Upload
phungnhu
View
233
Download
0
Embed Size (px)
Citation preview
When AI meets Network- New Business with AI, New Era of Standards and Industry
网络研究部 刘树成 博士,王雅莉 博士Network Research Dept., Dr. LIU Shucheng(Will),Dr. WANG Yali
Outline
New Business with AI: Toward Intent-Driven Network &
Autonomous Driving Network
New Era of Standards: Accelerating Intent-Driven Network &
Autonomous Driving Network
Reduces the probability of human errors.
Automates complex networks.
Accelerates service provisioning.
Makes networks understand
service intentions
Predicts and quickly rectifies faults.
Enables networks to optimize
themselves.
Improves the efficiency of detecting threats.
Enables networks to predict threats.
“AI+Connectivity” Enables Networks to Achieve New Business Goals of Enterprises
Simplicity
Automation brings
simplicity to users
Intelligence
Enables Networks
understand business intent
From Autonomous Driving Vehicle to Network
• Edge intelligence layer: UE MR/network performance/hardware status/topology data and fault data detection
• Local intelligence layer: Intent & Automation engine, and Analytics & Intelligence engine
• Cloud intelligence layer: AI training platform, full lifecycle app, network-wide data lake and expert labeled date
Achieving Autonomous Driving Network by Scenarios
Operation and optimization
Planning and design
Service provisioning
Deployment
Scenarios
• Survey
• Site design
• RF design
• Site verification
• Commissioning and integration
• Base station installation
• Fault management
• Big event assurance
• Performance optimization
• New service provisioning:
coverage-based/experience-based
provisioning
• Service-based network slicing
• L0: assisted monitoring capabilities, dynamic tasks have to be executed manually.
• L1: the system executes a certain sub-task based on existing rules to increase execution efficiency.
• L2: closed-loop O&M for certain units under certain external environments, lower the bar for
personnel experience and skills.
• L3: sense real-time environmental changes, optimize and adjust itself to the external environment to
enable intent-based closed-loop management.
• L4: predictive or active closed-loop management of service and customer experience-driven
networks, resolve network faults prior to customer complaints, reduce service outages and customer
complaints, and improve customer satisfaction.
• L5: closed-loop automation capabilities across multiple services, multiple domains, and the entire
lifecycle, achieving autonomous driving networks.
An example of Network Intelligence Categorization
Huawei Intent-Driven Network Enable Business Intelligent
Intelligence
Intelligent insights into
business intents
Simplicity
Ultra-simplified
network architecture
Industry-oriented
open platform
OpenNew service
provisioning Delay
Network O&M OPEX
Self-diagnosis and
automatic problem
solving rate
Unknown threat
detection rate
38%
80%
99%
99%
Context
Data-driven InsightsIntent-based API
Real-time Telemetry Configure NETCONF/YANG
Intent Engine
Automation Engine
Intelligent Engine
Analytics Engine
ETH PON Wi-Fi Microwave ZigBee Bluetooth 11ax/ac/n… Underlay
Network
Overlay
Network
Network
Service
VXLAN VPNs
Virtual Security VNF On-Premise Intelligence
SDN
Business Logic/Policy
Outline
New Business with AI: Toward Intent-Driven Network &
Autonomous Driving Network
New Era of Standards: Accelerating Intent-Driven Network &
Autonomous Driving Network
Activities in Related Organizations
CCSA: TC1/TC3/TC5/TC6/TC610 started related work in various fields of the network. 2017Q4,TC610-AIAN set up as the network intelligence industry group.
ITUT: 17Q4 sets up FG ML5G in SG13 to focus on machine learning related to 5G networks.
IRTF/IETF: 18Q3 began to discuss intents and intellectualization in IRTF NMRG (Network Management Research Group).
2018Q3 NMRG has held Side Meeting
about AI/ML in Networking and
Workshop about Intent Based
Networking.
Preliminary discussions on topics such
as intent and network intelligence for
IETF to work on are still not clear.
