UE attach procedure.
Fig. 8: OAI 5GC logs showing the gNB and the UE information.
COTS UE has completed the registration procedure and it is
connected to the network.
2) Computing capacity utilization: The computing resource
consumption experiment is conducted for a total duration of
six minutes, with a 3-sec interval between two consecutive
measurements. A tool, htop, is adopted to monitor the perfor-
mance of the RAN and CN software components in real-time.
The results of the computing performance with respect to
two metrics, CPU usage over time and memory usage over
time, for the Portable Network and the Full-Scale Network,
are shown in Fig. 9 and 10, respectively. The curves in these
figures illustrate the CPU and memory utilization for RAN
(i.e., gNB) and multiple core network functions. Only the
processes that are actively running in the background during
experiments (obtained via htop) are shown in these figures.
From both Fig. 9a) and Fig. 10a), it is evident that the RAN
requires much higher CPU utilization than the core network
functions, suggesting that the RAN is more resource-intensive.
In terms of memory usage over time, the RAN component
requires also a greater amount of memory than the 5GC
functions, as illustrated in Fig. 9b) and Fig. 10b) respectively.
C. Numerical Network Performance
This subsection performs network performance assessment
with respect to link capacity, ping, jitter, and packet loss, and
the experiments are performed based on the Portable Network.
1) Numerical performance: The measurements are taken
four times for a duration of fifteen minutes for each iteration.
(a) CPU usage for the OAI gNB and OAI CN components.
(b) Memory usage for the OAI gNB and OAI CN components.
Fig. 9: CPU and memory utilization: The portable network.
The experimental results for uplink and downlink link capac-
ity, delay, jitter, and packet loss are illustrated in Fig. 11.
(a) CPU usage for the srsRAN gNB and Open5GS CN components.
(b) Memory usage for the srsRAN gNB and Open5GS CN components.
Fig. 10: CPU and memory utilization: The Full-scale network.
The findings indicate that the obtained downlink link ca-
pacity varies from 94 to 98 Mbps whereas the uplink capacity
yields a median value of 4.79 Mbps. With respect to delay
obtained via Ping, the obtained values lie in the range of 25–
30 ms. For all four experiments, a stable jitter of 1 ms is
observed and the achieved packet loss ratio is null.
Fig. 11: Numerical results: Capacity, delay, jitter, and packet loss.
D. Further Discussions
In our current implementations, the RAN and CN are
deployed on a single bare-metal server. However, another
implementation alternative which appears to be attractive in
practice is to disaggregate RAN and CN onto separate bare-
metal servers. As we experienced in our experiments presented
above, the RAN component demands intensive RF processing
and frame decoding capabilities. Accordingly, deploying CN
and RAN components on separate servers may help mitigate
network performance issues, achieve greater numerical perfor-
mance, and provide more flexible CN functions.
Furthermore, we observe that the modulation and coding
scheme (MCS) adapts separately for uplink and downlink to
fluctuating channel conditions, resulting in deteriorated perfor-
mance at some time instances. When the channel condition is
ideal, the MCS automatically adjusts to a higher modulation
scheme (e.g., 64-QAM or even 256-QAM), resulting in higher
throughput, and vice versa. Therefore, it is imperative to de-
sign applications that should be adaptive to channel conditions
in order to ensure quality of service. However, this task is
beyond the scope of this paper.
V. CONCLUDING REM AR KS
This paper showcases the implementations of two distinct
network architectures: Portable and Full-Scale Networks, both
open-source based, using a combination of OAI RAN and
OAI 5GC, as well as srsRAN and Open5GS, respectively.
Through extensive experiments, we demonstrate that a COTS
UE can be successfully attached to the private 5G network on
the NR band N77. We also reveal that the RAN component
is more resource-intensive than the 5GC component. On the
basis of this finding with regard to computing performance, we
recommend that the RAN should be separated from the CN
and migrated to a server with greater processing capabilities.
ACKNOWLEDGMENT
The research leading to these results has received funding
from the EEA NO Grants 2014-2021, under project contract
no. 42/2021, RO-NO-2019-0499 - A Massive MIMO En-
abled IoT Platform with Networking Slicing for Beyond 5G
IoV/V2X and Maritime Services (SOLID-B5G)”.
REFERENCES
[1] S. M. A. Kazmi, L. U. Khan, N. H. Tran, and C. S. Hong, Network Slicing
for 5G and Beyond Networks, Cham, Switzerland: Springer, 2019.
