View
5
Download
0
Category
Preview:
Citation preview
www.huawei.com
HUAWEI TECHNOLOGIES CO., LTD.
Challenges of 5G Ultra Reliability
Li DehanHUAWEI RAS Technical Expert
ETR -RT 2018
HUAWEI TECHNOLOGIES CO., LTD. 2
Contents
What is 5G Ultra-high Reliability 1
Challenges of 5G Ultra-high Reliability2
Q&A4
Solutions for 5G Ultra-high Reliability3
HUAWEI TECHNOLOGIES CO., LTD. 3
Typical Usage Scenario for 5G
Sources: ITU-R: IMT Vision – Framework and overall objectives of the future development of IMT for 2020 and beyond. Recommendation.(09/2015),
URLLC scenario will bring big change for telecom network architecture.
Ultra-reliable is the basic requirement for entering mission critical
vertical industry.
HUAWEI TECHNOLOGIES CO., LTD. 4
Vertical Industry Requirements
Use Case Requirement text
Factories
of the
Future
Motion control
The 5G system shall support communication service availability exceeding at
least 99.9999%, ideally even 99.999999%.
The 5G system shall support UE speeds up to 20 m/s, even for
communication services with ultra-low latency and ultra-high reliability.
The cyclic data communication service of the 5G system shall be able to
support satisfy the safety requirements according to[*] for safety integrity
level 3 (SIL-3).
Control-to-control
communication (motion
subsystems)
The 5G system shall support communication service availability exceeding at
least 99,9999%, ideally even 99,999999%.
The cyclic data communication service of the 5G system shall be able
support to satisfy the safety requirements according to [*] for safety integrity
level 3 (SIL-3).
Mobile control panels with
safety functions
The 5G system shall support a communication service availability exceeding
at least 99,9999%, ideally even 99,999999%.
Process automation –
process monitoring
The 5G system shall support a communication service availability of about
99,99 % with a data transmission in intervals between 50 ms up to several
seconds.
Sources:3GPP TR 22.804 V1.0.0,Study on Communication for Automation in Vertical Domains *:IEC 61784-3: "Industrial communication networks – profiles – part 3: functional fieldbuses – general rules and profile definitions", 2016
Vertical industry like Factory of the future has ultra-high reliability ,
ultra-high availability, and SIL3 (PFH < 10−7/h) safety requirement.
HUAWEI TECHNOLOGIES CO., LTD. 5
Sources: Petar Popovski, Research Challenges towards Ultra--Reliable Wireless Communications
Many research target on packet loss rate of radio interface
Packet duplication [Popovski] *
Multi-connectivity [Fettweis]
Diversity-oriented approaches
Ultra-reliable of Radio Interface
* Mehdi Bennis, Building the foundations of Ultra-Reliable and Low Latency wireless communication.
HUAWEI TECHNOLOGIES CO., LTD. 6
5G Network Reference Structure
Automotive / Factory slicing will have edge cloud in the network, we
will have big problem to ultra-reliable requirements
Sources: View on 5G Architecture (Version 2.0)
HUAWEI TECHNOLOGIES CO., LTD. 7
• Human error• Software defect• Silence Failure• Overload problem
• COTS hardware • Virtualization layer• Openness/Dynamic• Cross Layer fault localization
• Radio ultra reliability• E2E Service ultra availability• Slicing fault isolation• Service defined availability
5G URLLC
Telecom Cloud
Telecom Box
Challenge of 5G Ultra-high Reliability
For target 5G Ultra reliability, we need to reach telecom grade cloud
reliability at first.
HUAWEI TECHNOLOGIES CO., LTD. 8
DFR in Cloud
Telecom cloud has same cloud native architecture with public cloud,
but all the DFR technology need to be enhanced to meet the telecom
grade reliability requirement.
DFR
Fault
ManagementArchitecture
Overload
Control
Hitless
Upgrade
Stateless VNF
Distribute DB
LB Pool
Fault detection
Fault localization
Self-healing
Fault isolationRedundancy
DFR: Design For Reliability
Fault prevention
Multi-DC/Region
Service
Degradation
Service Rejection
Auto scale in/out
HUAWEI TECHNOLOGIES CO., LTD. 9
Fault Detection: DeCentralized HA (DCHA)
COTS failure is normal,DON’Tlet it make service interruption become normal.
Decentralized Failure Detection
Fast fault detection is the basic feature for guarantee telecom grade
reliability.
VM Cluster Failure detection time 10+s or even 30+s, will cause service interruption
(call drops ).
COTS & large scale will make failures more frequently, then service interruption(call
drops) worsen much more.
DCHA supports sub-second VM failure detection.
Centralized Failure Detection
HUAWEI TECHNOLOGIES CO., LTD. 10
Self-healing: Fault Intelligent Self-Healing (xFISH)
Intelligent self-healing service provides best self-healing strategy, can
support Zero Touch fault management in Cloud.