The plan of NMRG is to start with intent
classification, cases, and system
architecture.
Led by the Fraunhofer HHI Research
Center in Germany, the management
team consists of research institutes,
colleges, ICT institutions, and carriers.
There are three groups:
WG1: Use cases, services and requirements
WG2: Data formats & ML technologies
WG3: ML-aware network architecture
Scope including traffic prediction,
resource optimization, user analysis and
profile, and data collection and analysis.
Title of the project that has been started. Responsible company
Case Study of Network Artificial Intelligence Application Scenario
China Mobile / China Telecom / Huawei
Intelligent Network Monitoring and Resource Scheduling Based on Hotspot EventsResearch on Artificial Intelligence Engine for IDC Network
China Telecom
Intelligent network control and traffic optimization China Unicom
Research on SDN/NFV Network Artificial Intelligence Application Architecture
China Telecom / China Mobile / Huawei
Research on the Application of Network Health Analysis and Early Warning
China Unicom / China Telecom / Huawei
Intelligent 5G network Datang
Research on Intelligent Deployment of Network Services (vCPE and vBBRAS)
FiberHome
Massive multiple-input multiple-output (MIMO) China Telecom
Intelligent management and optimization of data centers China Unicom
Since 2017, network intelligence has
become a hot topic in SDN/NFV
industry alliance (TC610).
In Oct 2017, China top operators and
vendors gather together set up the
AIAN group, and won the
enthusiastic participation of
mainstream players in the
communications industry.
Several projects are created about
use case and architecture.
ETSI: 17Q2 network intelligent standard group ENI was founded. In 17Q4, ENI released the industry's first network intelligent white paper.
16Q4 starts to operate and 17Q1 is established to balance the competitors' standard industry actions. The core concepts, phases, and reference architecture of intelligent network intelligence are proposed to provide important inputs for the intelligent network and IDN.
Level 1 Level 2
Use Case
Network Operations
Policy-driven IP managed networks
Radio coverage and capacity optimization
Intelligent software rollouts
Policy-based network slicing for IoT security
Intelligent fronthaul management and orchestration
Service Orchestration and Management
Context aware VoLTE service experience optimization
Intelligent network slicing management
Intelligent carrier-managed SD-WAN
Network AssuranceNetwork fault identification and prediction
Assurance of service requirements
Infrastructure Management
Policy-driven IDC traffic steering
Handling of peak planned occurrences
Energy optimization using AI
Name Rapporteur Company Current Status (FEB-2019)
Use Cases Yue Wang Samsung Rel.2 begun (Rel.1 Published 2018-04)
Requirements Haining Wang China Telecom Rel.2 begun (Rel.1 Published 2018-04)
Context AwarePolicy Modeling
John Strassner Huawei Rel.1 Published
Terminology Yu Zeng China Telecom Rel.2 begun (Rel.1 Published 2018-06)
PoC FrameworkLuca PesandoMostafa Essa
TIMVodafone
Rel.2 begun (Rel.1 Published 2018-06)
Architecture John Strassner Huawei Early draft v0.0.17
Definition of Networked Intelligence Categorization
Luca Pesando TIM Early draft v0.0.7
Leading Core Group: WP and 5 WIs Published are widely referred, 3 PoCs started and 1 proposed, and Intelligence Categorization WIs started
About 50 companies,Including 12 T(VDF/TI/PT/CT/NTT…) UCs in 4 categories 13 sub-cats
Work ItemsProgress: 17Q1 founded, Q4 WP,18Q1 UC/Req etc WIs
PoC – CT PoC -TIM & Samsung
PoC – PTPoC -TLF
PoCReview Team
Officials
Key WIs
Network Intelligence Core Standard Group - ETSI ENI ( Experiential Networked Intelligence )
ENI Vision
ENI
TraditionalTelecom Hardware
Manual
Operation
& Mgmt.
AI-based
Operation
& Mgmt.
Manual
Operation
& Mgmt.
