Classification of secure encrypted relationaldata in cloud computing

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org 159
CLASSIFICATION OF SECURE ENCRYPTED RELATIONALDATA IN
CLOUD COMPUTING
S.Narayanan1
, Naushin Ghani.V2
, Sangeetha.P3
, Manimegalai.K4
, Vishali.P5
1
Assistant Professor, B.Tech Information Technology, Valliammai Engineering College,Kattankulathur, Kancheepuram District -
India
2,3,4,5
Final Year Students, Department of Information Technology, Valliammai Engineering College, Kattankulathur,
Kancheepuram District - India
Abstract
Due to the increasing popularity of cloud computing, organisations have the choice to outsource their large encrypted data
content along as well as data mining operations to cloud the environment. Outsourcing data to such a third party cloud
environment can compromise the data security as cloud operations and data mining tasks cannot carry out computations without
decrypting the data. Hence, already present privacy-preserving data mining techniques are not efficient to address the security
and confidentiality problems. In the base paper, a k-NN classification algorithm over secure data under a semi-honest model was
developed using a Paillier cryptosystem for public key encryption. The usage of public key cryptosystems has security issues
during data transfer in the cloud. In this proposed work, we focus on solving the k-NN problem over secure encrypted data by
proposing a privacy preserving k-nearest neighbour classification on encrypted information in the cloud using private key for
encryption and decryption based on the symmetric AES cryptographic algorithm under the secure multiparty computations for
creating a complete homomorphic encryption (CHE) scheme which results in the reduction of space requirement and processing
time. Also, we aim to apply the same PPk-NN classification over encrypted images. The proposed protocol hides the input query
and data access patterns of the users and also preserves the confidentiality of text and image data.Finally, we present a practical
analysis of the efficiency and security performance of our proposed protocol for application in a Life insurance firm where the
clients are classified according to their risk-level.
Keywords: Data Mining, PPk-NN, Semi-Honest Model , Individual Key, Symmetric Homomorphic Encryption, AES
Algorithm,CHE, Less Space and Time.
---------------------------------------------------------------------***---------------------------------------------------------------------
1. INTRODUCTION
In today's computerized world, security is a major concern
in all sectors of the workforce. Many organizations, due to
extensive need for storage and resources, choose to
outsource their databases to the third party server. But due to
concerns about confidentiality and security issues, they have
to think twice before outsourcing their data to the cloud. [2]
Generally, data is encrypted before sending to the cloud. But
cloud operations such as data computations and data mining
tasks cannot be performed over such encrypted data[3][6]
.
Hence, data has to be decrypted by the third party cloud
before such data mining or cloud operations are performed.
Also, during processing queries, the cloud can also obtain
confidential and private information about the data by
analysing the data access pattern while user query and
information are encrypted[10]
. Therefore, from the
observations ,a privacy preserving query processing needs to
assure aconﬁdential setup for the encrypted data in cloud
and encrypt the user’s query while keeping data access
patterns hidden. [4][7]
In this paper, we mainly aim to avoid such pre-decryption of
data before performing the data mining classification tasks.
We propose a novel system for the semi-honest model[1]
which makes use of SMIN (Secure Minimum), SF (Secure
Frequency), SMC(Secure Multiparty Computation),
SMINn(Secure Minimum out of n numbers) protocols[1][3]
to
avoid leakage of intermediate data computation results. We
present a PPkNN(Privacy-Preserving k Nearest Neighbor)
classification method under the homomorphic encryption
scheme which aims to preserve the encrypted format of data
during mining operations by prohibiting decryption
activities by a third party evaluator.
Existing systems under semi-honest model only provide a
Somewhat Homomorphic Encryption(SHE) due to the
public key encryption system. In this paper, we use
symmetric AES algorithm[5]
for the private key generation
that creates a Fully Homomorphic Encryption (FHE)[4][7]
,
hence providing additional security. Also, not much work
has been done on preserving the encryption of images as
much as textual data records[5]
. Hence, we present a reliable
means of preserving encryption of images during data
mining classification.
2. RELATED WORK & BACKGROUND
Here, we ﬁrst present a light study of the existing secure k-
nearest neighbour methods. Then, we present the security
modelused in this paper along with the Complete

_______________________________________________________________________________________
homomorphic encryption scheme with symmetric key
cryptosystem and secret image sharing techniques.
Fully Homomorphic Encryption scheme
The view of privacy homomorphism was first introduced by
Rivest, Adleman and Dertouzos in 1978 [11].Generally,
many encryption techniques possess either a multiplicative
property or additive homomorphic property used in certain
areas. A fully homomorphic encryption scheme which can
compute any calculation of arbitary nature over encrypted
data was presented only in Gentry’s paper in 2009[12].
