5G ARCHITECTURE
1
Topics Covered
5G Architecture Introduction
5G Network Deployment Options
CU-DU Split and CP-UP Split
2
Introduction
5G has been introduced within the release 15 version of the 3GPP specifications, whereas 4G was introduced within
release 8
5G has been specified based upon the requirements of the following use cases:
enhanced Mobile Broadband ( eMBB)
Ultra Reliable and Low Latency Communications (URLLC)
massive Machine Type Communications (mMTC)
The Radio Access Network (RAN) belonging to 4G is known as Long Term Evolution (LTE), whereas the RAN belonging to
5G is known as New Radio (NR)
NR has been standardized to allow tight interworking with LTE. Tight interworking supports the inter-connection of LTE
and NR Base Stations. These Base Stations can then be used in combination to serve the population of User Equipment
(UE). 5G network architectures based upon tight interworking between LTE and NR are known as Non-Standalone
(NSA)
Non-Standalone architectures allow a smooth and relatively simple evolution towards a complete end-to-end 5G
System (5GS). Non-Standalone architectures allow re-use of existing LTE Base Stations and existing 4G Core Networks.
In general, a software upgrade is sufficient to allow interworking with a set of NR Base Stations
Standalone (SA) NR Base Stations provide connectivity to a 5G Core Network. The combination of NR Base Station and
5G Core Network is known as a 5G System (5GS). The benefits of 5G are maximized when using a SG System
3
5G System
5G System (5GS)
5G Core Network
(5G CN)
5G Access Network (AN)
3GPP Next Generation or Non-3GPP
Radio Access Network
Access Network
(NG RAN)
User Equipment (UE)
The 5G System included 5G Core Network ,the 5G Access Network and the User Equipment.
The 5G Core Network provides connectivity to the internet and to application servers.
The 5G Access Network can be a 3GPP NG-RAN ,or a non-3GPP Access Network
4
Types of Base Stations in 4G-5G Network
LTE Base Station is known as e Node B.
A New Radio Base Station is known as g Node B (next 'generation' Node B).
An LTE Base Station which has been upgraded to allow connectivity with the 5G Core Network is known as
an enhanced eNode B or a Next Generation eNode B
Next Generation eNodeB : It will connect to 5G
Core and it will connect to other gNode B over
Xn interface
5
5G Core Architecture
The 3 most important nodes of Core Network which interact with RAN Network are : AMF
,UPF ,SMF .
AMF : Access and Mobility Function.
UPF : User Plane Function
SMF : Session Management Function
The AMF replaces the MME (of LTE).
We have Gateways today (SGW and PGW) in LTE.
The Gateways have been split in to 2 parts –the user plane and Control plane.
The control plane part of the Gateway is called the SMF.
The user plane part of the Gateway is called the UPF.
In LTE the IP allocation to the UE is done by the PGW , In 5G the IP allocation to the UE is
done by the SMF.
6
5G Architecture
7
3GPP 5G Architecture
3GPP has specified both’ Reference Point’ and ‘Service based’ architecture for the 5G System (5GS )
Point to Point Point to Point
Interface Interface
Network Network Network
Element 1 Set “A” Procedure Element 2 Set “B” Procedure Element 3
Specified for this Specified for this
Interface Interface
Concept of Reference Point system architecture
Service A Service C Service E
Network Network Network
Service F
Function 1 Service B Function 2 Service D Function 3
Concept of Service based system architecture
8
Reference Point System Architecture
Point to Point Point to Point
Interface Interface
Network Network Network
Element 1 Set “A” Procedure Element 2 Set “B” Procedure Element 3
Specified for this Specified for this
Interface Interface
Concept of Reference Point system architecture
The Reference point Architecture is based upon a set of Network Elements.
Reference Point Architecture uses point to point interfaces to interconnect those Network
Elements.
Signaling Procedures are specified between each point to point interface.
LTE NW Architecture is an example of Reference Point Architecture.
9
Service Based Architecture.
In Service based Architecture we have a common bus.
The Nodes are called Network Functions.
The Network Functions connect to the common bus.
Most likely Operators will not go for Point to Point Architecture.