SDN&NFV Telecom Software & Hardware
SDN &
NFV
Step 1 Step 2 Step 3
ETSI ISG NFV ETSI ISG ENI
SDN&NFV Telecom Software & Hardware
ENI PoC
Title PoC Team Members Main Contact Start TimeCurrent Status
(Mar-2019)
Intelligent Network
Slice Lifecycle
Management
China Telecom
Huawei,CATT,DAHO Networks,Intel,China Electric Power Research Institute
Haining Wang Jun-2018 Stage 1 finished
Elastic Network Slice
Management
Telecom Italia S.p.A.Universidad Carlos III de Madrid, CEA-Leti, Samsung R&D Institute UK, Huawei
Marco GRAMAGLIALuca Pesando
Nov-2018 Started
Securing against Intruders and other threats through a NFV-enabled Environment (SHIELD)
TelefonicaSpace Hellas, ORION, Demokritos (NCSR)
Diego R. Lopez Antonio Pastor
Jan-2019 Started
Predictive Fault management of E2E Multi-domain Network Slices
Portugal Telecom / Altice LabsSliceNet Consortium (Eurescom,University of the West Scotland,Nextworks S.R.L,Ericsson TelecomunicazioniSpA,IBM,Eurecom,Universitat Politècnica de Catalunya,RedZinc Service Ltd.,OTE – The Hellenic Telecommunications Organisation, SA,Orange Romania / Orange France,EFACEC,Dell EMC,Creative Systems Engineering,Cork Institute of Technology)
António GamelasRui Calé NA
Discussed,To be Proposed early 2019
ENI PoC review team:
PoC project wiki: https://eniwiki.etsi.org/index.php?title=Ongoing_PoCs
https://eniwiki.etsi.org/index.php?title=PoC#PoC.231:_Intelligent_Network_Slice_Lifecycle_Management
• For generating new scale up/down and converting the intent to suggested configuration.
• LSTM is used for traffic prediction.
AI-based predictor:
TNSM:
CNSM:
• Provides underlay network control to satisfy the network slice requests.
• FlexE and a FlexE-based optimization algorithm are used for underlay network slice creation and modification.
• Provides core network control to satisfy the network slice requests
PoC Project Goal #1: Demonstrate the use of AI to predict the change of traffic pattern and adjust the configuration of network slice in advance.• PoC Project Goal #2: Demonstrate the use of intent based interface to translate tenant requirements to network slice configuration and
intelligent network slice lifecycle management on demand.
Host/Team Leader Team members
ENI PoC project #1: Intelligent Network Slice Lifecycle Management
Orchestration Architecture
Hardware and Software Setup
PoC Team
Network Slice Blueprinting &
Onboarding
Elastic Intelligent Features:
Horizontal and Vertical VNF
Scaling
Intelligent Admission Control
One innovative service through 2
network slices
Virtual Reality application
One eMBB slice for 360 video
One URLLC slice for haptic
interaction
Main Features
https://eniwiki.etsi.org/index.php?title=PoC#PoC.232:_Elastic_Network_Slice_Management
ENI PoC project #2: Elastic Network Slice Management
Time Plan: Proposal submission Dec. 2018, PoC demo target date Feb. 2019
Host/Team Leader:
Team members::
PoC Concept: network security management, Isolation at network layer based on NFV
PoC Goal #1: Demonstrate an AI framework able to detect network attacks over NFV network based on the combination of several Machine Learning algorithms
PoC Goal #2: Present a policy-driven control loop:
ML-based attack detection
Mitigation recipes through an intent-based security policy
Translated & implemented in a NFV environment with security VNFs (vNSF)
PoC Goal #3: Remote attestation technology to avoid device and data collecting corruption or tampering.
