Вы находитесь на странице: 1из 29

Big Data Analytics in Mobile Cellular

Networks

Presented By
Rakesh Bhramar

INTRODUCTION
With recent advances of wireless technologies
and ever-increasing mobile applications, mobile
cellular networks have become both generators
and carriers of massive data.
When geo-locating mobile devices, recording
phone calls, and capturing mobile applications
activities, an enormous amount of data is
generated and carried in mobile cellular
networks.

What is big data


analytics???
Big data analytics is the process of
collecting, organizing and analyzing
large sets of data (calledbig data)
to discover patterns and other useful
information. Big data analytics can
help
organizations
to
better
understand
the
information
contained within the data and will
also help identify the data that is
most important to the business and

Drawbacks of Traditional Data


Analytics
traditional data analytics deals with structured
data.
the implementation of data analysis is
traditionally confined within a department, or a
business unit. analytical conclusions come from
very limited, local angles, rather than global
perspectives.
the analytics mainly aims at transaction data,
and pays less attention to the operational data,
due to its incapability to make real-time
decisions.

Why it is required?
data constantly accumulated in the
database
the technologies of big data
analyticsrapidly developed
to improve the performance of
mobile cellular networks and
maximize the revenue of operators

How it is going to help?


As the technology that helps an
organization to break down data silos
and analyze data improves, business
can be transformed in all sorts of
ways.

Big Data Analytics for Mobile


Cellular Networks
In mobile cellular networks
Traditional data analytics:basically
centralized, such as from charging and billing
systems, operation systems, etc.
the huge amount of data is scattered across
the organization, like the device data, cell
site data, network data, back office data, etc.
Higher dimensionality of data implies better
inference, the convergence rate of the
empirical Eigen value distribution to its limit.

Big data analytics supports an global point


of view

Semi-structured, unstructured data


cannot be processed until big data
analytics comes to the scene.

Foundation of big data analytics is


based on three ingredients:
(1)high-dimensional statistics,
(2)matrix analysis, and
(3)convex optimization.

Big Data With Large Random Matrices


The essence of big data analytics is to exploit the highdimensionality of the spatial-temporal datasets.
Denote the random vector byxof dimension n,
wherenis large in value.
For example, for 100 antennas with each orthogonal
frequency-division multiplexing (OFDM) symbol of 128
modulated tones,

These ndata samples are corresponding to some


spatial points (modulation tones are functions of
spatial points).

Big Data With Large Random Matrices

High-dimensional statistics suggests


that a great numberNsamples
ofndimensions are jointly taken into
account, to extract corre
N independent realizations of the
random vector x, i.e.,
.
A random (data) matrixXof size is
formed using
Where

Tstands

for

the

matrix

Big Data With Large Random Matrices

From the viewpoint of analysis, the data


matrix is the basic departure point for
any big data analytics.The matrix can
be decomposed into eigenvalues and
eigenvectors using the eigenvalue
decomposition.
Let us say we obtain neigenvalues of
the
sample covariance matrixS
whereHrepresents the transpose and
conjugate
of
a
matrix.
These
neigenvalues are non-negative.

Big Data With Large Random Matrices

Here, a new matrix transform is


defined

whereUof size
is the unitary
Haar matrix, a random matrix.

Comparison the two paradigms.


The neigenvalues of the data
matrix is supported on the nonnegative real-axis.Theneigenvalues
of the transformed matrixYare,
however, supported on the WHOLE
complex plane!