The Common configuration expected is Service based Architecture
Service A Service C Service E
Network Network Network
Function 1 Service B Function 2 Service D Function 3
Concept of Service based system architecture
10
Reference System Point Architecture
Application Function
Network Slice Selection
(AF)
Function (NSSF)
N5
N7
Policy Control Function
Data
(PCF)
Network
N22 N15 (DN)
Authentication Server Function Unified Data Management
(AUSF) (UDM)
N12 N8 N10 N4 N6
N11
Access and Mobility Management Function Session Management User Plane Function
(AMF) Function (SMF) (UPF)
N1 N2 N3
N14 N9
N26
5G Access Network (AN)
Uu
User Equipment (UE)
11
Interfaces
N1: UE to AMF ; N1 is more of a logical interface similar to S1-MME (S1-AP). The
connectivity of UE to AMF is through gNode B –cannot be direct.
N2: g Node B to AMF
N3:g Node B to UPF
N4 : UPF to SMF
N5 : PCF to AF
N6 : UPF to Data Network (outside world)
N7 : SMF to UPF
N8 : AMF to UDF
N9 : Inter UPF
N10 : UDM to SMF
12
Service Based System Architecture
N nef Network Exposure Function Path for User Plane Data
(NEF) Control Plane Network Elements
Network Slice Selection N nssf N nrf Network Function Repository
Function(SMF) Function(NRF)
Data
N pcf N af Network
Policy Control Function Application Function
(PCF) (AF) (DN)
N ausf N udm
Authentication Server Function Unified Data Management
(AUSF) (UDM) N6
N4
Access and Mobility Management N amf N smf Session Management User Plane Function
Function (AUSF) Function(SMF) (UPF)
N26 N1 N2 N3
5G Access Network
Towards MME (AN)
Within 4G EPC
Uu 3GPP Radio Access
Note :Service Based Architecture is applicable to the control plane only User Equipment
The user plane still remains Point to Point only (UE)
13
Service Based Architecture
Service Based Architecture is applicable to the control plane only.
The user plane still remains Point to Point only.
All the functions connect to the Bus- there is no point to point interface defined now.
There is a common bus and through this bus all the Network Functions connect.
The HOD of all the functions is Network Function Repository Function (NRF).
All the Functions give their attendance to NRF saying – that I am a function A and I give this
service.
If any Function need any service it will ask NRF- who gives this service.
So NRF contain details of all Network Elements and what all functions they give.
Most likely all the Operators will go for Service Based Architecture and not for Reference
Point Architecture.
14
5G NETWORK
DEPLOYMENT OPTIONS
15
Standalone vs Non-standalone
NR Supports non-standalone mode at first phase, which NR relies on LTE EPC for
core network functionalities
NR Standalone NR Non-Standalone
Core NW Core NW
Air Interface Air Interface
data data
control control
NR has a dedicated core network NR relies on LTE EPC for core network functionalities
NR does not need to rely on LTE EPC NR UE must be implemented to simultaneously
NR UE can be implemented to support NR only support both LTE and NR
16
5G Deployment Scenarios
17
5G Architecture Options
Support dual connectivity between gNB and LTE eNB / eLTE eNB
Dual Connectivity Single Connectivity
Option 3 (EN-DC) Option 3a (EN-DC) Option 3x (EN-DC) Option 2 (NR Stand-alone)
EPC EPC EPC
NGC
S1-C S1-U S1-C S1-U S1-U S1-C S1-U S1-U
NG-C NG-U
LTE eNB gNB LTE eNB gNB LTE eNB gNB
gNB
Option 7 (NGEN-DC) Option 7a (NGEN-DC) Option 7x (NGEN-DC)
Option 5 (Rel.15 LTE WI)
NGC NGC NGC
NG-C NG-U NG-C NG-U NG-U NG-C NG-U NG-U
NGC
eLTE eNB gNB eLTE eNB gNB eLTE eNB gNB NG-C NG-U
Option 4 (NE-DC) Option 4a (NE-DC) eLTE eNB
NGC NGC
NG-C NG-U NG-U NG-C NG-U
eLTE eNB gNB
eLTE
18
eNB gNB
Stand Alone Architecture
Figure illustrates two Standalone (SA)
Base Station architectures. These are
known as 'option 2' and 'option 5'.
Connectivity towards the AMF is
based upon the Next Generation -
Control Plane (NG-C) interface.
