1. 4th SG13 Regional Workshop for Africa on “Future Networks for a better Africa: IMT-2020, Trust, Cloud Computing and Big Data” (Accra, Ghana, March 2016)
Big Data – The  Analytic Vision Abdallah Ajlani Ph.D., Principal Consultant, ERICSSON
Byline

2. Ericsson Expert Analytics
Outline
APPROACH TO BIG DATA
EEA USE CASES
INTEGRATION IN OOREDOO TN
© Ericsson AB 2015

3. Analytics is vital piece of our OSS/BSS Software suite
Ooredoo TRM - OSS/BSS in a Networked Society
Analytics is vital piece of our OSS/BSS Software suite
Customer & Partner Interaction
Digital Self-Service
Digital Storefronts
Service Exposure
Customer Service
Sales
Marketing
Analytics
Reporting
Product & Service
Analytics
Customer Analytics
Network
Analytics
Customer & Product Management
Fulfillment Management
Revenue Management
Experience & Assurance Management
User & Partner Management
Order Management
Charging & Billing
Customer, Product, Service, Resource Assurance
Product Catalog
Configuration Management
Mediation
Service Quality Management
Service Catalog
Activation
Policy Control
Trouble Management
Service & Resource Management
Network Management
NFV/SDN/Cloud Management
Network Inventory
Network Design
Network Optimization

4. Telecom Big Data Unique Challenges
Ericsson Expert Analytics
Telecom Big Data Unique Challenges
Facebook
VIP
Roaming
Service
Subscriber
Device
Care
Tethering
Speed
Voice
How to interpret
Incident
Churn
QoE
Coverage
Location
GUMMEI
BCCH
Iups
X2AP
kHz
RNTI
Ic/No
DRX
RTT
QCI
PHICH
NBAP
ISUP
HSDSCH-CC
ARQ
RAB
ICIC
RACH/FACH
RANAF
PSP
QPSK
RSCP
S1APID
CQI
How to correlate E2E
Telecom big data has some specific characteristics that are not present in typical big data systems:
1) Data Technology challenge
Geo distribution
Data is highly cross-referenced: the interpretation of one event depends on another event
No unique identifier
No easy way to partition
Timing and order of events carries a meaning
-> hadoop/mapreduce and other standard data tools cannot be directly applied
2) Interpretation challenge
-> You need to combine data processing science with domain knowledge to make sense of telecom big data
OSS/BSS
CONTENT
NETWORK
RADIO
CORE
© Ericsson AB 2015

5. REAL TIME INSIGHTS AND ACTIONS
September 2015
Device
Device Profile
Tethering
Device usage
Replacement patterns…
Service
OTT usage
Service Profile & Sessions
Service Level Index
Quality Indicators (S- KPIs)…
Subscriber
Subscriber Profile
Service Usage
Lifecycle, CE index, CLV
Location, mobility, roaming…
Network
Ericsson Telecom Data Models & Insights
UNIQUE, END-TO-END EXPERTISE THAT EMPOWERS REAL WORLD USE CASES
Incidents and Root Cause
Traffic, Load and Forecast
Performance
Radio, Core, Data Center…
Networks, SOC, Care, Marketing
Closed loop action, e.g. Self Care
Analytics application ? ?
Query
Real-Time
Human “data scientist”
TELECOM DATA MODELS AND INSIGHTS
Big Data Store ?
Raw data
Stream analysis
Real-Time
Google analogy: we “index” telecom big data the way google is indexing the internet
Closed loop action, e.g. Policy Control
ACTIONABLE INSIGHTS FROM COMBINATION OF OFFLINE AND ONLINE DATA
© Ericsson AB 2015

