Professional Documents
Culture Documents
Page 1
nsn.com
FutureWorks
CONTENTS
Executive summary
Use cases
10
11
12
13
14
Conclusion
15
Page 2
nsn.com
FutureWorks
Executive summary
Ask a typical mobile broadband operator what their key optimization
problem is from a business perspective and they may well answer:
How do I provide the best experience for my subscribers, while at the
same time utilizing my limited network resources well and maximizing
my revenues?
In reality, todays networks are mostly a case of one size fits all.
Exponential traffic growth is not matched by revenue growth. High
average revenue per user (ARPU) subscribers churn due to poor
network experience. All traffic, whether high or low value, is treated
equally. Unfortunately, such mismatches are likely to worsen as many
more applications, devices and sensors become available.
By Personalizing the Network Experience many operators can solve
their major optimization problem - aligning resources, revenues
and experience. To do this, a telco-scale system is vital to enable all
signaling data across the network to be captured in real-time, and
then correlated with customer and business data to allow immediate
action. Such an approach would enable the network to accurately
predict what will happen next and then act on these predictions, even
down to the level of an individual connection. In practical terms, the
system would, among other things, be able to:
Automate and correct in real time any faults that are degrading
services for subscribers
Ensure a consistent video experience by using predictions for a high
value user traveling a specific route
Dynamically adapt the network to sudden changes in use
Dynamically alter policy and radio settings according to user value
and network load
Analyzing the colossal amount of raw network data and turning it
into immediate business value can only be achieved with big data
technologies applied to telco environments and enhanced by machine
learning and machine reasoning . Nokia is running several innovation
projects to realize the personalized network experience. For example,
a FutureWorks innovation project, overcomes the most critical
challenge of telco big data - scaling up processing capacity in line with
the growing volume of data. The project uses parallel computing and
a virtualized architecture to handle massive volumes of telco big data
in real time, with no theoretical capacity limit. It has already processed
one million messages per second, reflecting the peak hour data
volume of a real 3G network with 10 million subscribers.
Page 3
nsn.com
FutureWorks
Page 4
nsn.com
FutureWorks
Customer
Experience
Differentiate
the customers
Differentiate
the offering
Personalize
network
experience
Network
resources
Revenue
Page 5
nsn.com
FutureWorks
Personalize
network experience
Differentiate
the customers
Customer
Experience
Personalize
network
experience
Network
resources
Guidelines
Business
Differentiate
the offering
Revenue
Management
Experience
Teach networks to
be self-aware
Network
Optimization
Self-management
Networks
Devices
Metrics
Applications
Insight
Ins ight
Metrics
Metrics
Users
Revenue
Fig. 2: Optimize network, experience and business simultaneously - leveraging data and machine
intelligence technologies
Page 6
nsn.com
FutureWorks
User perspective
8:15 am
Predicted cell
congestion
8:20
95%
8:25
85%
8:30
70%
The operator collects data on the network and user behavior for
a period of time. The network realizes that the subscriber uses
music streaming along the same route at about the same time
each day. The operator can even predict when cell congestion will
happen, based on long term measurements from network prediction
algorithms.
One particular morning, foul weather is forecast which will lead to
heavy congestion along the route. From immediate monitoring,
the network can see the high value user, subscribed to a streaming
service, is traveling through the area and streaming music.
The network sets a higher session priority for the user and reduces
the scheduling frequency of subscribers using interactive, but delaytolerant, services like Facebook and Twitter.
The user is able to pass through the congested area with break-free
music playing at the same quality as always. Meanwhile, other users
notice no change in their service since only delay-tolerant services for
budget users were scheduled less frequently.
Long term data gathering allows networks to accurately predict
network performance down to the cell level which, when combined
with information on user behavior and subscribed service levels,
allows the operator to sell and deliver higher service packages
customized to users needs. The high value user gets whats been
paid for - a consistent streaming performance - while the operator
is able to efficiently schedule other users in a way that doesnt affect
their quality of service.