UE
NFVI
UE
Network
Local Intelligent Self-healing
RAN 1
Local Intelligent Self-healing
RAN 2
Local Intelligent Self-healing
SDN Controller
Local Intelligent Self-healing
5G Core VNF
Central Intelligent Self-healing
XFISH System in Telecom Cloud
HUAWEI TECHNOLOGIES CO., LTD. 11
0
2
4
6
8
10
12
14
16
18
Temporal
Spat
io
tttt RSTM
tM
tT
tS
tR
Data
MiningFailure Analysis
• 20% failures 80% business impact
• Hard to detect, diagnose and recovery
COTS/Decoupling/Virtualization of NFV will make it worse and more challenging
Silent
Failure
Gray
FailureFalse Alarm
COTS,Decoupling
Fast prevention: DM Failure Analysis for gray failure
prediction (DMFA)
Sick but not dead
“Gray Failure: The Achilles’ Heel of Cloud-Scale Systems”Peng Huang, Chuanxiong Guo, Lidong Zhou, etc, HotOS ’17
The fault of hardware which has life limitation (hard disk) can be predicted.
System degradation also can be predicted.
HUAWEI TECHNOLOGIES CO., LTD. 12
Fault Localization: Converged Localization Flow-
based ML (CLFM)
Cross Layer Reliability
Fault Localization
Active Probing
Automated
Correlation Analysis
VM VMVM VM
Key Enabling Technologies
≥ 90% < 1s
Cross layer fault localization is one of the biggest challenge in the NFV.
CLFM uses Hierarchical unsupervised machine learning to localize typical
fault in the NFV cloud in one second.
HUAWEI TECHNOLOGIES CO., LTD. 13
Overload Control: Intelligence OverLoad Control in
cloud (IOLC )
Predict the service change, automatic scale out to avoid overload in the cloud
OLC inside VNF must guarantee the VNF can survive before automatic scale out.
If the resources in the pool is limited, IOLC can borrow resources for overload VNF
form other VNF which is not busy.
Pool
VNF1 VNFi VNFn
IOLC
VM1 VM1 VM2VM3
(halt) VM1 VM2 VM3VM4
(new)VM2 VM3
OLC OLC OLC
IOLC in a Resources Limited Cloud
HUAWEI TECHNOLOGIES CO., LTD. 14
From “5 9s “to “6 9s” and SIL 3
Telecom Cloud Safety Critical*
Software HA
Architecture
and
Technology
• Stateless VNF
• Distributed Database
• LB Pool
• All Active Disaster Recovery
• DCHA
• XFISH
• DMFA
• IOLC
• CLFM
• VM/Container Fault Isolation
• Grey Upgrade
• FMEA/FTA
• Fault Detection & Diagnosis
• Error Detecting Codes
• Diverse Monitor
• Functionally Diverse Redundancy
• Stateless Software
• Graceful Degradation
• Static Resources Allocation
• Semi-formal Methods
• Formal design and refinement Methods
• Event tree analysis
• Fault tree analysis
• Software functional failure analysis
*IEC 61508-3 Functional safety of electrical/electronic/programmable electronic safety-related systems –Part 3: Software requirements
But how to reach ultra high availability and SIL 3? We still need to
borrow some ideas from safety critical industry.
HUAWEI TECHNOLOGIES CO., LTD. 15
E2E Channel Redundancy
Borrow ideas from safety critical industry (TMR/NVP/Voting)
E2E channel redundancy is one choice to guarantee E2E service
high availability, ms level switchover is the first step.
E2E Channel redundancy in 5G network
UE
RAN Channel 1
RAN Channel 2
User Plane
Channel 1
User Plane
Channel 2
Control Plane
Critical App Server
5G Core
HUAWEI TECHNOLOGIES CO., LTD. 16
Different reliability and availability for different slicing, Not Built to Peak
Different slicing has different network resources allocation strategy.
Different slicing has different HA service in cloud.
UE
5G RAN 3
5G RAN 4
UE5G RAN 1
5G RAN 2
5G Core 1
5G Core 3
UE
UE
Basic HA Service Basic HA Service
Higher HA Service
Higher HA Service
5G Core 2
Higher HA Service
Ultra HA Service
Ultra HA Service
Slicing 1
Slicing 2
Slicing 3
Service Defined Availability (SDA) for 5G slicing
5G Core 5
Ultra HA Service
5G Core 4
Ultra HA Service
SDA (Service Defined Availability) for different slicing
HUAWEI TECHNOLOGIES CO., LTD. 17
Ultra high reliability/availability and SIL 3 safety
requirements bring challenges for telecom industry.
“5 9s” reliability of telecom cloud is the foundation of 5G
ultra high reliability/availability.
Ideas form safety critical industry can be borrowed for 5G,
but cost is the big problem.
AI is the interesting direction for increasing the 5G ultra
high reliability/availability.
Conclusions
Copyright© 2011 Huawei Technologies Co., Ltd. All Rights Reserved.
The information in this document may contain predictive statements including, without limitation,
statements regarding the future financial and operating results, future product portfolio, new
technology, etc. There are a number of factors that could cause actual results and developments to
differ materially Sources those expressed or implied in the predictive statements. Therefore, such
information is provided for reference purpose only and constitutes neither an offer nor an
acceptance. Huawei may change the information at any time without notice.
Thank youwww.huawei.com
Recommended