5g/nr architecture is it revolutionary or...
TRANSCRIPT
5G/NR Architecture – Is it
revolutionary or evolutionary?
The 5G-MoNArch view
David M. Gutierrez Estevez
Senior 5G Researcher
Samsung Electronics R&D Institute UK
IEEE CSCN, September 2017
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Initial Thoughts
2/14
Evolution or revolution? It depends how “revolutionary” these key concepts sound:
Network function (NF)-centric view of architecture
Cloudification of networks
However, cloud computing has extensively occupied itself with virtualization
Solutions from cloud frameworks will need to be enhanced considering central/edge cloud
trade-offs:
Centralising RAN provides mux gains, but timescales involved in RAN functions are
much smaller than those considered in cloud solutions.
Cloud resources are typically limited at the edge but centralisation consumes substantial
resources in the x-haul and poses strict requirements on the network
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Initial Thoughts: ETSI NFV
3/14
How “revolutionary” is ETSI NFV architecture?
Many of its concepts are key to 5G architecture
ETSI NFV’s main focus is on management and orchestration aspects that are purely
related to virtualized deployments of network functions defined in other forums (e.g.
3GPP)
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Initial Thoughts: 3GPP SA2
4/14
How “revolutionary is 3GPP’s SA2 architecture?”
Protocol and reference points are defined for each NF
Have both functional behaviour and interfaces.
Can be implemented either as a network element on a dedicated hardware, as a software instance
running on a dedicated hardware, or as a virtualized function instantiated on generic hardware
Separation of control and user planes
Guarantees each plane resources to be scaled independently.
Allows UPFs to be deployed separately from control plane functions in a distributed fashion
AMF PCF
UE (R)AN UPF DN
NG13
NG7
NG3 NG6
NG2 NG4NG1
AFNG5SMFNG11
NG9
AUSF
NG8NG12
UDM
NG10
NG14 NG15
NW Arch w/ Point-to-Point Interface Network Functions
AMF Access & Mobility Management Function
SMF Session Management Function
PCF Policy Control Function
UPF User Plane Function
AUSF Authentication Server Function
UDM User Data Management
AF Application Function
New Interfaces
NG10 SMF – UDM
NG11 AMF – SMF
NG12 AMF – AUSF
NG15 AMF – PCF
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
The 5G-MoNArch View
5/14
EU’s 5G-PPP phase 2 project
5G-MoNArch: 5G Mobile Network Architecture for diverse services, use cases,
and applications in 5G and beyond
Project builds upon key end-to-end architectural concepts of ETSI NFV and 3GPP
SA2 using as basis an end-to-end multi-slice system
Flexible function decomposition to tackle challenges associated to allocation of
NFs to central/edge clouds
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Project Key Facts
6/14
Project runtime: July 2017 –
June 2019 (24 Months)
Leadership team:
Coordinator: Nokia Munich
(Germany)
Technical Management:
Universidad Carlos III de
Madrid UC3M (Spain)
Innovation Management:
Deutsche Telekom,
Darmstadt / Bonn
(Germany)
Effort: 7.7M EUR / ca. 750 PM
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Project Overview
7/14
Approach:
3 enabling innovations,
which fill gaps not addres
sed in 5GPPP Phase1
2 functional innovation
s (one per testbed), com
plementing architecture
with specific functionality
Testbeds:
Vertical industry use case (smart sea port)
Mobile operator deployment (touristic city)
Motivation:
5G architecture: Flexible,
adaptable, programmabl
e
Proof-of-Concept through
real-world testbed imple
mentations
Impact:
Commercial potential for enhanced products and novel services
Paving the way for new market players
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Functional Innovation I: Resilience & Security
9/14
Innovations underway:
Multi-connectivity in virtualised architecture setups with limited computation capabilities,
focusing on reliability rather than throughput
Network coding techniques tailored for reliability in virtualised network architectures
Mechanisms to properly isolate segments of a virtualised environment from network
faults and security threats
Correlation of network function faults in virtualized environments, identifying root
cause
Failsafe operation – 5G islands: Semi-autonomous operation
Efficiency of specialised lightweight cryptographic algorithms when deployed in
virtualised network environments
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Functional Innovation II: Resource Elasticity
10/14
Elasticity: ability to gracefully adapt to load changes in an automatic manner such that at
each point in time the available resources match the demand as closely and efficiently as
possible.
Dimensions of elasticity: VNFs, infrastructure, slices, services.
Three key contributions of resource elasticity functional innovation:
Introduce elasticity at both edge and central cloud taking into account the associated
constraints of the cloud infrastructure and the mobile network
Consider that in 5G cloud resources are shared by different slices and their availability
may change according to the dynamic request of tenants
Provide an edge cloud architecture where computational and storage resources can
be federated, assigned, and scaled where needed according to the load
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Testbeds
11/14
Traffic control/Surveillance
Barges measurements Smart Sea Port
(Hamburg)
Touristic City
Applications
Traffic light control (cMTC): Traffic lights which are connected through wireless connection; reliable and resilient; data integrity
Video surveillance (MBB): Video surveillance required to control entrance to areas, current status of areas, etc
Sensor measurements (mMTC); Sensor measurements on barges which must be connect through wireless terminals
Focus areas:
On-site Live Event Experience by means
of VR
Immersive and Integrated Media: People
will see a part of a touristic city full of real
and imaginary people
Cooperative Media Production: People will
cooperate in real time with imaginary and
real people who are feeling the same VR
experience.
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Additional 5G Architecture Topics
12/14
Artificial Intelligence (AI) for management and orchestration:
Extension of NFV/SDN orchestration principles towards automated network provision
and dynamic resource allocation through application of AI-based autonomics
Optimisation of network performance for multi-network connectivity, network efficiency,
cost per bit, energy consumption, …
Consolidation of service deployment, policy control, resource management, network
monitoring and analysis and prediction under single AI-driven control.
Novel UE-centric roles
Conventional role constrained to supporting eNB/RAN operation without decision power
New 5G architecture and AI may enable new roles:
Slice selection and control
UE as part of the mobility management orchestrator
UE-centric cells: further enhanced D2D, virtual cells
ⓒ 2017. Digital Media & Communications R&D Center. All rights reserved.
Conclusions
13/14
5G architecture will be the result of the complete merging of the
cloud computing and networking worlds Result is a telco cloud that brings together nodes of different nature
(edge and core) and enables flexible allocation of NFs to any of the
nodes within this cloud.
Network slicing paradigm will provide the “logical architectural
units”
Whether that is evolutionary or revolutionary is probably just a
matter or perspective, but cellular architecture is definitely
experiencing quite a few changes