04 - 5G Ready Transport Network - Workshop Proximus 2017 12 08 v.1.2 (Huawei)
04 - 5G Ready Transport Network - Workshop Proximus 2017 12 08 v.1.2 (Huawei)
1
Agenda
1 Architecture Evolution
2 Proposed Architecture & Solutions
4 Q&A
2
3GPP 4.5G & 5G Standardization Process
Mar 17
5G Previous Timeline 5G Phase 1 5G Phase 2
Global
Non-SA Full IMT-2020 Launch
5G Current Timeline SA-NR
NR NR
Dec 17 June 18 Dec 19
17H1/H2:5G RFI/RFP Trial 2017H2 TN Ready Feb 2018 5G Trial, aimed at the 2018 Winter Olympic Games
NG-TN/MSAN/WDM RFP 2018: Test & Trial 2019: Ready July 2020, 5G Olympic
commercial use
2017~18: Innovation & Standard 2018: Test & Field Trial 2019: Ready 2020 5G commercial
launch
2017H2: 5G RFI & 5G RAN Trial 2018: Test & Field Trial 2019: Early Deployment 2020 5G commercial
launch
2017H1: IT RFI, 17H2: IT Trial 2018:Field Trial @ IT & ES 2019: Ready 2020 Launched in
developed markets.
20/21 5G commercial
2017/18: Test & Demo 2018/19: Field Trial 2019/20 commercial trial with friendly user
launch
4
Understanding of European SP 5G Deployment Timeline
UCNC
…
…
Urban Inter Site Distance:
2 7 13
Carriers
2T2R 4T4R
… 64T64R 500m 300m … 100m
6
CloudRAN: Future Proof Elastic Architecture
Unified Architecture
for Multi-technologies and Multi-layers
CloudRC LBO
VR Server mAOS
RAN-NRT
CBS
(Centralized Base
Station)
V2V App
Small Cell Macro 5G NR
RAN-RT RAN-NRT
RAN-RT User
UL Plane RAN-RT
RAN-RT DL Control RAN-RT
Plane Gbps 1 ms
Dual-connectivity
7
Standardization with 5G NR
• CU/DU split is the high-layer split option 2 (April 3rd 2017, 3GPP TSG-WG3 Meeting # 95-bis in Spokane, US)
IP Transport, latency >1ms, throughput equivalent to S1 interface Backhaul
CU DU RU
Coordination: CA, CSPC, eICIC
Midhaul
CU DU RU
• CPRI/eCPRI is the low-layer split between baseband and RF
Midhaul Fronthaul
Core CU DU RU
FH Latency requirement
PDCP RU – Radio Unit
Coordination gain
TSG-WG3 # 95-bis in Spokane,
5G add requirements to the Transport network in the
FH BW requirement
RLC-L USA, accepted to use the split Access : Back Haul, High Layer Spllit (HLS, Mid-Haul) and
DU MAC-H
RAN RT mode option 2 which centralized
PDCP / RRC and distributed RLC / Low Layer Split (LLS, Front-Haul) with Bandwidth, Latency,
MAC / PHY in the first 5G version
MAC-L R15. Synchronization
(Consistent with Huawei's
PHY-H recommended type of split on Bandwidth : MH similar to BH, CPRI 100G, eCPRI 25G
CloudRAN) Latency : MH <2ms BH : 10ms, CPRI 100us, eCPRI 100us
Fronthaul
eCPRI RRU
Synchronization : HLS, BH : ,1us , CPRI :10ns , eCPRI : 130ns
PHY-L
CPRI CU-CP/CU-CP separation
RF (3GPP under discussion) High Low High
Flexible
8 function split: Option2/option7/option8
RRU or combined option2&7 or option2&8
NG Core: Reconstruct for Simplified and Flexible Deployment
CP
HSS ANDSF PCRF MME Common DB
Control DB DB DB DB
Plane Control Plane Atomic Functions Library
Reconstruct AMF SMF PCF …
AAA DHCP IMS
DB DB DB Service Framework
NG Core
Optimized infrastructure and Flexible routing are needed
9
Cloud RAN Drives Transport Network Architecture Evolution
NG CORE
CS/PS EPC
CP/UP
Backhaul
RNC RNC UP UP UP
Backhaul CU
CU CU
Midhaul
Backhaul
Centralized Centralized
BBU DU
Fronthaul Fronthaul
10
Cloud-Native Architecture: Provides Flexibility for Diversified
Service
Typical Latency & Distance in E2E Architecture On-demand Deployment for Services
Higher
Efficiency
Regional DC DC
(POC1) IoT Server
< 200 km Latency:10~30m Latency:10~30m NGC-UP NGC-CP
s s RAN-NRT
DC
Aggregation DC Latency:5~10m …
Latency:5~10m Cache
(POC2) s s NGC-UP APP
< 80 - 100 km
RAN-NRT
DC
Latency: 0.5~2ms Latency: 0.5~2ms V2X Server
CO (POC3) NGC-UP …
< 20 km RAN-NRT
CloudBB