Gentry starts from a Partial Homomorphic Encryption(PHE)
that handles only a finite number of computations
performed homomorphically on cipher texts.
Based on Gentry's work, a fully homomorphic scheme [13]
over integers appeared at 2010. The primary attraction of
the work is that it has a conceptual ease; as inall operations
are performed over the integer values instead of lattice
values. But, the public-key size was very large. Coron et al
[15] presented a way to minimize size of the public key to
Õ( 7  ), by saving only a smaller range of the public key.
The public key was lowered to Õ( 3.5  ) as preset in [16]
by using a decryption algorithm based on probability .
Paillier Cryptosystem:
Pascal Paillier proposed a new cryptosystem[14] in the year
1999. This method is based on composite residuosity
classes.The evaluation of these classesis is said to be too
complex. It is a cryptosystems using asymmetric algorithm
based on probability and makes use of additive
homomorphic properties, where the product of two
ciphertexts will be decrypted based on the total of their
plaintexts.
Image sharing Techniques
The latest techniques in instant communication require new
methods of protecting, storing and transmitting important
data and involves security strategies such as encryption
watermarks ,digital signatures.Shamir[17] presented a novel
method ofsecret sharing in 1979.The secret image sharing
[5]investigates the integrity and security of shared secret
images and proposes new methods for polynomial secret
image sharing. Normal security systems for sharing secret
images assume that all pixels are independent. However,
this assumption may not be always correct. Advancements
in secret image sharing means increasingly sophisticated
ways of producing shares. Enhancing the security of
traditional secret image sharing has 2 parts –1)Encoding and
2)decoding. All computations are to be done in the Galois
Field GF(N) which is a finite field with values in the range
of [0, N] and pixel depth is 1.The proposed method
attempts to prevent dealers from knowing the position of
any pixel. A user who doesn’t know the pixel positions
cannot formulate a relationship basis to identify the original
pixel value when the amount of shares is less than the
critical value.Firstly, the entire image is encoded using an
AES encryption system along with a key. Secondly, the
secret image sharing method is adjusted from prime value
251 to GF(256) to attain a lossless recovery. Thirdly, the
shared key is assigned to some temporary keys by the
Shamir secret sharing method [18]. Finally, the temporary
keys and images are combined to produce a complete set of
shares. This algorithm is effectively more secure compared
to traditional secret image sharing.
3. PROBLEM STATEMENT
In a two-party communication,one party A will encrypt the
database D using a secret key Eskcontaining n records and m
attributes in bulk and outsource the encrypted database D’
to a third party cloud for storage and evaluation. The second
party B will want to extract the data by sending the query qu
for retrieving a specific record containing text and images
from the database D’(r, i).But during the retrieval process,the
third party cloud decrypts the data in order to perform the
k-NN computation based on the attribute classifiers
m+1.Hence,we aim to preserve the encrypted format of the
database by applying Privacy Preserving k-NN that
prevents such decryption. Also, encrypted images can be
decrypted by mapping the individual pixel positions of the
images.So, preserving the privacy of encrypted images is
done using fault tolerance key.
Therefore,
Esk(D(r,i)) D’
qu D’
4. SEMI-HONEST MODEL
Assume a situation where the parties involved in
communication sends its inputs to a trusted third party that
calculates the functional component ,a arbitrarily random
process that designates and maps n inputs to n outputs. This
trusted third party gives both the sender and receiver its
output , so we would prefer to perform computations
without this trusted party.So a semi-honest party is used
where all the protocols are followed by the trusted party
only with the exception that it keeps all its intermediate
computational results to itself. The semi-honest party while
tossing a fair coin,will only toss a fair coin and perform or
reveal anything else.A separate protocol securely calculates
what a semi-honest party can obtain after participating in the
computations, and also from its inputs and outputs.
Use an encoding of 0’s and 1’s Alice chooses a random
encoding of a random bit b and sends Bob the one-way
function (or more exactly bit commitment) of the bit.Bob
sends a random bit c to Alice , Alice reveals the
commitment to b The common random bit is b+c.
We considered in classes, the functionality in which A
outputs a random bit and B outputs nothing andcannot learn
A’s output bit. We showed a protocol that is clearly
insecure: A chooses a random bit and outputs it to B (B
ignores the bit). This protocol is insecure since B learns A’s
output. We could have proved this protocol to be secure if
we used a security definition that compares the view of a
corrupt B in the real execution to the output of a simulator
that only gets B’s input and output: The simulator should
generate a transcript that contains a single random bit sent
from A to B. This demonstrates that a security definition

_______________________________________________________________________________________
that does not take into account the joint distribution of both
parties is insufficient. Note that the protocol cannot be
proved to be secure according to the security definition that
looks compares the joint definition: In the real execution the
bit sent to B is identical to the bit output by A. In the
simulation the simulator does not have access to A's output
and, therefore, the value that it generates for B's transcript
would be independent of A's output.