https://www.shield-h2020.eu/
ENI PoC project #3: Securing against Intruders and other threats through a NFV-enabled Environment (SHIELD)
15
Title of the project that has been started Responsible company
Case Study of Network Artificial Intelligence Application Scenario China Mobile/China Telecom/Huawei
Intelligent Network Monitoring and Resource Scheduling Based on Hotspot EventsResearch on Artificial Intelligence Engine for IDC Network
China Telecom
Intelligent network control and traffic optimization China Unicom
Research on the Architecture of SDN/NFV Network Artificial Intelligence Application
China Telecom/China Mobile/Huawei
Research on the Application of Network Health Analysis and Early Warning
China Unicom/China Telecom/Huawei
Intelligent 5G network Datang
Research on Intelligent Deployment of Network Services (vCPE and vBBRAS)
FiberHome
Massive multiple-input multiple-output (MIMO) China Telecom
Intelligent management and optimization of data centers China Unicom
At the 2018 Alliance Conference, Mr. Wei's keynote speech, "The New Phase of Network Architecture Reconstruction -The Network of Artificial Intelligence Enablement" emphasized the importance of this direction and became a hot spot in network intelligence.
The 2018Q2 SDN/NFV Industry Alliance officially released the white paper on the network AI application. The above
figure shows the level-1 directory.
SDN/NFV Industry Alliance and ENI Joint Office Network Intelligence Forum
Since 2017, network intelligence has gradually become a hot topic of the SDN/NFV industry alliance. After AIAN group established, the mainstream players in the communications industry participated in the discussion and created WIs.
In the 2018 alliance summit, network intelligence was hop topic from the main conference site to the branch sessions.
During the Beijing meeting of 17Q3 and ETSI ENI, the SDN/NFV Industry Alliance and ENI jointly
organized a network intelligent forum to attract 140+ participants, laying a foundation for the
establishment of the AIAN team.
17Q4, SDN/NFV Industry Alliance -AIAN Team is officially established.
Released the White Paper for the SDN/NFV Industry Alliance Conference in April 18th, 2016.
18Q2, SDN/NFV The industry alliance is incorporated into CCSA and becomes TC610.
The 18Q3 ENI-TC610 network intelligent forum attracted 100+ participants and demonstrated multiple
demos. The first PoC-intelligent network slice PoC of ENI was officially released at the conference.
Progress of TC610 SDN/NFV Industry Alliance -AIAN (Artificial Intelligence Applied to Network )
• Network reconstruction requires manual intelligence.
• Multiple technologies, such as SDN, NFV, cloud computing, 5G, and IoT
• Diversified user requirements require network flexibility and operability.
• The network becomes more complex, heterogeneous, and dynamic.
• Global focus on artificial intelligence (state-of-art):
• ETSI ISG ENI; 3GPP SA2; ITUT FG-ML5G
• Three phases of network AI development:
• Phase 1: Complete a large number of repetitive and inefficient manual labor and human-machine interaction.
• Phase 2: Provides network awareness and prediction capabilities.
• Phase 3: Align with high-level operation intentions and strategies to greatly improve operation economic benefits.
• 11 telecom application scenarios of two categories and four categories:
• Fault alarm, performance optimization, mode analysis, and deployment management
网络人工智能应用场景
网络资源类应用场景 网络业务类应用场景
网络告
警关联
和故障
定位
网络故障告警应用场景 网络性能优化应用场景
网络流
量优化
与拥塞
防控
热点事件分析预测
网络模式分析应用场景 网络部署管理应用场景
用户移动模式分析
网络服务优化部署
智能化
网络切
片编排
管理
网络健康分析预警
基于人
工智能
的网络
节能
无线网
络覆盖
和容量
优化
CCO
5G
Massive
MIMO
自优化
配置
移动网络负载均衡
• Key technologies:• Machine learning, natural language processing, intent API and
strategy, big data technology, artificial intelligence chip• Follow-up work:
• Promote model algorithm design, technology research, and live network pilot in the industry.
Core Contents of the AIAN White Paper
Set up in October 2017. Focus on industry ecosystem forces, jointly carry out artificial intelligence technologies, standards, and industry research, explore new models and new models of artificial intelligence, promote technology, industry, and application R&D, and carry out pilot demonstration, international cooperation is widely carried out to form a global cooperation platform.