Latency: 0.1ms
< 10 km Latency: 0.1ms
11
Agenda
1
Industrial Trends & Challenges
2
Proposed Architecture & Solutions
4 Q&A
12
5G E2E Requirement
Orchestration
E2E Slicing
Central DC
RAN-Real Time
Macro (LTE/5G)
NG Core CP
Regional DC
Local DC
RAN-Real Time 10G/ 50G 200G/400G NG Core UP APP 400G/1T NG Core UP APP
NG Core UP APP
Micro(LTE/5G) RAN-Non Real Time RAN-Non Real Time
IP+Optical RAN-Non Real Time IP+Optical IP+Optical
RAN-Real Time
E2E: 130ns BITS BITS
Pico(LTE/5G)
E2E: 1.5us
ITU-R M.2083-0 – IMT Vision – Framework and overall objectives of the future development of IMT for 2020 and beyond
DRAFT NEW REPORT ITU-R M.[IMT-2020.TECH PERF REQ] – Minimum requirements related to technical performance for IMT-2020 radio interface(s)
13
Proposed 5G Transport Network Architecture
200/400G 400G
eNB
Access Ring
50G
CloudBB
OLT
OXC
OXC
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
14
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Automation Mgmt.Controller
Lifecycle Mgmt. Automatic service provisioning
Vertical Visualized management
IOT, Automotive, Drones,
Netconf/YANG Telemetry
eHealth, Energy, Smart Simplified
City… NGC UP
SRv6 VPN Service Layer Flatten 3 layers
业务连接:EVPN+VxLAN
NR CU NFC
CSG AGG CR SR, protocol simplified
业务连接:EVPN+VxLAN NFC
eMBB slice Unified protection, TI-LFA
mMTC slice
EVPN unify VPN technology
uRLLC slice
Edge DC
4G - eNB Connection Layer FDC/BDC
IPv6 + SRv6
5G – gNB CSG AGG CR
Internet
FMC Access
FlexE FlexE
FlexE FlexE
FlexE FlexE
15
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
T-LDP BGP
BGP
Challenges: Benefit:
Multi-Protocol, Complex Configuration Easy Maintenance w. Protocol Simplification & Unification
Isolated O&M for Each Domain, Low efficiency 100% Topology protection switching <50ms (TI-LFA)
Protocol consolidate help reduce CAPEX & OPEX and easily migrate to SDN
16
DC Centric High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Single Sector Average Rate: Peak Rate/Site 9Gbps Average Rate/Site 3.75Gbps
Core 5*50Mbps+1*1G(100MHz*10 bit/Hz)bps=1.25Gbps
400GE
Single Sector Peak Rate: Single Site
Single Site
5*300Mbps+1*5G(100MHz*50bit/Hz)bps=6.5Gbps Average Rate ×
IP+Optic Peak Rate ×1
9
al Shift
Down to Aggregation
ASG 200/400GE 3 sector Base Station Bandwidth
Single Access Ring: 42.75Gbps
Average Rate/Sector
Peak Rate/Sector 6.5Gbps
1.25Gbps
Single Aggregation Ring Capacity 400Gbps
Acc
ess
50GE and Single Sector Single Sector Single Sector
compatible to Peak Rate ×1 Average Rate ×2 Average Rate ×3 5 Access Rings per Aggr. Node
100GE
6 Sites per Aggr Ring, 4:1 Convergence Ratio
LTE&5G 5*42.75Gbps*6/4=320Gbps
Single Site Peak Rate: Single Site Average Rate:
9Gbps 3.75Gbps
Single Aggr. Ring: 320Gbps
17
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
5G CSG
Macro
AGG CR
Fiber
CSG
100GE
AGG CR
Radio Sites Access PtP or Ring Aggregation DWDM Ring Core DWDM Ring with PtP
with PtP connection connection
Radio : E-Band MW, Fiber, PtP PON, XGS-PON
CSG : Dual 10GE uplinks or 50GE Ring to AGG
AGG : 100G DWDM
CR : n*100G, 400 DWDM
18
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
50GE High Speed Physical Port, the Best Cost Performance Interface
25GE SFP
50GE will widely adopted, server port forecast 2019 > 7M
DRV TOSA
25G NRZ
NRZ
TIA ROSA
50GE QSFP
(Single Lane, small difference with
25GE )
DRV TOSA
50G
PAM4
TIA ROSA
Cost per 10Gbps estimation of optical module
100GE LR4 CFP/CFP2 10
8 10GE
4*25G NRZ 6
MUX
4*DRV 4*TOSA 4
40GE
50GE
2
NRZ 100GE
0
DEMUX
4*TIA 4*ROSA 2017 2019
Ready on 2017 Q4
Based on single lane implementation, 50GE represents the best cost performance, 50GE per 10Gbps cost will be 2/3 of 100GE, and
1/2 of 100GE in the future.