6. UNIQUE data modeling   Technical Network Performance
September 2015
UNIQUE data modeling
Technical Network Performance
Session QoE per User
User Overall Perception of Service Quality
User Satisfaction
Resource-KPIs & Counters
Service-KPIs & E2E session record
Service Level Index (SLI)
Net Promoter Score (NPS)
When shifting from network to per customer understanding a structured approach is needed. Traditionally service providers are either monitoring the network resources or relying on interviews to get customer satisfaction trends or the Net Promoter Score. But with the user centric approach in Expert Analytics for every session it is possible to bride these two extremes.
This is done by obtaining user centric KPI’s for each session and then predict each users overall satisfaction of the service quality. This slide show how to bridge the gap and with a single indication about satisfaction per user. This is even going beyond the NPS survey satisfaction indicator as the Service Level Index is calculated for each and every user while NPS surveys are for a subset.
Took too long to start
Video didn’t start
Video stopped
Video/App crashed
Buffer time is too long
Video freezes often
Video quality is bad
Part of the video is lost
Accessibility
Video Accessibility
Video Access Time
Retainability
Video Retainability
Integrity
Video Freeze Rate
Video Quality  
Resolve & Retain
Drive ARPU
© Ericsson AB 2015

7. The Service kpi concept
Assurance in CEM
Sales Training Ericsson Expert Analytics
The Service kpi concept
Accessibility e.g.
Video Accessibility
Video Access Time
Retainability e.g.
Video Retainability
Integrity e.g.
Video Freeze Rate
Video Quality
How to measure service experience?
A/R/I For Video
End-users
S-KPIs are not derived from R-KPIs in EEA!
The USP concept serves as a solid foundation for a management system on
which an enterprise can base business and investment decisions, and from
which its service offerings can be expanded.
Implementation of the concept leads to the creation of an enriched set
of network-level information with the granularity of individual customers
and transactions, enabling a wide range of customer and service-centric
applications.
Indicators and probable cause incident data from vast amounts of network
and other data requires a paradigm shift from isolated or ‘siloed’ data, to truly
end-to-end-correlated data from multiple sources, and multiple vendors.
The USP Concept is built-up based on the following:
• Definitions of SySe and S-KPIs, Resource Services (ReSe) and R-KPIs,
and a single value QoSS (Quality of SySe) representing all S-KPIs
• Formal Structure between SySe (with S-KPIs), ReSe (with R-KPIs) and
Resources (with PIs)
• Methodology to give priority and identify S-KPIs and also to obtain operator
unique S-KPI quality targets
• Methodology to classify and identify which resource service is responsible
for a potential S-KPI degradation
A SySe, is a service consumed by the user, for example, TV. It is delivered
by all resources that make possible for an end user to consume this SySe,
for instance, Terminal, fixed or radio or core networks and internet as shown
The SySe entity is created in order to clearly define what Ericsson means
instead of using the commonly used word “service”. A SySe is consumable,
usable and possible to experience by a consumer, a human or a machine. A
terminal provides the interface through which the consumer consumes a SySe.
Services & Apps
Identify
Understand
Act
Network
© Ericsson AB 2015 7

8. Understand the Why of issues
Sales Training Ericsson Expert Analytics
Understand the Why of issues
Determine the most probable cause when an S-KPI has a threshold violation using the Correlate Subscriber QoE with Device, Radio and Core network Resource KPIs (R-KPIs) for each session
Device
“Individual Customer”
Subscription
Coverage
“Holistic”
“Real Time”
End-to-End Session Record (ESR)
User
Service
QoE
Device
Location
Radio
Core
Transport …
App usage
Handovers
Cell load
Web experience
Throughput
PDP connectivity
UE type
Topology
Subscription
Content type
QoE
Interference
Coverage
Video freeze
Service providers
Attach
Roaming
Service Catalog
CRM data
Service quality
Usage
Congestion
CPU load
Radio bearer
Drop/outage
Page delays
Voice quality
QoS class
Load
Charging
Profile
Settings
Probes 2G
…..
Internet
OSS/BSS
CONTENT
NETWORK
ENB
MME
SGW
PDN GW
Terminals
© Ericsson AB 2015