Member Conference of the Artificial Intelligence Industry Development Alliance
AI Industry Development Alliance Council
Work Team&Research Team
Gen
era
l team
Steering Committee
Secretariat
Po
licy a
nd
Reg
ula
tion
Wo
rk
Team
Sta
nd
ard
izatio
n a
nd
P
rom
otio
n T
eam
Aca
dem
ic an
d In
telle
ctual
Pro
perty
Wo
rkin
g G
rou
p
Tech
no
log
y&
Ind
ustry
Wo
rk
Team
Wo
rk T
eam
for C
on
verg
en
ce
an
d A
pp
licatio
n o
f Pro
du
ction
, R
ese
arch
, an
d R
ese
arch
Secu
rity W
ork
Team
Asse
ssmen
t an
d C
ertifica
tion
W
ork
Team
Inte
rnatio
nal E
xch
an
ge
Wo
rkin
g G
rou
p
Wo
rkg
rou
p a
nd
So
cial
Resp
on
sibility
Wo
rkin
g G
rou
p
Expert Committee
AI Open Source Open Promotion Team
AI Computing Architecture and Chip Promotion Team
AIIA Organizational Structure
Standardization and Promotion Workgroup -Telecom Project Team
Telecom is an industry domain of AIIA coverage. Currently, the telecom project team is the sub-team of the application work team of the R&D department.
The standard is provided by the CCSA.
Interaction with ITU-T FG ML5G
Project team
Team leader: Cheng Qiang China Information Communication Research Institute
Deputy team leader: Dr. Liao Jun (China Unicom) Deng Chao (China Mobile)
No.Grou
pTopic Led By
Person who leads the project
1Network
Application of Artificial Intelligence in 5G Mobility Management Technology
Telecommunications Science and Technology Research Institute Co., Ltd.
Ai Ming
2Network
Research and application of artificial intelligence to improve network efficiency and word-of-mouth
China Unicom Network Research Institute
Wei Guanglin
3Network
Intelligent network optimization and O&M system AIOpsChina Mobile Communications Corporation Design Institute
Wang Xidian
4Network
Research on Dynamic Broadcast of Massive MIMO Based on Artificial Intelligence
ZTE Communication Co., Ltd. Zheng Li Ming
5Network
Network Knowledge Map and Key Technologies of Network Intelligence
China Mobile Zhu Lin
6Network
Hierarchical Research on Intelligent Capability of Mobile Communication Network
China Mobile Cao Xi
7Securi
tyResearch on Network Countermeasures Based on Big Data
Harbin Institute of Technology
Wang Xin
8Servic
e
Research and Application Practice of Artificial Intelligence Capabilities for Telecom Operators' Market and Service Scenarios
China Mobile Ren Zhijie
AIIA China Artificial Intelligence Industry Development Alliance
Forum on Network Intelligence, Dec’16, Shenzhen, China
• Orange, NTT, CableLabs, China Telecom, China Unicom, China Mobile, NEC, HPE, HKUST etc, 40+ participants joined
• Study of the scope of network intelligence and related technologies, collected ideas of operators and academics
• Participants discussed and get rough consensus on the ENI ISG Proposal
ENI & SDNIA Joint Forum on Network Intelligence, Sep’17, Beijing, China
• China Telecom, China Unicom, China Mobile, Huawei, Nokia, Intel, Samsung, ZTE, H3C, Lenovo, HPE, etc , 140+ joined
• ENI main players presented the progress of ENI
• Operators shared use cases and practices while manufacturers shared solution ideas
• Demo prototype that embodies the ENI concept
• Deep cooperation between SDNIA and ENI was discussed, then SDNIA created AIAN(AI Applied to Network) industry group
ENI & H2020-SliceNet Workshop, Dec’17, London, UK
• Collaboration on PoC and Common partners identified between ENI and SliceNet
• ENI plan to set up a PoC team
• SliceNet then could take PoCs as an opportunity to feed requirements, use cases and input for ENI work programme,
contributing one or more PoC proposals
ENI & 5GPPP MoNArch Workshop, June’18, Turin Italy
• Collaboration on PoC and Common partners identified between ENI and 5G=MoNArch
• Need to Formally submit the PoC Proposal
ENI & CCSA TC 610 AIAN Joint Forum on Network Intelligence, Sep’18, Beijing, China
• ENI main players presented the progress of ENI
• Operators shared PoC experience and demos with support of Manufacturers
• Demo prototype that embodies the ENI concept
ENI Forum in SDN World Congress, Oct’18, Hague, Netherlands
• ENI main players presented the progress of ENI, including introduction, use cases, PoC and architecture
Presentations was made in IETF / IRTF / BBF / ITUT / SDN Congress / ONS/ TMF / MEF / CCSA …
Forum on Network Intelligence, Dec’16
ENI & SDNIA Joint Forum on Network
Intelligence, Sep’17
ENI & SliceNet workshop, Dec’17
Network Intelligence Activities in 2016 - 2018
EcosystemStandard & Industry
ETSI ENI(Concept, UC/Req, Top-level Arch, Categorization, PoC, etc.)