[2] J. A. S. Øgaard, F. Y. Li, G. Jevne, K. Kjellstadli and A. Martian,
“Internet connected open-source based 4G MIMO and 5G NSA systems:
Prototyping and performance,” in Proc. 25th Int. Symp. Wireless Pers.
Multimedia Commun. (WPMC), 2022, pp. 187–192.
[3] K. Mubasier, “UiA 5G private network: Implementation and experi-
ments,” Technical Report (Internal), University of Agder, Norway, Jun.
2023.
[4] srsRAN Contributors, srsRAN Project Update. [Online]. Available: https:
//www.srsran.com/srsran-project-update.
[5] OpenAirInterface, OpenAirInterface 5G Features. [Online]. Available:
https://gitlab.eurecom.fr/oai/openairinterface5g/blob/develop/doc/
FEATURE SET.md.
[6] Magma India Documentation, Magma Services 5G. [Online]. Available:
https://docs.magmaindia.org/services/services.html.
[7] S. Lee, Open5GS Features. [Online]. Available: https://open5gs.org/
open5gs/features/.
[8] Etus Research, About USRP bandwidths and sampling rates.
[Online]. Available: https://kb.ettus.com/About USRP Bandwidths
and Sampling Rates.
[9] Ettus Research, USRP X300 Specifications. [Online]. Available: https:
//www.ettus.com/all-products/x300-kit/.
... For instance, based on proprietary "all of Ericsson in a box", Ericsson private 5G offers 5G standalone (SA) connectivity for both 110 indoor and outdoor environments with micro-or macro-cell radios, providing end-to-end solutions for many applications including Industry 4.0 and smart agriculture [18]. On the other hand, open-source based software implementations relying on universal software radio peripheral (USRP) software defined radios (SDRs) have been a popular approach for many applications including 115 on-campus 5G networks [19]. No matter a private network is built relying on proprietary or COTS hardware, Open RAN based software implementations represent a salient feature for 5G/B5G network deployment and service development. ...
... Fig. 5 presents measured coverages for the implemented private network 960 transmitting in the n7 frequency band (2600 MHz). Based on the empirical results obtained from real-life, a comparison has been made between the coverage estimated by six standardized radio propagation models (ITU-R 1225, ITU-R 528-3, ITU-R 2001-4, ITU-R 1546-6, ITU-R Sonica [78] a Lenovo laptop functioning as the CN and a full-scale static network with a 980 dedicated Dell Server as the CN, have been implemented and tested [19]. The hardware and software components adopted in our private 5G network are listed in Tab. 7. ...
... The experiments performed on this network includes three categories of trials, including connectivity, numerical performance, and service satisfaction respectively. In Fig. 7, we illustrate one example of each category based on the empirical results we obtained [19,70], as a) Connectivity (left): a COTS UE is 990 connected through an Attach procedure; b) DL and UL data rates, round-trip time (RTT), and packet loss (top-right): when a Samsung A52 is connected to a B210 RU; and c) Service satisfaction (bottom-right): three types of services (Netflix, Skype, and Spotify). For network connectivity, it is worth mentioning that with a special permit from Telenor Maritime, our prototype network 995 has been successfully connected to Telenor's commercial home subscriber server 30 Figure 7: Examples of empirical results performed based on UiA's private 5G network: Connectivity, downlink data rate, and service satisfaction. ...
Book
This book provides a comprehensive guide to the emerging field of network slicing and its importance to bringing novel 5G applications into fruition. The authors discuss the current trends, novel enabling technologies, and current challenges imposed on the cellular networks. Resource management aspects of network slicing are also discussed by summarizing and comparing traditional game theoretic and optimization based solutions. Finally, the book presents some use cases of network slicing and applications for vertical industries. Topics include 5G deliverables, Radio Access Network (RAN) resources, and Core Network (CN) resources. • Discusses the 5G network requirements and the challenges therein and how network slicing offers a solution • Features the enabling technologies of future networks and how network slicing will play a role • Presents the role of machine learning and data analytics for future cellular networks along with summarizing the machine learning approaches for 5G and beyond networks
UiA 5G private network: Implementation and experiments
  • K Mubasier
K. Mubasier, "UiA 5G private network: Implementation and experiments," Technical Report (Internal), University of Agder, Norway, Jun. 2023.