9. Patent-Pending Service Level Index (SLI)
September 2015
Patent-Pending Service Level Index (SLI)
VoLTE Sessions
Service Level Index (SLI)
Predictor of Customer Satisfaction Available for Every User, Continuously Updated
Correlates with Net Promoter Score
Voice / SMS Sessions
Web Browsing Sessions
Video Streaming Sessions
Objective Service-KPIs: Video Streaming
Subjective Weight – Early Adopter
Subjective Weight – Family =
Subjective Weight – Student
Video Start Time
Subjective Weight – Professional X
Video Image Quality
Service Significance X
Incident Repetition X
Memory Fading
When shifting from network to per customer understanding a structured approach is needed. Traditionally service providers are either monitoring the network resources or relying on interviews to get customer satisfaction trends or the Net Promoter Score. But with the user centric approach in Expert Analytics for every session it is possible to bride these two extremes.
This is done by obtaining user centric KPI’s for each session and then predict each users overall satisfaction of the service quality. This slide show how to bridge the gap and with a single indication about satisfaction per user. This is even going beyond the NPS survey satisfaction indicator as the Service Level Index is calculated for each and every user while NPS surveys are for a subset.
Video Freeze
Video Retainability
© Ericsson AB 2015

10. SLI Individualized Scoring Based on user grouping
Ericsson Expert Analytics
SLI Individualized Scoring Based on user grouping
UG1
UG2
UG3
UG4
© Ericsson AB 2015

11. Ericsson Expert Analytics
Case Study SLI Correlation with NPS
© Ericsson AB 2015

12. Ericsson Expert Analytics
Outline
ERICSSON APPROACH TO BIG DATA
EEA USE CASES
INTEGRATION IN OOREDOO TN
© Ericsson AB 2015

13. Proactive Service Operations
CEM & Analytics
Proactive Service Operations
June 2015
Productized Use Cases
KEY HIGHLIGHTS:
EEA Service Operations Center (SOC) enables Operators & Enterprises to:
Know the Impact – How many Customers are impacted, and where?
Understand the Impacted – Who is impacted? What do they have in common?
Improve the Impacted – What are the “Most Probable Causes”? What is the “Next Best Action”?
KEY BUSINESS BENEFITS:
Minimize time to resolution through “proactive operations”
High efficiency in operations management
Deflect / reduce call volumes to contact centers
Specific Users
E2E Real- Time Insights
Leading Usability
Service KPIs
Patented SLI
© Ericsson AB 2015

14. Empowered Customer Care
CEM & Analytics
Empowered Customer Care
June 2015
KEY HIGHLIGHTS:
Proactive handling of customer experience issues in Customer Care
Predict Reason for Call – Why might the customer be calling? How to resolve the call?
Validate the Complaint – Did the customer really have an issue? If so, when and where did it happen?
Take Action – Identify the “Most Probable Cause” and trigger the “Next Best Action”
KEY BUSINESS BENEFITS:
48% Reduction of Average Handling Time (AHT)
35% Increase in First Call Resolution Rate
Reduced number of trouble tickets created
Increase Net Promoter Score
subscriber
Incident
timeline
Leading Usability
Patented SLI
This capability allows Customer Care organization to better achieve its business goals of reducing Average Handling Time (AHT); improving First Call Resolution rate (FCR) and improving NPS by delighting the customer with visibility, understanding and next-best-actions
Service KPIs
Most probable
cause
© Ericsson AB 2015

15. Executive insights / Dashboard
Ericsson Expert Analytics
Executive insights / Dashboard
KEY HIGHLIGHTS:
Enable executives to have a birds-eye view of real time subscriber experience and satisfaction
KEY BUSINESS BENEFITS:
Experience & SLI – How happy and satisfied are your customers now, by service type?
Trend and Causes – What is making your customers unhappy, and what are the causes?
Subscriber Impacts – What is impacting your subscribers now? Where and why?
Focus Area – Closely monitor and track new product and service launches and associated customer experiences
© Ericsson AB 2015

16. Experienced Based Marketing
CEM & Analytics
Experienced Based Marketing
June 2015
KEY HIGHLIGHTS:
Utilize unique EEA insights about each subscriber to drive revenue and reduce churn
What is the satisfaction of each and every subscriber? (Service Level Index)
What is their profile and behavior (e.g. Social media fanatic, Heavy Netflix Usage in evening)
KEY BUSINESS BENEFITS:
Drive ARPU
Which customers are most satisfied and most likely to buy additional services and products?
What Upsell/Cross-sell offers are best fit?
Reduce Churn
Which customers are most unhappy, and what are the contributors to their dissatisfaction?
What can be offered to make them stay?
© Ericsson AB 2015