5. PROPOSED SYSTEM
In this system, we propose a k-NN classification over
semantically secure encrypted data in the cloud
environment. A privacy preserving k-NN is used to classify
the data based on the user’s query(which is also preserved in
encrypted format) and also to provide security to the
encrypted data. This PPkNN classification solves the
DMED(data mining over encrypted data) problems such as
user data confidentiality, privacy of the user’s query and
encrypting the access patterns. encryption .We have based
the security premises on a Semi-honest model where the
revelation of the interparty computations is prevented.The
secure privacy-preserving kNN is built using various sub-
protocols, namely secure multiplication (SM), secure
minimum out of numbers (SMINn), secure Bit
Decomposition(SBD), Secure Shortest Euclidean
distance(SSED) for performing secure PPkNN, along with
formal security analysis and proof under the semi-honest
model[1].
We also propose a new concept of image encryption in the
cloud environment by using the safe secret image sharing
with fault tolerance key. In this process, an advanced
method of secret image sharing is performed compared to
the traditional method where independent neighboring
image's pixel values are mapped secretly where a co-
efficient is required in sharing the images to preserve the
pixel values. To achieve this , a fault tolerant key along with
AES cryptographic technique is used for implementing safe
image sharing under the semi-honest homomorphic model.
The database encryption and user query encryption is based
on a (FHE)Complete/fully homomorphic encryption scheme
that is achieved using a private key AES(Advanced
Encryption Scheme), a block cipher algorithm is used for
the symmetric key. This scheme is used for evaluation of
expressions with encrypted operators, handling memory
operation (such as assignment and lookup operations) and
execution of loop done under a secure environment.
In our setting , let us consider two users Bob and Alice that
make use of two semi-honest cloud service providers C1 and
C2.Bob outsources his encrypted database D' to one cloud
server C1 and the secret key to C2.For added security,after
outsourcing Bob cannot involve in any future
computations.Now,Alice would want to retrieve a record
from the database in a secure manner. Hence,we implement
our goal to classify users’ records in D’ using PPkNN
protocol for secure retrieval.
6. PRIVACY-PRESERVING PRIMITIVES
We have presented a privacy reserving k-NN that is based
on semi-honest model and fully homomorphic
encryption(FHE). The semi-honest model is applied over
two parties (P1 and P2)communication scheme. Here,We
apply a symmetric private key AES cryptosystem where a
single secret key S is used for the purpose of encryption and
decryption and only when the secret key is shared to the
decryption party P2, the data can be decrypted.
6.1.1 AES Encryption
AES (Advanced Encryption Standard) is a symmetric
algorithm which is used for encryption. It was designed to
be efficient for both hardware and software.It has a various
block lengths such as 128 bits, 192 bits and 256 bits.
KEY
LENGTH
(Nk words)
BLOCK
SIZE (Nb
words)
ROUNDS(Nr)
AES 128 4 4 10
AES 196 6 4 12
AES 256 8 4 14
Steps in AES algorithm:
 Convert the plain text and given cipher key into the
state table using the ASCII table and the state table has
4x4 square matrix of bytes. The key that is provided as
input expanded into the array of 32-bit words (a0
a1………an) this key would be used as a round key for each
round.
 Perform the XOR operation for the state table and
cipher key and you will get another state array of the 4x4
matrix.
 Key expansion:

_______________________________________________________________________________________
 It takes as input as four words(16 bytes) key and
produces a linear array of 44 words(176 bytes) and these
will be used for the 10 rounds of the cipher.
 The key is copied to first four words of the expanded
key. The remainder will be filled in four words at a
time.Each added word will depend on the immediately
preceding word.In these cases, a simple XOR function is
used.
 After that, we have to take the RCON table first column
to XOR with the previous result.
To understand the PPkNN classification better we have
given a sample dataset containing life insurance details of
clients with their medical history into to identify which risk
level they belong to certification of an insurance policy
scheme.
 Like this, the operation gets continues till the end of the
operation.
 Every 4x4 matrix will be taken for the next round key
for the encryption process.
 After the expansion of the cipher key for the first round
there will be 4 rounds:
They are
 SUBBYTES
 SHIFTROWS
 MIX COLUMNS
 ADD ROUNDKEY
 Substitute Bytes: Uses an S-Box to perform a byte-by-
byte substitution of this block.
 Shiftrows: A simple permutation i.e we have to shift the
2nd
row by one left byte and vice versa for the 3rd
and 4th
byte
 Mixcolumns: A substitution that makes a use of
arithmetic over GF.
