智慧网络论坛

AI in Network Seminar –
Powered by Beta Labs
 Keynote
主题演讲

Henry Calvert
Head of Future Network
未来网络负责人
GSMA
NETWORK AUTOMATION IN 5G ERA
HENRY CALVERT, GSMA
 Overall it is estimated that AI will add $16tr to the global
economy by 2030…

$16tr GDP contribution +14%

By 2030
GDP growth
By 2030

Adoption by Absorption
72% By 2030
enterprise 48%
By 2030
by enterprise

Source: PwC analysis 2018, McKinsey, 2018 notes from the new frontier
6
 Not exhaustive

Operators are exploring on AI in different areas.

The roundtable we will focus on network operations
Operations Services Platforms

Network operations Digital assistants Micro-services

Current Area of Focus

Network planning &

Smart Devices & Robotics Incubator
deployment

Security Marketing & Sales Strategic Investment

Customer care Advertising AI as a Service

Operators propositions tend to be discreet services

 NOT EXHAUSTUVE

AI in the networks: status of deployment of applications

Operations
Planning & Deployment
Security

• Threat / anomaly
Site maintenance
In deployment detection

or Launched Hardware detection Subscriber mobility &

usage patterns
• Fraud detection
• RAN planning and prevention
Load balancing / • VM failure
Traffic prediction prediction
• MiMO QoS degradation
• Fiber roll-out beamforming prediction • Alarm failure
Level of maturity

VOLTE optimisation • Resource

Trials / utilisation • prediction
Enhanced
Pilots carrier VIDEO • Threat hunting
Site Energy Root causes
aggregation optimisation
efficiency analysis
• DDoS
• Orchestration for load prevention
• Ran model for balancing
LAB energy efficiency
Lifecycle
• Dynamic allocation / • Policy composition/
Network slicing Network orchestration management
• AI enabled trouble-shooting
handover Radio QoE prediction for
propagation Real time data path
dynamic allocation

ACCESS NETWORK CORE NOC & SOC*

* Centralised functions that monitors across core and access networks 8

Source: GSMA Analysis
 AI in networks: from automation to zero touch

Closed feedback Autonomous

Automation Era
loop networks
Today Short-medium term Long term

• Automation is pervasive • Realise the promise of self- • Networks adapt continuously

healing, self-optimising learning from experience
• Execution of repeatable
instructions based on well defined • ML orchestrates and input from • AI systems take over decision
rules different models to close the loop making and execution

• ML capabilities are introduced, • Execution still left to rule based • Human supervision, event based
focus on discrete applications policies

9
Source: GSMA Analysis
 ILLUSTRATIVE

AI in networks: from automation to zero touch examples

Closed feedback Autonomous

Automation Era
loop networks
Today Short-medium term Long term
Action taken by human
Traffic VOLTE
rule based automation Domain expert
prediction optimisation
supervision
Feedback
End to End predictions
AI powered decisions
Load Overarching AI
balancing input Feedback
input and
output actions
Resource output
Video utilisation Traffic Volte Video Load Traffic Volte Video Load
optimisation prediction optimisation optimisation balancing prediction optimisation optimisation balancing

10
Source: GSMA Analysis
 However companies need to overcome business
challenges…

1 2 3

AI Skillsets AI infrastructure Data strategy

In-house vs outsource Public/hybrid/private cloud Acquisition, training, managing

4 5 6

Use cases & Organisation Societal and privacy

business case & Culture
Prioritisation
concerns

11
… and technical limitations of the technology

1 2 3

Labelling & Need for massive Difficulties of

training data data sets explaining results

4 5

Generalisability Potential bias

of learnings Data and algorithms

12
 Network operations: examples of holistic approach

In-house development In-house & open source Open source framework

Network Clouds
TANGO AI-powered OSS Jiutian platform
Virtual machine boxes
To improve network CM Network Brain Edge computing
performance and network systems and apps
monitoring Network services
AI marketplace

13
Source: Company websites, press releases
 Network operations: example of applications

Self maintenance /
Predict Peak traffic
Automated AI

VoLTE optimisation MIMO beam forming

Enhanced carrier aggregation/ Failure prediction in

Load balancing virtualised networks

Infrastructure
SD WLAN
maintenance

14
Source: Company websites, press releases
 Network planning and deployment

Planning & Network

Design deployment

Predict best route for Fiber

RAN planning and design
roll-out

15
Source: Company websites, press releases
 Security: Fraud and advanced threat protection

Fraud
Threat Hunting
prediction / Investment
services
prevention

Prevent Subscriber Identity Enterprise services

Module Box (SIM box) fraud

16
Source: Company websites, press releases
 How GSMA is helping operator to tackle these
challenges

Deep dives Applied AI Forum Applied AI initiatives

Strategy
GSMA Global
group led
AI Challenge

Regulation External Affairs &

Public policy Technology Policy
Modernisation Industry Purpose
group led Competition policy, AI activities AI / big data for social good
customer and AI and toolkit

Technology Future Network programme IoT Programme Identity+

group and Sharing best practices on
Pilots on IoT Pilot on risk
programme network automation,
verticals scoring
led energy, etc.

17
 The move to 5G will further push these developments and
increase complexities that will need to be solved

Intent driven Billions of IoT New products and

networks & slicing devices connected services

18
…with integration of AI & 5G across multiple verticals

Entertainment Financial services Health

Energy Manufacturing Agriculture

Transport First response Autonomous vehicles

19
 4 AREAS OF FOCUS IN BETA LABS

AI &
Energy Infrastructure Backhaul
Automation
 BETA LABS LIBRARY -
www.gsma.com/betalabs

Website Blogs & Interviews Case studies

22