Finisar considered 50G will be the same cost with 10GE in long-term.
19
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Cloud BB
Cloud BB Delay 100~150us
1800V
5G RRU
Interface ETH:25GE
100G N*25G
N*25G 3.Passive Colorless
810 Solution Synchronization +/- 130ns
GE/10GE
eCPRI/CPRI
GE/10GE eCPRI/CPRI
2. Active FO
Solution 2/3/4G BBU 5G RRU
5G RRU
2/3/4G BBU
• High Reliability and Quality OAM
20 • Easy for the future evolution
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
10 20 30 40 50 60 70 80 90 100
Beyond 100GHz
Milli-meter wave bands are introduced and combined with traditional bands MW to provide 10G-level capacity
Urban Suburban Rural
<2Km,10Gbps+ 2-7Km,3~10Gbps >7Km, 2Gbps
Bundling 4 continuous or
fragmentary CS
Low-
priory
service
80GHz E-band 140GHz D-band
20Gbps+ capacity 50Gbps+ capacity Carrier Aggregation (CA) : 4x4 MIMO
High-priory Super Dual Band ~ XGbps in Traditional Band
Millimeter-wave
service
Swap with E-band/D-band Adding E-band Traditional Band with CA/MIMO
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
AAU
OSN OSN
3* 25G eCPRI/CPRI
CPRI
100G
6~12*CPRI 2~8
RRU RRU RRU eCPRI
RRU RRU RRU
RRU RRU RRU 10 km 5G Board 4G Board
RRU RRU RRU 4G Board
FAN
Site BBU
22
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Huawei Patented
Features Active Solution Passive WDM
Colorless WDM
23
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
24
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
mMTC UP
uRLLC UP eMBB UP CP
DCI DCI
Edge DC Metro DC Core DC
3GPP TR 22.866
RTT < 2~4ms (uRLLC)
5G eV2X < 5ms
E2E RTT < 10-12ms (eMBB) 3GPP TR 22.863
Latency streaming 8k 3D video : RTT 10-12ms
RTT < 30-50ms (mMTC)
ACC Convergence
AGG CORE
layer
230km
26
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
1 CU/MEC 1 CU/MEC
CU/MEC CU/MEC AGG1 AGG2 L3
L3 AGG1 AGG2 L3 DC2 L3 DC2
DC1 DC1
Slave Slave
Master Master
Primary Backup Route
Route Routing Protocol
L2 CSG: L3 CSG: 2
2
• Static Routing in BBU • Dynamic Routing in CSG with security
• BFD between BBU and AGG mechanisms
• 50 ms protection could not be • Optimized BFD between CSG and AGG
achieved: • 50 ms protection achieved :
BFD Software slow timers L2 L2 CSG L2 Fast failure Detection L3 L3 CSG L3
BBU ARP learning in AGG TI-LFA
Slow switch-over in BBU 3 Switch-over in NP
Static 3
Routing
DU DU DU DU DU DU
1- DC1 failure : L2 Edge Solution 1- DC1 failure : L3 Edge Solution
L2 Edge CSG : failure not detected by BBU. When DU learns DC2 becomes master, L3 Edge CSG : failure is advertized by AGG1. DC2 becomes master.,
traffic uses non optimal path through AGG1 not suitable for rLLC services advertized to CSG by AGG2. Traffic to DC2 with optimal path.