Connectivity towards the UPF is
based upon the Next Generation -
User Plane (NG-U) interface.
19
Multi RAT Dual Connectivity (MR-DC)
Non-Standalone (NSA) Base Stations use Multi-RAT Dual Connectivity (MR-DC) to provide
user plane throughput across both the 4G and 5G air-interfaces.
This naming convention orders the air interface technologies according to Master Node
followed by Secondary Node, i.e. E-UTRA is the Master Node so is named first in EN-DC.
MR-DC : EN-DC ; NE-DC ;NG-N-DC.
NE –DC : NR is the Master ,E-UTRA (LTE) is Secondary.
NGN-DC : NG (LTE eNode B upgraded) is the Master and NR is the Secondary
20
Multi RAT Dual Connectivity (MR-DC)
This requires an eNode B and gNode to to operate together.
The eNode B and gNode B can be connected using non-ideal backhaul, i.e. a realistic
transport connection.
The eNode B and gNode B have their own independent packet schedulers for allocating
resources to the UE
Legacy 4G UE can continue to use the eNode B in the normal way, while newer UE with a
4G/5G capability can take advantage of the Multi-RAT Dual Connectivity.
Non-Standalone Base Stations can be used to provide an evolution path from 4G to 5G.
21
Control Plane in Dual Connectivity
The Control plane Connectivity will always be with the Master.
In EN-DC Control plane connectivity is with E-UTRA (LTE)
In NR-DC Control Plane Connectivity is with NR.
NGEN-DC Control Plane Connectivity is with NGE (LTE upgraded BTS).
In EN-DC MME signaling will come to EUTRA (LTE).
In NR-DC AMF Signaling will come to NR.
In NGEN-DC AMF signaling will come to NGE (LTE upgraded BTS)
22
E-UTRA New Radio Dual Connectivity
The deployment options illustrated (3,3a and 3x) are known as E-UTRA New Radio Dual
Connectivity (EN-DC).
The E-UTRA refers to the 4G air-interface and New Radio refers to the 5G air-interface.
This naming convention orders the air interface technologies according to Master Node
followed by Secondary Node, i.e. E-UTRA is the Master Node so is named first.
23
Option 3
4G eNode B provides control plane connectivity
towards the Core Network and acts as the Master
Node (MN).
The 5G gNode has control plane connectivity across
the X2 interface and acts as the Secondary Node
(SN).
These configurations require the gNode B to
support the X2 interface rather than the Xn
interface.
Option 3 has the drawback of tunneling all user
plane data through the legacy eNode B. It is likely
that the eNode B hardware has been dimensioned to
support the air-interface throughputs offered by 4G.
The hardware may not be capable of supporting
the higher throughputs offered by 5G (20 Gbps)
The PDCP layer within the eNode B (Master Node}
dynamically splits the downlink data between the
eNode B and gNode B. The data allocated to the
gNode B is forwarded across the X2 interface
24
Option 3A
Option 3a resolves the drawback associated with
option 3 by providing user plane connectivity
between the S-GW and gNode B.
Option 3a has the drawback of not supporting the
transfer of application data across the X2 interface.
The coverage provided by the gNode B may be
smaller than the coverage provided by the eNode B
(the gNode B may be using a higher operating band
with higher air-interface attenuation).
A UE may move out of the coverage of the gNode B
making it necessary to switch all data transfer to the
eNode B.
This can be done using a Path Switch procedure
initiated by the eNode B (the eNode B instructs the
MME to request the SGW to switch its downlink data
path from the gNode B IP address to the eNode B IP
address).
However, the Path Switch procedure is relatively
slow and is not a dynamic solution
25
Option 3X.
A third Non-Standalone Base Station
architecture has been developed to overcome
the drawbacks associated with Options 3 and
3a.
This architecture is illustrated in Figure below
and is known as 'Option 3x'.
In this case, the user plane paths are full
meshed between the eNode B, gNode B and S-
GW.
The eNode B remains the Master Node and is
able to control the selection of the downlink
data path from the S-GW, i.e. the eNode B can
provide the MME with the IP address of the
gNode B for some EPS Bearers, while it can
provide the MME with its own IP address for
other EPS Bearers.