The eigenvalues of the transformed random matrix defined in(3)are distributed within
the single ring, as predicted by the single ring theorem. Outliers are clear from the figure.
Simulation parameters:n=1000

ANARCHITECTURALFRAMEWORK TOSUPPORTBIGDATAANALYTICS INMOBILE


CELLULARNETWORKS

Data Collection
data collection is the process of forming
data matrix X
data matrixX as a large random
matrix.
Big data in mobile cellular networks can
be gathered from either internal or
external sources.
Data collection 1. through data sources
2. through auxiliary tools

Big Data Analysis and Preprocessing

From a statistical analysis point of


view,
extract correlation contained in the data
matrix X
sample covariance matrix
matrix transform, to use the fundamental Singe
Ring Theorem

Three common data preprocessing


techniques are: integration, cleaning
and redundancy elimination

Big Data Analytics Platforms


and Tools
Apache Hadoop is an open-source
software framework for distributed
storage and distributed processing of
large-scale datasets.The power of
clusters enforces Hadoop to store
and process data at an amazing
speed.

Big Data Analytics Applications


The applications of big data analytics
in mobile cellular networks can be
divided into two categories:
internal business supporting
applications and
external innovative business model
development.

CASESTUDIES OFBIGDATAANALYTICS IN
MOBILECELLULARNETWORKS
Big Signaling Data

Big Traffic Data

Big Location Data

Big Radio Waveforms Data

Big Heterogenous Data


One critical task of big data analytics in mobile cellular
networks is the integration of very heterogeneous data:
correlation mining in massive database. Data sources are
rich in types such as data rate, packet drop, mobility, etc.
Different base stations host these data over time. They
need to be aggregated across space and time to obtain big
data analytics.For example, for cyber security there are
many different heterogeneous sources, such as numerous
distributed packet sniffers, system log files, SNMP traps
and queries, user profile databases, system messages, and
operator commands. Essentially, data fusion is a
technique to make overall sense of data from different
sources that commonly have different data structures.

RESEARCHCHALLENGES
privacy may be among the most important
challenges
How to filter out un-useful data is another
significant challenge
automatically generate the right metadata
Data analysis is considerably more
challenging than simply locating, identifying,
understanding, and citing data. For effective
large-scale analysis, all of these have to
happen in anautomatedmanner.

CONCLUSIONS
Big
data
analytics
will
be
an
indispensable part of the mobile cellular
operators consideration of network
operation, business deployment, and
even the design of the next-generation
mobile cellular network architectures.
The connection between big data
analystics and mobile cellular networks
has been systematically explored.

REFERENCES

1. C. Liang, F. R. Yu and X. Zhang


"Information-centric network function virtualization over 5G mobile wireless networks"
IEEE Netw., vol. 29, pp. 68-74,2015
2. C. Liang and F. R. Yu
"Wireless network virtualization: A survey, some research issues and challenges"
IEEE Commun. Surveys Tuts., vol. 17, no. 1, pp. 358-380,2015
3. J. Liu, F. Liu and N. Ansari
"Monitoring and analyzing big traffic data of a large-scale cellular network with Hadoop"
IEEE Netw., vol. 28, no. 4, pp. 32-39,2014
4. S. Bi, R. Zhang, Z. Ding and S. Cui
"Wireless communications in the era of big data"
IEEE Commun. Mag., vol. 53, no. 10, pp. 190-199,2015
5. J. Liu, N. Chang, S. Zhang and Z. Lei
"Recognizing and characterizing dynamics of cellular devices in cellular data network through
massive data analysis"
Int. J. Commun. Syst., vol. 28, no. 12, pp. 1884-1897,2015
6. R. C. Qiu, Z. Hu, H. Li and M. C. Wicks
Cognitive Radio Communication and Networking: Principles and Practice
2012,Wiley
7. A. Guionnet, M. Krishnapur and O. Zeitouni
"The single ring theorem
Ann. Math., vol. 174, no. 2, pp. 1189-1217,2011

8.C. Zhang and R. C. Qiu


"Massive MIMO as a big data system: Random matrix models and testbed"
IEEE Access, vol. 3, no. 4, pp. 837-851,2015
9. A. M. Khorunzhy, B. A. Khoruzhenko and L. A. Pastur

"Asymptotic properties of large random matrices with independent entries"


J. Math. Phys., vol. 37, no. 10, pp. 5033-5060,1996

THANK YOU