Big Data Analytics use Big Data Analytics use Big Data Analytics use Big Data Analytics use cases cases cases cases Zhao Lifen China Mobile
Big Data Analytics use Big Data Analytics use Big Data Analytics use Big Data Analytics use casescasescasescases
Zhao Lifen
China Mobile
Which areas offer the most promising opportunities for big data and analytics?
Opportunities for big data and analytics
Source: Managing and mining big data (TMF Insight research)
Value of Big Data for CSP
� Improve Customer ExperienceImprove Customer ExperienceImprove Customer ExperienceImprove Customer Experience
� Assist Marketing DecisionsAssist Marketing DecisionsAssist Marketing DecisionsAssist Marketing Decisions � Optimize Network PerformanceOptimize Network PerformanceOptimize Network PerformanceOptimize Network Performance
requirements
Traffic Traffic Traffic Traffic AAAAnalysisnalysisnalysisnalysis
Customer Customer Customer Customer AnalysisAnalysisAnalysisAnalysis
Revenue Revenue Revenue Revenue ManagementManagementManagementManagement
Network Quality and Performance
Analysis
Network Value Analysis
End-to-End Analysis of Service
Customer Behavior Analysis
PINetwork Element Layer
KPINetwork Performance Layer
KQIService quality layer
QOECustomer Perception Layer
CSP should extract value of big CSP should extract value of big CSP should extract value of big CSP should extract value of big
data based on cloud computing data based on cloud computing data based on cloud computing data based on cloud computing
technology and advantages of CSP. technology and advantages of CSP. technology and advantages of CSP. technology and advantages of CSP.
For China Mobile, the following For China Mobile, the following For China Mobile, the following For China Mobile, the following
are our priorities.are our priorities.are our priorities.are our priorities.
Use Case1:Improve
Customer Experience—query
of bill and detailed record
Implementation Architecture
CDR process CDR query� Process CDR queries from Process CDR queries from Process CDR queries from Process CDR queries from all channels, all channels, all channels, all channels,
portalportalportalportal, , , , customer service customer service customer service customer service hhhhotline, selfotline, selfotline, selfotline, self----
service terminals, etc.service terminals, etc.service terminals, etc.service terminals, etc.
� Traditional solution: UNIX server +disk Traditional solution: UNIX server +disk Traditional solution: UNIX server +disk Traditional solution: UNIX server +disk
array,array,array,array,--------High costHigh costHigh costHigh cost,,,,difficult scaledifficult scaledifficult scaledifficult scale----
out,lowout,lowout,lowout,low performance when heavy loadperformance when heavy loadperformance when heavy loadperformance when heavy load,,,,etc.etc.etc.etc.
� Based on HadoopBased on HadoopBased on HadoopBased on Hadoop and and and and HbaseHbaseHbaseHbase, the system , the system , the system , the system
architecture can:architecture can:architecture can:architecture can:
• HighHighHighHigh----efficiency queryefficiency queryefficiency queryefficiency query
• ScaleScaleScaleScale----out easilyout easilyout easilyout easily
• Agile statisticsAgile statisticsAgile statisticsAgile statistics
Use Case2: Assist Marketing Decisions
— customer tags
SGSN
BTS/NodeB
eNodeB
BSC/RNC
GGSN/SAE-GW
GatewayGn
DPI① Interface: Gn
AC
③ Location infoMSC
BTS BSC
Interface: A
BillingSystem
Big Data Analytics System
Billing CDR
CG
Accurate customer tags,including customer’s features& preference, range of activities. This will help us to:• Know customer better• Do customer care• Recommend appropriate
product to customers• Remind customers about
the usage of resources in time
② WLAN logs
Use Case2: Assist Marketing Decisions
— precision marketing
Send e-coupons of the shopping mall to
customers
Send e-coupons of Starbucks Coffee Bar
to customers
Based on customer’s info, including:�tags�sex�age�ARPU�range of activities�usage of Internet�usage of voice call,SMS, etc�social attribute�browsing history on related e-commerce website�Shopping history on related e-commerce website
�Target users of Starbucks Coffee Bar
�Target users of the shopping mall
�Near shopping mall on weekends
�At Central Business District on workdays� There is Starbucks Coffee Shop
nearby(for example, within 200 meters)�At 11:30
1、Choose target usersfor e-coupons
2、Define rules about when to send e-coupons
Big Data
Analytics
System
Precision Marketing Platform
1Analytics and pick up the target users
3Personal Recommend-action Engine① Algorithm② Rules
Choose time slot① Dinner time on
weekdays② Night time near
shopping malls
Marketing Scene Settings① Send electronic
e-coupons
2
Signaling,User location info
User-defined settings:1)receive or not2)parameters
Match and decide the
target users(near real-time)
3、example
Send e-coupons of Starbucks Coffee
Shop to this person
� With the development of digital service, digital traffic puts strain on 2G network.
� Based on the following model, we can analyze the network,,,,service,,,,customer and terminal
info under 2G/3G/WLAN and support the synergic utilization and expansion of these networks.
Use Case3:Optimize Network Performance— synergic utilization and expansion of 2G/3G/WLAN networks
2G cells with high load2G cells with high load endendLow/normal
3G coverage analysis3G coverage analysis WLAN coverage analysisWLAN coverage analysis
Service analysis
Service component
analysis
Service component
analysis
Traffic analysis of
associated TD cells
Traffic analysis of
associated TD cells
Traffic analysis of
associated WLAN hotspot
Traffic analysis of
associated WLAN hotspot
No
High
Performance analysis
Network coverage analysis
YesYesNo
Fit for TD Fit for WLAN
TD quality analysisHigh
Low
Abnormal
normal
TD terminal analysis
few many
AP quality analysisHigh
Low
Abnormal
normal
WIFI terminal analysis
few many