PosterPDF Available

5G Experimentation Framework: Architecture Specifications, Design and Deployment

Authors:
  • Orange Labs Poland
Poster

5G Experimentation Framework: Architecture Specifications, Design and Deployment

Abstract and Figures

In this paper we present a 5G design and deployment framework within the 5G-EVE European project which aims at interconnecting multiple sites in order to form a single 5G end-to-end facility. Each partner of the project proposes 5G solution elements that will host use-cases brought by verticals. In this paper we present Plug'in as a design framework allowing different partners of the 5G-EVE project and verticals to create and test innovative components for 5G. We also present how components such as VNFs are on-boarded within 5G-EVE and finally we give a quick overview of a vertical's workflow.
Content may be subject to copyright.
5G Experimentation Framework: Architecture
Specifications, Design and Deployment
Louiza Yala, Sihem Cherrared, Grzegorz Panek∗ †, Sofiane Imadali, Ayoub Bousselmi
Orange France, Paris, France
Orange Polska,
Warsaw University of Technology, Warsaw, Poland
first.second@orange.com
Abstract—In this paper we present a 5G design and deploy-
ment framework within the 5G-EVE European project which
aims at interconnecting multiple sites in order to form a single
5G end-to-end facility. Each partner of the project proposes 5G
solution elements that will host use-cases brought by verticals.
In this paper we present Plug’in as a design framework allowing
different partners of the 5G-EVE project and verticals to create
and test innovative components for 5G. We also present how
components such as VNFs are on-boarded within 5G-EVE and
finally we give a quick overview of a vertical’s workflow.
Index Terms—5G, 5G experimentation, NFV, VNF, CNF, xNF
I. INTRODUCTION
The fifth generation (5G) of mobile and wireless com-
munications will constitute a distributed intelligent commu-
nications, sensing, and computing platform. To support this
vision of 5G and be an actor of its evolution, the 5G-EVE1
european project is building an end-to-end facility composed
of four interconnected sites located in Greece, Italy, Spain and
France [1]. Each site is capable of handling the requirements
of various use cases brought by verticals and proposes 5G
solution elements that will be deployed and implemented. The
use cases that 5G-EVE will host are mapped to corresponding
business cases identified by vertical industries and fit the three
main 5G requirements. The use cases are mainly on smart
Transport such as Intelligent railway for smart mobility, smart
tourism based on augmented fair experience and industry
4.0 to enable connected factories. In order to achieve these
use cases, existing technologies such as Network Functions
Virtualisation (NFV) and Software Defined Networking (SDN)
arised to satisfy the requirements of 5G and vertical industries.
In particular, NFV decouples software and hardware of a net-
work function to allow the resulting Virtual Network Function
(VNF) to be implemented by multiple providers and more
easily deployed on generic computing resources [2]. A VNF
is typically pulled from a VNF catalogue and deployed on the
Operator’s mobile network cloud, in an on-demand fashion,
the networking being managed by an SDN controller.
5G-EVE will cover a very broad spectrum of scenarios,
addressing a multitude of 5G PPP use cases [3]. For this
purpose we propose in this paper a 5G experimentation frame-
work divided in two parts: design framework and deployment
1https://5g-ppp.eu/5g-eve/
framework. First, we present Plug’in, our design framework
which is a 5G experimentation platform to develop and test 5G
software and VNF chains. Second, we present the VNF on-
boarding phase. Finally, we present a vertical experimentation
setup and detail relevant workflows of its journey.
II. 5G DES IG N AN D DEP LOY ME NT F RA ME WORKS
The work presented in this paper is a part of the 5G-EVE
European project which aims to build a European 5G end-to-
end facility that will host multiple use-cases to be deployed
by vertical industries (a.k.a. OTTs, which we call verticals in
the rest of the paper). Details of the architecture of 5G-EVE
facilities can be found in our previous work [4].
In 5G-EVE framework, verticals can test and evaluate the
performance of their services in a realistic 5G environment
that replicates the characteristics of an operational 5G in-
frastructure. This approach allows the verticals to verify the
compliance of the relevant network-related Key Performance
Indicators (KPIs) with the expected service performance, be-
fore the deployment of the services in the real operational
networks and their delivery to the final customers. Moreover,
the flexibility of the 5G-EVE framework allows the verticals to
validate the behaviour of their services in different conditions
(e.g. changing the service request load, the traffic,etc.)This is
crucial to support the tuning of the service to guarantee its
performance in different environments and conditions. In this
section, we describe our two-tier system that allows developers
and verticals design and experiment 5G software components.