17. DYNAMIC EXPERIENCE MGMT
CEM & Analytics
DYNAMIC EXPERIENCE MGMT
June 2015
KEY HIGHLIGHTS:
Real time analytics to identify highly valuable customers who are receiving low QoE
Real time actuation of policy adjustments via PCRF to alleviate QoE problem
Pre-integration with Ericsson SAPC
KEY BUSINESS BENEFITS:
Ensure high QoE for most valuable customers
Retain the highest revenue providing customers
© Ericsson AB 2015

18. Ericsson Expert Analytics
Analytics use cases
SoC: Proactive Service Operations
SoC: VoLTE Analytics
CC: Improve call resolution in customer care
Marketing Analytics:
Dynamic Experience Management
SoC: Real-Time Enterprise SLA Assurance
Marketing Analytics:
Customer Retention & Drive Spend
Use cases: Icluding real-time video experience management, enterprise SLA, customer retention and upsell, location data monetization and OTT application analytics
Improve 1st call resolution in customer care Optimize experience for valuable customers Customer retention with improved marketing Optimize network after value and experience Improve experience for enterprise customers
© Ericsson AB 2015

19. Ericsson Expert Analytics
Outline
ERICSSON APPROACH TO BIG DATA
EEA USE CASES
INTEGRATION
© Ericsson AB 2015

20. ANALYTICS ARCHITECTURE
Sales Training Ericsson Expert Analytics
ANALYTICS ARCHITECTURE
JMS actuation**
Trouble ticket request**
Operator’s
Map server*
3PP Exposure and Analytics tools *
Exposure and Visualization Layer
Web service
(Api’s)
Customer Care Views
Service Operations Views
Executive Dashboard
Dynamic QoE Management Views
Marketing Analytics
Web application server
Storage Layer
Online repository (GREENPLUM)
Offline repository
(HADOOP)
In memory data grids
Processing Layer
SNCD loader
Incident
generator
Esr data loader
Aggregator
Stream analytics
Data
loaders
Real time Actuation
PCRF
Subscriber Profiler and Sli calculation
Reference data
CRM
Cell info
IMEI TAC
Billing
Correlation Layer
Extended session records
Real time event stream
Real Time Correlator
E/// CTR
adapters
E/// EBM
Generic KPI
Interface
E/// ctum
IMS
Probes
Connection Layer
External
Subscriber
data
Generic RNC /
eNodeB If
E/// GPEH
adapters
Generic SGSN /
MME Interface
GTP
Probes
CDR Adapters 3G
rnc
OSS
sgsn / MME
sgsn UE
MSC
EMM
ggsn 4G
enb UE
© Ericsson AB 2015

21. MULTI-VENDOR ADAPTATION LAYER
Sales Training Ericsson Expert Analytics
MULTI-VENDOR ADAPTATION LAYER
Continuously increasing libraries / integration points …
Multi Vendor Libraries
OSS/BSS
Multi Vendor Libraries
OSS/BSS
Multi Vendor Libraries
OSS/BSS
OSS/BSS Integrations Topology, IMIETAC, CRM
3PP Probe Libraries (1)
3PP Terminal Libraries (1)
Multi Vendor Libraries *
OSS/BSS Integrations*
Multi Vendor Libraries *
OSS/BSS Integrations *
Generic KPI Adapter (comes with a reference integration model)
Generic Network Adapter (comes with a reference integration model)
Terminal & Survey Data Adapter (comes with a reference integration model)
Reference Data API (comes with a reference integration model)
RAN: Ericsson (continuous support for all relevant nodes for all versions)
Core: Ericsson (continuous support for all relevant nodes for all versions)
Today
Expert Analytics 15.1
Expert Analytics 16
© Ericsson AB 2015