Page 7
nsn.com
FutureWorks
User perspective
Session drops
Corrected
Quality monitoring of
all network sessions
Page 8
nsn.com
FutureWorks
User perspective
High Value
Budget
Data from the network shows that a users video session, which
started well, is being degraded by video buffer starvation. The
operator needs to know whether this is because the user is in a
bad coverage area, whether there are YouTube server issues due to
sudden demand, or whether there is cell congestion or a network
fault.
Data from several sources is analyzed - around the user at the cell
level, compared to other users watching the same video, and YouTube
network throughput, to determine the severity of the problem.
Actions need to be taken quickly since YouTube videos are relatively
short, on average around 3-4 minutes, stalling of video being far more
disturbing for a user than lower quality.
Network analysis shows that this is a congestion issue in the cell and
since the operator has adopted a business rule to always have good
quality for video as a differentiator, the network enforces hard limits
for minimum acceptable video data rates in that cell. In addition,
delay-tolerant services are scheduled less often to give preference to
video over best effort services. This ensures all users can watch the
viral video without stalling.
The ability to make these kinds of experience optimizations comes
from insights into network performance, application performance and
usage, guided by business logic.
Page 9
nsn.com
FutureWorks
User perspective
Aggregated traffic 3G/LTE cell
Morning
Afternoon
Page 10
nsn.com
FutureWorks
Predictive
performance
based user habits
Customer Care
Value based QoE
Application
performance
> 15 minutes
minutes
Immediate
Dynamic adaptation
seconds
NOW
Instant
NW reflex
Page 11
nsn.com
FutureWorks
Wireless network
Verified for
accuracy
1,000,000
Every second in
real-time
Music most
popular in the
mornings
Massive
network data
WhatsApp*
Unique
footprint
578,000
QoE equals
user policy
Twitter*
Video quality
links to
coverage
143,000
Visa*
18,500
Data
Variety
Volume
Velocity
Veracity
Fig. 8: In telco data is not big its colossal and most of it is real-time
Page 12
nsn.com
FutureWorks
Page 13
nsn.com
FutureWorks
Proven linear
scalability
Parallel computing and
virtualized architecture
Page 14
nsn.com
FutureWorks
Conclusion
Nokia believes that with Personalizing the Network Experience,
operators will be able to align data traffic growth, revenues and
customer experience by changing from a conventional network
model to actions at the speed of the user for profitable delivery.
Personalizing the Network Experience is the fundamental approach to
increase operator profitability:
1. Exponential growth of data traffic can better be monetized by
differentiation of traffic, users and offerings.
2. Churn of high ARPU customers can be lowered as network and
service experience accounts for 30 to 40% of the loyalty of
mobile users.
3. Different traffic streams can be better handled according to their
individual requirements. This will become ever more important
as the variety of different traffic patterns increases with new
applications such as machine-to-machine and HD video services.
Personalizing the Network Experience can be summarized as the nexus
between network automation, CEM and predictive analytics.
Nokia is putting all these building blocks in place to create a
personalization solution with cognitive networks to solve operator
challenges and eliminate the one size fits all network model.
The data in the network is immense, rich and can unlock business
value if processed for dynamic adaptation to user needs in less than
15 minutes or in timescales longer than 15 minutes for long term
predictions.
Cognitive networks can be made a reality, as Nokia has solved one
of the biggest hurdles for exposing the data. This technique is being
used by more than ten innovation projects which Nokia is working on
with operators. With its Technology Vision 2020, Nokia will continue
to develop the technology enablers for Personalizing the Network
Experience, which is good for delighting the user but even more
essential for the future of mobile operators.
Page 15
nsn.com
FutureWorks
Nokia is a registered trademark of Nokia Corporation. Other product and company names mentioned herein may be trademarks
or trade names of their respective owners.
Nokia
P.O. Box 1
FI-02022
Finland
Visiting address:
Karaportti 3, ESPOO, Finland
Switchboard +358 71 400 4000
Product code C401-00985-WP-201405-1-EN
Nokia 2014.
nsn.com