A. Design framework
In this section we present our design framework ”Plug’in”
a platform that allows the development, integration and test
of 5G software components, also called atoms. Plug’in is
an open-platform that can be used by all 5G-EVE partners
for 5G components development and reuse. In particular, its
functionalities are comparable to those of an Integrated De-
velopment Environment (IDE) and include: automatic project
generation, code version control, online publication store,
automatic documentation generation, and execution and ex-
perimentation environment (i.e. compute, networking and stor-
age resources). These tools are built on top of other tools,
preferably open source, commonly known as DevOps tools to:
version, review and publish the code (e.g. Gitlab, Artifactory);978-1-7281-5127-4/20/$31.00 c
2020 IEEE
schedule continuous integration (CI) and continuous delivery
jobs (e.g. Jenkins CI, Gitlab CI); collect, store and process
logs (e.g. ELK stack); and communicate. In the following
subsections, we describe the design framework tools from a
user perspective as illustrated in figure 1.
Platform tools NFVI VIM/VNFM
AtomStore
Use/Create atoms
AtomDocs
Read/Write
documentation
PlayGround
On demand compute
to Experiment on
atoms Wall
Publish results
Toolboxes
Learn/Deliver
Environments
AtomGen
Generate an atom’s
project structure
Cloud Native
SDK
Develop cloud native
VNFs
Fig. 1. Overview of the Platform’s tools and the underlying technologies
classified according to the ETSI-MANO standard.
1) AtomStore & AtomDocs: Efficiently locating all the
needed tools and information for a research and/or develop-
ment project can be a tedious task. It often involves exchanging
emails, posting on social networks or using search engines
before starting a project. AtomStore is proposed to cope with
this problem. It allows the users to publish their atoms using
either AtomStore’s user interface or Application Programming
Interface (API). AtomStore aims to gather the needed informa-
tion about a project in a single place. It provides an ergonomic
presentation of atoms in multiple categories (e.g. security,
API, networking, monitoring). Each atom is provided with
illustrations; an overview; documentation and code repository.
The goal of the design framework is to provide a self-
care experience for developers and integrators. Hence, a
documentation service, AtomDocs, is also proposed to store
atoms documentation (e.g. usage, configuration, implementa-
tion, contribution guide). The documentation is written by the
atom developers in the form of markdown files. Those files
are automatically compiled and a HTML view is generated
and integrated using of a CI pipeline.
2) AtomGen: Open source software with a fast growth rate
may suffer from code bases divided into multiple packages
structured differently. Due to this variations enhanced by the
great number of contributors, working on multiple modules
might become a challenging task. We propose to simplify
the project’s management and maintainability by introducing
a common project template that fits most needs, generated
using our tool called AtomGen. Its allows 3 options: OS-
level virtualization using Docker, hardware virtualization using
Vagrant Virtual Machines and no virtualization using the user’s
computer as a working environment.
3) PlayGround: Once the user has found a convenient
atom on the AtomStore, he/she can easily deploy it on
the PlayGround, a lightweight Docker in Docker (DIND)
containers orchestrator. This service provides an on-demand
experimentation environment and helps tests reproduction [5].
It creates time-limited sessions where users can create a
number of compute instances. Plugin’s user can use this tool
to instantiate custom or pre-configured images (Sandboxes)
to run experiments. It offers multiple functionalities such as:
save the current session, share the experiment artifacts with the
community, upload files, etc. The PlayGround offers multiple
sandboxes (e.g. Alpine, Ubuntu, CentOS).
4) Cloud Native Software Development Kit (SDK): The
Cloud Native SDK provides an API model and implementation
that can be integrated by users in their development. The
model is based on the ETSI-MANO specification [6] and
allows cloud native VNF design.
As mentioned, the Cloud-native VNFs (CN-VNFs) run in
a container rather than a Virtual Machine, their life cycle is
orchestrated by a container orchestration engine (e.g. Kuber-
netes), using cloud orchestration paradigms.