22. Sales Training Ericsson Expert Analytics
Adapters In general
Adapters interface between the data sources and the Correlator
to gather and format the data into the expected structure
pre-integrated adapters for E/// network devices
generic adapters for inter-operability with other vendor’s equipment (SI integrated)
Correlation Layer
Real Time Correlator
Connection Layer
Adapters interface between the data sources and the correlator to gather and format the data into the structure expected by the correlator
EEA comes with pre-integrated adapters for E/// network devices
EEA provides generic interfaces for inter-operability with other vendor’s equipment
SI activity is required for other vendor’s equipment
The pre-integrated adapters do filtering of key IMSIs that should not be seen by EEA (Blacklisting)
Reference data
CRM
Cell info
IMEI TAC
Billing
External
Subscriber
data
Generic RNC /
eNodeB If
E/// GPEH
adapters
E/// CTR
adapters
E/// EBM
adapters
Generic SGSN /
MME Interface
E/// ctum
adapters
Generic KPI
Interface
GTP
Probes
CDR Adapters
IMS
Probes 3G
rnc
OSS
sgsn / MME
sgsn UE
mMsc, mtas, cscf, sbg
IMS
* Not Delivered as part of EEA
** SI work required
ggsn 4G
enb UE
CORE
RAN
VOICE
© Ericsson AB 2015

23. Sales Training Ericsson Expert Analytics
RAN integration
RAN adapters are concerned with items like
Radio Bearer Handlings
Mobility (e.g. handovers)
Radio resource metrics
Cell metrics (e.g. cell load)
Enabling
root cause analysis / most probable cause detection in relation to the RAN performance
e.g. coverage issues, cell load bottlenecks
e.g. suspecting cell outages
analysis of location dependency of end-user experience
analysis of connection drop causes
Ex: Nokia OSSii interface
R-KPIs
Most Probable Cause
© Ericsson AB 2015

24. Sales Training Ericsson Expert Analytics
PS Core integration
Enabling
root cause analysis / most probable cause detection in relation to connectivity issues
e.g. no subscription for service
tracking subscriber mobility
detailed analysis of core network signaling performance
e.g. paging success, RAU succcess
Core NW adapters are concerned with
Mobility Management
Subscription of Service
R-KPIs
Most Probable Cause
Alternative: PS Core Signalling probe
© Ericsson AB 2015

25. CDR (2G/3G Voice) integration
Sales Training Ericsson Expert Analytics
CDR (2G/3G Voice) integration
2G/3G CDR Adapter for
MSC CDRs
files from the Mediation Platform / CV Billing
Enabling
Analysis of the subscribers’ data service experience together with their CS activity, for example:
Originating / Terminating Call Success
SMS Originating / Terminating Success
Call Forwarding
Emergency Call
Call Hold Time
EA MSC ADAPTER
MSC
Analytics pulls the files over SSH/SCP
Mediation collects the data from the MSC
MSC creates the CDRs
Mediation
S-KPIs
Pre-requisites : enable un-answered/ failed CDRs
© Ericsson AB 2015

26. Big data analytics Probes In the network
Sales Training Ericsson Expert Analytics
Big data analytics Probes In the network
Vendor independent
Capturing standard interfaces:
Gn GTP-C
Gn DPI
Pre-requisites : taps or span ports or splitters
EEA GTP Probe
Active tap (e.g. Gigamon) 3G
RNC
SGSN-MME
The EEA probes:
Capture user service activity
Detailed analysis of video and web
TCP performance analysis
Signature-based traffic classification (mostly possible even if the application content is encrypted)
Do Deep Packet Inspection
Analyze 100% of the encapsulated GTP-C and GTP-U traffic
Are Compliant with 2G/3G/4G Core Networks
Analyze each session for every subscriber
The generic KPI interface can be used for data from third party probes (not an apple to apple comparison)
Network taps are required along with the flow balancer (when traffic is greater than a single probe can support)
Gn, S11
Gn-3GDT
LTE
S1-U
eNB
GGSN-S-GW
© Ericsson AB 2015

27. Integration with OSS/BSS SUITE
Sales Training Ericsson Expert Analytics
Integration with OSS/BSS SUITE
3PP Exposure and Analytics tools
Trouble Ticketing
API for CRM GUI
CRM
Self Care
Reference data
CRM Customer info
SLA / SLI
Cell topology (NIM)
billing
ETSI IMEI / TAC db
EEA
SMS Notifications
NPS
Customer Surveys DB
© Ericsson AB 2015