If coverage from the gNode B becomes weak
then the gNode B can dynamically forward
data across the X2 interface towards the
eNode B
26
CU-DU Split & CP-UP Split
27
CU-DU Split Base Station
28
CU-DU Split
The Centralized Unit (CU) Distributed Unit (DU) Split Base Station architecture allows the
gNode B to be deployed using two physically separated units.
These two units are connected using an open interface standardized by 3GPP.
It should be possible to use a CU provided by one network vendor and a set of DU provided
by another network vendor.
The CU provides support for the higher layers of the protocol stack.
There is a single CU for each gNode B.
The Centralized Units belonging to multiple gNode B can be implemented using a shared
hardware platform.
Cloud computing and Network Function Virtualization (NFY) can provide benefits when
deploying the population of CU.
29
F1 Interface
The F1 interface connects the CU to each DU.
Both control plane signalling and user plane data are transferred across the F1 interface.
Control plane signalling is based upon the F1 Application Protocol (F1AP) specified within
3GPP TS 38.473.
F1AP uses SCTP over IP and supports Interface Management, UE Context Management and
RRC Message Transfer procedures .
User Plane Data is transferred across F1 Interface using GTP-U over UDP over IP.
A GTP-U tunnel is setup across the F1 interface for each Data Radio Bearer (DRB).
Each tunnel is identified using the combination of {source Tunnel Endpoint Identifier,
destination TEID, source IP address, destination IP address}
30
Upper and Lower Layer Split
3GPP also discussed the possibility of a 'Lower Layer' split and generated TR 38.816.
3GPP concluded not to progress the specification of a 'Lower Layer' split but noted that
organizations outside of 3GPP (OEM) may do so.
The Lower Layer split divides the DU into a Radio Access Unit (RAU) and a Radio Unit (RU).
The Radio Unit is typically an Active Antenna with baseband processing capability
31
Upper and Lower Layer Split
The transport network which connects the Base Station to the Core Network is known as the
Backhaul,
The connection between the CU and DU is known as the Midhaul
The connection between the Radio Access Unit and Radio Unit is known as the Fronthaul.
The Fronthaul has more demanding transport requirements in terms of bandwidth and latency
when compared to the Midhaul and Backhaul.
The Fronthaul is not an interface standardized by 3GPP. Instead, it is typical to use the Common Public
Radio Interface (CPRI) or enhanced CPRI (eCPRI) specifications for the Fronthaul.
CPRI and eCPRI have been developed by a consortium of companies and are intended to complement
the work of 3GPP.
32
CP-UP Separation
CP-UP separation allows independent
scaling of the Base Station hardware.
For example, the UP capability can be
increased to support higher throughputs
without having to increase the CP
capability.
CP-UP separation also provides benefits in
terms of geographic location flexibility.
The CP capability may be centralized using a
cloud based platform, whereas the UP
capability may be distributed to minimize
latency towards the Distributed Unit (DU)
33
CP-UP Separation
Figure illustrates CP-UP separation for the
CU.
The CP includes the RRC and PDCP
protocol stack layers and is responsible for
RRC signaling.
The UP includes the SDAP and PDCP layers
and is responsible for handling application
data.
The CP and UP sections of the CU are
connected using the E1 interface.
The E1 interface is only used for control
plane signaling and does not transfer user
plane data.
34
CP-UP Separation
The connection of multiple CU-UP to a single
DU means that each cell can have a choice
of CU-UP.
A first CU-UP could be deployed at a
centralized location to take advantage of
cloud technology.
This CU-UP could be used for eMBB
applications.
A second CU-UP could be deployed at the
Base Station location to help minimize user
plane latency. This CU-UP could be used for
URLLC applications.
The selection of appropriate CU-UP can
form part of a Network Slicing
implementation
35
CP-UP Separation
The connection of multiple CU-UP to a single
DU means that each cell can have a choice
of CU-UP.
A first CU-UP could be deployed at a
centralized location to take advantage of
cloud technology.
This CU-UP could be used for eMBB
applications.
A second CU-UP could be deployed at the
Base Station location to help minimize user
plane latency. This CU-UP could be used for
URLLC applications.
The selection of appropriate CU-UP can
form part of a Network Slicing
implementation
36
Option 3X-ENDC
37
