International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 02 Issue: 03 | June-2015 www.irjet.net p-ISSN: 2395-0072
© 2015, IRJET.NET- All Rights Reserved Page 86
BOOSTING THE PRIVACY OF REAL TIME DATA WITH DIFFERENTIAL
PRIVACY
Ms.Neeta Patil
1
, Prof.Pankaj Agarkar
2
1
PG Student, Computer Engineering, Dr.D.Y.Patil School of Engineering,Lohgaon,Pune,India
2
Assistant Professor, Computer Engineering, Dr.D.Y.Patil School of Engineering,Lohgaon,Pune,India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Performing data mining of real time data
has great value. Generally it has importance in real
time applications like traffic congestion applications,
disease surveillance applications. In these types of
applications data values are highly correlated so
providing differential privacy to this data is essential.
Hence while maintaining the confidentiality of data, the
purpose of statistical database is to allow third parties
or users to study the properties of traffic congestion
applications or contents in the databases but individual
users contribution is keep safe from damage. Trusted
party carry the whole database and provide requested
information to third party in the form of count. User or
third party make query to the trusted party for
accessing some data from trusted server. In the form of
aggregate function trusted server or party sends the
requested data count to user. This aggregate function is
formed as answer to the query plus some random noise.
So to provide strong privacy guarantee all appreciable
data is examine methodically.
Key Words: Differential privacy, Geometric
mechanism, Exponential mechanism, Laplace
perturbation, filtering, sampling
1. INTRODUCTION
While considering some applications sharing of real time
data has great value. Because many times such data is
private data and needs to maintain privacy of this data.
For that purpose real time data is mostly shared in the
form of aggregate function which provide count to user.
When real time data is shared in the aggregate function
form security of that data is maintained properly. Thus to
provide strong guarantee about confidentiality of data all
appreciable data is examine methodically. Over the past
few years, privacy of data is preserved by maintaining
some terms. Privacy is provided to databases, the general
principles of science(theory) and other kind of data. This
privacy is provided by means to increase the correctness
of queries from statistical databases while decreasing
opportunity of identifying its records (differential
privacy). In many data mining applications, real time data
is provided in the form of aggregate function to provide
the security to private data. Applications where such
privacy can be provided are Disease Surveillance,
monitoring of road traffic, medical records, voter
registration information, and email usage. The goal of this
differential privacy is analysis of confidential data by
preserving the privacy of data. Consider following
examples:
Disease Surveillance: Providers of health care collects data
from individual visitors. After that all this collected data is
shared with third party, who maintains record of this data
properly and securely. This third party has all the
collected data. So third party may share this data to other
parties. Other parties are may be trusted or not.
Monitoring of traffic: Providers of GPS service collect data
from individual users. This data is related to location,
speed of vehicle and mobility. Again here providers of GPS
service share data with third party which is trusted or not.
In recent years lot of research is done on privacy
preserving of whatever data get published. Different terms
or concepts are there to preserve the privacy of statistical
databases. Among those differential privacy has lots of
importance. As different applications like disease
surveillance, bank database or other maintain lots of data
in their databases. This data needs to collect and then
stored. All stored data is maintained properly so that
guarantee about privacy is provided. So preserving
privacy of secure data is big question in front of different
organizations and government. Attacker can do any type
of attack to access the data from database. While making
access of data available to user, strong privacy guarantee
needs to provide. Trusted party done not have attackers
background knowledge.
2. LITERATURE SURVEY
2.1 Calibrating Noise
Trusted servers carry the database or private information
of individual users. User make query to the trusted server
for accessing some data. Trusted server sends the
requested data to user in the form of aggregate function..
This aggregate function is formed as answer to the query
plus some random noise. So to provide strong privacy
guarantee all appreciable data is examine methodically.
There are two basic categories for privacy i.e. input and
output perturbation: In first approach, data are changed
randomly and answer to the user created query is
calculated from changed data. In second approach i.e.
output perturbation, using real data correct answer to the
query is calculated and noisy data is provided.