5) Design framework use cases: The design platform helps
the development of multiple atoms, for example:
Machine Learning For Mobile QoS (ML4MQ) which
allows estimating the reachable throughput by a mobile
user on a 4G network without any data consumption;
The Infrastructure Provider (InP) Registration which pro-
vides an interface to identify, discover and register a new
infrastructure provider and shared resources;
Oko which is an extension of Open vSwitch-DPDK that
provides runtime extension with BPF programs. BPF
programs act as filters over packets: they are referenced
as an additional match field in the OpenFlow tables and
cannot write to packets. They can however read and write
to persistent maps to retain information on flows.
A complete list of the developed software is available in [7].
B. Deployment framework
The deployment framework of 5G-EVE project include the
physical and software infrastructure layer. In this section we
summarize these components and the technologies used in the
project in order to satisfy and achieve vertical’s requirements.
Regardless of site location and device capabilities, 5G-EVE
platform ensures transparent wireless connectivity to answer
the needs of the envisioned use-cases. For example, the choice
of the radio access technology and its configuration parameters
is done in a manner that guarantees seamless connectivity and
ensures the required service constraints.
For all the 5G-EVE sites, physical infrastructure is com-
posed of a set of reusable open-source or proprietary software
components as well as hardware equipment. The software
components include for example, CNFs and VNFs based on
Docker containers or Virtual Machines, and elementary or
composite RAN software functions running as micro-services.
III. VERTICAL EXPERIMENTATION SETUP
The 5G-EVE portal is the architectural entity that serves
the single point of access to the 5G-EVE system and services
for the verticals, abstracting the backend functionalities and
tools composing the 5G-EVE platform. Through the 5G-EVE
portal, the verticals are able to define their experiments, launch
their execution, verify their progress and finally collect the
results.The portal backend takes care of translating the service-
oriented requirements from the verticals into the formal defi-
nition of the vertical service and experiment settings that are
running in 5G-EVE to execute the experiment. The definition
of such elements involves the specification of network service
descriptors, VNFs packages for the single components of
vertical services or experiments’ context, and descriptors of
physical functions available in each of the 5G-EVE facilities.
A. VNF Onboarding
Within the 5G-EVE project, the VNFs onboarding is differ-
ent from one site to another, this is mainly due to the different
orchestrator that the partner chose in order to orchestrate the
VNFs. For example the Italian and the Spanish sites chose
OSM as their main orchestrator, while the French site chose
ONAP as the de facto orchestrator.
In particular in the French site, two types of NSDs arise, first
a high-level service descriptor based on TOSCA2and second a
low level VNF descriptor. In the case of the french site it is an
OpenStack Heat descriptor (in ONAP internal catalogue). As
already mentioned, the French site uses ONAP as orchestrator.
It controls several VIMs, located in different sites. Following
the VNF onboarding demand ONAP North Bound Interface
is triggered by the interworking layer and the latter provides
ONAP the information about the request (service name, VIM
location and where to deploy the resource). For the Italian site
the NSDs are on-boarded either manually using the dashboard
or automatically through OSM northbound REST APIs.
We propose in 5G EVE multiple CNFs provided to the
vertical. These VNFs allow the vertical to experiment 5G
and in order to provide the list of CNFs/VNFs, we first on-
board the VNF Descriptor (VNFDs).The workflow depicted in
Figure 2 detail the onboarding phase of a VNFD in a 5G-EVE.
5G-EVE Portal Multi-site catalogue Catalogue sitex Catalogue sitey
Online periodic check of site catalogues
Start initial synchronization of the multi-site catalogue
Request all VNFDs in 5G-EVE
List of VNFDs
Request all VNFDs in 5G-EVE
List of VNFDs
Loop: if VNFD/NSD not available, add to local catalogue
List of VNFDs
Fig. 2. VNFDs onboarding workflow
B. Vertical’s workflow
The vertical workflow is composed of the following 4 steps:
Step 1: Log into 5G-EVE portal
Step 2: Define the experiment
2http://docs.oasis-open.org/tosca/tosca-nfv/v1.0/tosca-nfv-v1.0.html
Step 3: Define the vertical service by (i) browser tools:
select one or multiple element from the list of compo-
nents, drag&drop the required elements and connect these
elements if necessary or (ii) by blueprints: which are
templates generated for specific services, where a vertical
has to complete the information for its service. The
vertical can either select, among all available Blueprints,
the most suitable for its service or complete the missing
information such as the number of users, availability.