2- Primary Link failure : 2- Primary Link failure :
L2 Edge CSG : slow detection with software BFD, Slow switch-over to backup route, L3 Edge CSG : Fast failure detection, Fast protection with TI-LFA Fast
ARP learning in AGG2 Service impact Convergence, no Service Impact
3- DU in cell site choose different CU 3- DU in cell site choose different CU
L2 Edge CSG : Traffic to DC2 uses non -optimal path through AGG1 L3 Edge CSG : Traffic to DC2 uses optimal path through AGG2
27
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Standard evolution:
FlexETH
FlexETH Discussion FlexETH Standard
FlexETH (Agreement FlexETH (Agreement (Agreement
and Preparation of Project Start
1.0)Standard Released 2.0)start 2.0)standard
standards (OIF)
Released
28
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
29
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Timing
Scenario services Timing Requirement Impacts Scenario Services Requirement Impacts
Inter-Base station Low Precision of Sync will cause
Basic FDD/TDD 5G Low
±0.05ppm handover switching the basic 5G service failure (low
service frequency(sub- Basic 5G service < ±1.5us
failure frequency)
4G 6Ghz)
Basic TDD TDD cell closed if time
service < ±1.5us error exceeds 10us 5G High Low Precision of Sync will cause
frequency(above Basic 5G service < ±500ns the basic 5G service failure (high
4.5G UL CoMP -6Ghz) frequency)
< ±1.5us Zero gain for CoMP
function 5G Low collaborate
< ±130ns Zero gain for CoMP//SFN
frequency(sub- services(CoMP/
6Ghz) SFN)
Synchronization Design factors: Clock Server Sync precision, number of hops, NE Sync precision
Source: HW RAN
30
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
SyncE + Hop-by-Hop 1588V2 is by far the most reliable solution for Time Synchronization
Full Timing Support (FTS):
G.8275.1, “Precision time protocol telecom profile for phase/time synchronization with full timing support from the network”, June 2016
G.8271.1, “Timing characteristics of enhanced primary reference time clocks”, November 2016 (Network limits)
G.8273.2, “Timing characteristics of telecom boundary clocks and telecom time slave clocks”, January 2017 (Clock specification)
31
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Current Precision PRTC: ± 100ns Transport : ± 1.0us (FTS) End Application: ± 150ns
Target Precision PRTC: ± 40ns Reserved time Error: ± 250ns
Transport : ± 60ns (FTS) End Application: ± 20ns
Reserved time Error: ± 10ns
ETH ETH
Router ETH ETH ETH WDM/
OTN T-BC T-BC
T-BC T-BC < ±30ns < ±30ns
< ±30ns < ±30ns < ±10ns < ±10ns
< ±10ns < ±10ns
<±50ns
< ±20ns < ±100ns
MicroWave < ±100ns < ±40ns
32
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
±50ns
33
Architecture High Bandwidth Low Latency Precise Sync Agile Network Slicing
34
Architecture High Bandwidth Low Latency Precise Sync Agile Network Slicing
1 2 3
Enable service automation Network autonomous: network traffic optimization + Full Lifecycle Network Mgmt
automated troubleshooting
35
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Step1: Controller collects topology, bandwidth resource from network Management Control Analystic&
Slicing lifecycle E2E Connection Slicing Tunnel Assurance
Step2: Controller receives service requirement from GUI/NBI, calculates service path,
Management Management Path Path Traffic
allocates resource Control Control collect
Device Resource Management
Step3: Controller creates VS and deploys dedicate protocol instance for each slice
Step4: Controller deploys slice by create channelized sub-interface and FlexEthernet sub- NCE Platform OS
interface, allocate bandwidth to the sub-interface, associate sub-interface to Virtual-system IaaS&Paas
Step5: Controller deploys VPN inside the VS, so that the service use the resource of the slice
VRF
100M Netconf/BGP-LS/PCEP
Green VLAN1001
VRF VLAN1002
300M