CCSA TC610 – AIAN and AIIA CT Group(Industrial development in China)
ITU-T, MEF, TMF(Policy, Models, Categorization)
IETF, 3GPP(Protocols, Architecture)
BBF, GSMA(Fixed / Mobile Industrial Development)
ITUT ML5G
IEEE ETI( Intelligence Emerging Technology Initiative)
IRTF NMRG
EU H2020 & 5G-PPP Projects
Open Source
Linux FoundationONAP, Acumos, PNDA
(Data collection and decision, AI )
Research
• Expand contributions on existing WIs and PoCs• ETSI ENI established contact with other SDOs and industry
• Organizations: IETF, BBF, MEF, ITU-T, LF, …• Other ETSI groups: NFV, ZSM, NGP, MEC, NTECH, OSM, …
• Cooperate with mainstream operators, vendors and research institutes in Europe, USA and Asia• Guide the industry with consensus on evolution of network intelligence
Further Opportunities in ENI and Related Organizations
Related Organization Activities
IETF-ANIMA ANIMA (Autonomic Networking Integrated Model and Approach) WG on autonomic networking integrated model and approach
GSMAGSM (Global System for Mobile Communications) Association issued a new report, ' Intelligent Connectivity: How the Combination of 5G, AI and IoT Is Set to Change the Americas’, highlighting how the region is set to benefit from the age of 'intelligent connectivity 'or the fusion of high-speed 5G networks, AI and the IoT.
ITU-T Focus on machine learning for future networks including 5G
TM Forum Proposed AI & Data Analytics Project including AI data model for Telco., Management standards for AI, Creating an AI data training repository, and AI maturity model and metrics.
CCSA TC610 AIAN (Artificial Intelligence Applied in Network) industry group & AIIA (Artificial Intelligence Industry Alliance)
IEEE-ETI ComSoc network intelligence Emerging Technology Initiative (ETI)
IRTF NMRG Network Management Research Group (NMRG) provides a forum for researchers to explore new technologies for the management of the Internet.
ITU-T FG-ML5G The Focus Group will draft technical reports and specifications for machine learning (ML) for future networks, including interfaces, network architectures, protocols, algorithms and data formats.
EU H2020 & 5G-PPP Projects SliceNet, SelfNet, 5G-MoNArch
Linux FoundationONAP (Open Network Automation Platform) allows end users to design, manage, and automate services and virtual functions; PNDA (Platform for Network Data Analytics) is a platform for scalable network analytics, rounding up data from multiple sources on a network; Acumos AI is a platform and open source framework that makes it easy to build, share, and deploy AI apps.
Contact Details
ENI Technical Manager:
Dr. LIU Shucheng (Will) [email protected]
Thank you!
ETSI ENI#9 meeting and workshop will be held by Samsung in
Warsaw, Poland, on 09-12 Apr, 2019.
You are welcome to join us!
Looking forward cooperation with ML5G and AIIA!