Step 4: Select the conditions of the experiment such as
background traffic and the parameters of the test flow.
Note that while the experiment is running, the vertical can
monitor the components of the experiment and at the end
collect the results.
IV. CONCLUSION & FU RTH ER WO RK
In this paper, we presented a two-tier framework to design
and experiment 5G network functions. Plug’in, our design
framework for 5G experimentation, provides a set of services
to publish, document and experiment a software brick. It
facilitates the integration of VNFs by providing a generic
API model. For the deployment framework, we presented
the components used in 5G-EVE project in order to achieve
the use cases brought by the vertical and its requirements.
Finally we propose a vertical experimentation setup where
we highlight the vertical journey and the back-end VNF on-
boarding process. Currently we are experimenting 5G-EVE
platforms in order to validate 5G KPIs based on use-case.
ACKNOWLEDGMENT
This work was partly funded by the European Commission
under the European Union’s Horizon 2020 programme-grant
agreement number 815074 (5G-EVE project). The paper solely
reflects the views of the authors. The Commission is not
responsible for the contents of this paper or any use made
thereof.
REFERENCES
[1] T. Doukoglou et al., “Vertical industries requirements
analysis & targeted kpis for advanced 5g trials,” in
European Conference on Networks and Communications
(EuCNC), IEEE, 2019, pp. 95–100.
[2] J. G. Andrews et al., “What will 5g be?” IEEE Journal
on Selected Areas in Communications, vol. 32, no. 6,
pp. 1065–1082, Jun. 2014.
[3] M Maternia et al., “5g ppp use cases and performance
evaluation models,5G PPP, vol. 1, 2016.
[4] L.Yala et al., “Testbed federation for 5g experimentation:
Review and guidelines,IEEE Conference on Standards
for Communications and Networking, 2019.
[5] C. Boettiger, “An introduction to docker for reproducible
research,” ACM SIGOPS Operating Systems Review,
vol. 49, no. 1, pp. 71–79, 2015.
[6] ETSI, “Network Functions Virtualisation; Protocols and
Data Models, ETSI GS NFV-SOL2 V2.3.1,” 2017.
[7] AtomStore, https://atomstore.pluginthefuture.eu/. (visited
on 10/16/2019).
Article
Mobile network operators are at the verge of distribution and allotment of existing mobile communications with 5G. It is a high time that we should be focused on the forthcoming sixth generation (6G) networking. Though, it is mandated that 6G would leverage best of the existing network functions and cover an extended geographical range, we need to first understand the importance of 6G. In this article, we discuss about the vision of 6G and elaborate how it should look like with relevant elaborations. Key objective of this paper is to present analysis of the crucial requirements to develop 6G network infrastructure. We present a list of important networking and communication technologies that shall indeed play the vital role to design 6G. We also present how key enablers of 6G shall highlight their significance to emerge 6G. The article also presents important use case scenarios which may be overserved in the 6G era. Lastly, we find open research challenges and discuss way outs. Discussion about future road map designing concludes this literature.
Article
Full-text available
As computational work becomes more and more integral to many aspects of scientific research, computational reproducibility has become an issue of increasing importance to computer systems researchers and domain scientists alike. Though computational reproducibility seems more straight forward than replicating physical experiments, the complex and rapidly changing nature of computer environments makes being able to reproduce and extend such work a serious challenge. In this paper, I explore common reasons that code developed for one research project cannot be successfully executed or extended by subsequent researchers. I review current approaches to these issues, including virtual machines and workflow systems, and their limitations. I then examine how the popular emerging technology Docker combines several areas from systems research - such as operating system virtualization, cross-platform portability, modular re-usable elements, versioning, and a `DevOps' philosophy, to address these challenges. I illustrate this with several examples of Docker use with a focus on the R statistical environment.
5g ppp use cases and performance evaluation models
  • M Maternia
M Maternia et al., "5g ppp use cases and performance evaluation models," 5G PPP, vol. 1, 2016.
Network Functions Virtualisation; Protocols and Data Models, ETSI GS NFV-SOL2 V2.3.1
  • Etsi
ETSI, "Network Functions Virtualisation; Protocols and Data Models, ETSI GS NFV-SOL2 V2.3.1," 2017.