Red
Channelized VLAN1001 FlexEthernet VRF
10G sub-interface Green
sub-interface
VLAN1002
VRF
VRF 30G Red
Blue VLAN1003 VLAN1003
10G VRF
50M
Blue
Access Aggregate
VS #1
ISIS, BGP, RSVP VR Slicing
VS #2
Game Slicing
ISIS, BGP, RSVP
VS #3
IOT Slicing
ISIS, BGP, RSVP
37
Architecture High Bandwidth Low Latency Precise Sync. Agile Network Slicing
Phase 2
Migration Stage VS + FlexEthernet : Network
infrastructure slice
Phase 1 Dedicated NE Hardware, Management
1st Deployment Control isolated with Virtual System
FlexEthernet : Hard-Pipe slice
Dedicated network management slices
Physical interface split into multiple
for vertical industry services in global
independent, isolated time-slots and MAC to
5G deployment in phase 3
ensure service stable and guaranteed
VPN + QOS : Soft-Pipe Slice forwarding latency. VR+FlexEth1: eMBB VR+FlexEth3: mMTC
VPN3:
mMTC
1
Industrial Trends & Challenges
4 Q&A
39
5G Transport Network Solution Implementation Recommendations
Phase1 Phase2 Phase3
Simplified architecture 50GE uplink for 5G site when 26G
10GE uplink for 5G site high spectrum used
High Throughput
50GE@access ring Upgrade to 100GE@access ring
100G DWDM Aggregation/Core 200G/400G DWDM Aggregation/Core
Clock Sync. 1588v2 FTS + Sync. E Transport devices support
Class B Clock precision to class C clock precision
support E2E ±1.5us
Network Protocol & Edge DC @CO
Edge DC IPv6 + SRv6, L3 to ACC SRv6 E2E, SRv6 to DC TOR
VPN +QoS for Services and FlexETH for low latency E2E Slicing mgmt. for massive
Network Slicing SLA isolation service isolation vertical industries
LoRa
FE
2G BTS GE GE(Optical) ESS SR
3G NodeB CISCO ADVA ESS
SR
4G ESS
eNodeB
LoRa
FE
2G BTS 10GE(Optical)
Titan SN
3G NodeB GE CSG LN
4G
eNodeB 10GE(O)
5G gNB
Key Requirements:
- Port scalability: from FE to GE then to 10GE.
- 5G ready: PTP, network slicing.
CSG: Cell Site Gateway
- Low OPEX: PnP, joint one site-visit with mobile.
41
CSG for Macro Site : ATN910C, 1U Highest-density 10GE
Router (Tree Topology)
42
CSG for Macro Site : ATN 950C, 50GE Ready, Smooth Upgrade
to 100GE (Ring Topology)
Multi-service Bearer
Higher Specification F
CXP CXP
ATN950C
PIU
2*10GE, 20*GE 2*10GE,20*GE
A
Routes MAC LSP 2*10GE, 20*GE 2*10GE,20*GE
PIU N
4*10GE,20*GE 4*10GE,20*GE
512K 256K 32K
2U
43
CSG for Small Cell Site : ATN905, Best Choice for Small Cell
Backhaul
1 Compact design
245x180x52(mm) Size as iPAD Weight as Laptop
ATN 905A, 6G capacity, 2.55kg
2GE(O)
Similar with
2GE/FE(O)
2GE/FE Combo
-40~65OC
ETH/OAM
GE/FE Combo
130W PoE
GE/FE (O) ATN905A BTS3911E
ETH/OAM bracket bracket
44
ATN9xx Ready for 5G Evolution: 4 Key Requirements
2
uRLLC Slice 1
3 eMBB Slice 2
Slice 3
SDN Controller
GE Copper mMTC
3G NodeB Netconf/YANG
4
ATN910C 1G/XGE upgrade to 50GE EVPN
10GE Fibre
5G gNodeB
1 BGP
EPC
GE/xGE CO SR
4G eNodeBFibre
ATN910C
10GE Fibre
5G gNodeB
1 Ultra-high bandwidth: bandwidth 10x increase, and Peak 10Gbps of base stations, 8 ports of 10GE/GE, flexible for evolution.
2 Network slicing: network isolation with slicing, and automation management of slicing, To ensuring different service SLAs.
4 Automation: Segment Routing + EVPN Enables Protocol Simplification. Support SDN based Automatic Service Provisioning.
45
Atom GPS Timing System Flexible Installation (Outdoor)
Antenna Feeder
Atom GPS Module ATN 905A
Surge
protector
Ground
cable
46
Atom GPS Timing System Flexible Installation (Indoor)
Atom GPS used in building with Indoor ATN Atom GPS used in base station with Indoor ATN
Atom GPS module
Ensure that the bending radius of the feeder is 20 times the feeder diameter.
Route the feeder along the cable rack that is mounted on the wall and use feeder fasteners to fix the feeder. The
interval between feeder fasteners is 2.5 m.
47
Highlights of Huawei 5G Transport Network Solution
Innovative Cost-effective 50GE Microwave Smoothly Evolution