 AddRoundKey: A simple bitwise XOR operation of the
current block with the portion of the end key.
 The above four rounds will be continued for 9 rounds
and the last round is the 10th
round that will not have a
mix column step.
 Finally, the 4x4 square matrix will consider the
encrypted data or cipher data.
Consider the following plain text that has to be encrypted
using the given key.
PLAINTEXT:0123456789abcdeffedcba9876543210
KEY : 0f1571c947d9e8590cb7add6af7f6798
Step 1: The plain text is changed into the 4x4 square matrix
state table and the key will also change to the array of
bytes.There will occur an XOR operation and produces the
cipher state table for the next round.
For example
01=0000 0001
0f=0000 1111
0000 1110=0e
The XOR function will be performed on all the bytes in the
state table and the produced state table will be applied for
further process.
Step 2: Round 1 (Sub Bytes)
In this round, it has four steps to be performed to provide
more security.for that, the next step would be substitute
bytes using the S-BOX. After substituting the bytes from the
s-box the resultant matrix would be:
ab 8b 89 35
05 40 7f f1
18 3f f0 fc
e4 4e 2f c4
01 89 fe 76
23 Ab dc 54
45 Cd ba 32
67 Ef 98 10
0f 47 0c af
15 d9 b7 7f
71 e8 ad 67
c9 59 d6 98
0e Ce f2 d9
36 72 6b 2b
34 25 17 55
ae b6 4e 88

_______________________________________________________________________________________
Step 3: Shifting Rows
In this the above matrix will get the shifting process i.e 1st
row will not be shifted.But the 2nd
row is shifted towards the
left by one byte likewise the 3rd
will be shifted by two bytes
and the 4th
row will be shifted by three bytes.The bytes
which are goes beyond the MSB will occupy the empty
place in the respective row.BeforeShifting,
b9 94 57 75
e4 8e 16 51
47 20 9a 3f
c5 d6 f5 3b
After shifting,
ab 8b 89 35
05 40 7f f1
18 3f f0 Fc
e4 4e 2f c4
Step 3: Mix Columns
The forward mix column transformation will be considered
as Mix columns which operate on each column
individually.Each byte of a column will be mapped into a
new value that is a function of all four bytes in that
column.The matrix multiplication has the following state
After multiplying each column with the abovemultiplication
table there will occur a resultant matrix to be XOR with the
round key which has to be moved for the next round
Like this matrix multiplication will be performed on all the
columns and the result matrix will be:
ab 8b 89 35
40 7f f1 05
f0 Fc 18 3f
c4 e4 4e 2f
6.1.2 Privacy Preserving K-NN
After the encryption has taken place, the encrypted
databaseD’ will be outsourced to cloud.when a cloud user
would wish to classify and download the data, the k-NN
algorithm is used.But for the cloud to perform this
operation,
it has to decrypt the data.We propose a method to avoid
such decryption under a Fully Homomorphic encryption
scheme by using a Privacy Preserving K-NN algorithm.The
secure k-NN takes place in two stages,
Stage I: Secure retrieval of k-NN.
Stage II: Secure calculation of the Majority Classes
6.1.2. Stage I
Alice sends her query to C1.After which, both C1 and C2
involve in securely retrieving the class labels k nearest
neighbor value of the user query q by using certain sub-
protocols such.
Secure squared Euclidian distance metric(SSED)
This protocol[19] does not take the square root of the value
unlike Euclidean Distance metric hence provides faster
computations during privacy preserving classification
where the database records are classified according to the
attributes. For every data object, P1 and P2 calculate the
SSED with a minimum k neighbor value.
The Secure Shortest Euclidean distance (SSED) protocol is
used to determine the relative distance between the nearest
neighbors.
The Euclidean distance is calculated using the formula,
dist((y, z), (m, n)) = √(y - m)² + (z- n)²
wherex,y denotes the user query parameters which are the
criteria for classification.
Secure bit-decomposition (SBD)
P1 has input data and P2 securely computes the encryptions
of the individual bits of the input data. The output computed
is known only to P1.
Step1: generation of Bytecode.
Step2:swapping of the byte code values(rearrangement).
Step3: Cross multiplication of rearranged values.
Step4: encryption using the secret key.
Step5: further computations.
Decryption occurs in a similar manner except the steps take
place in reverse order.
02 03 01 01
01 02 03 01
01 01 02 03
03 01 01 02
02 03 01 01 ab b9
01 02 03 01 * 40 e4
01 01 02 03 f0 47
03 01 01 02 c4 c5

_______________________________________________________________________________________
After the Shortest distance is calculated the Secure Bit-
Decomposition(SBD) protocol is used to provide encryption
for the classified data.
Secure minimum out of n numbers (SMINn)
In SMINn, we consider a party P1 with encrypted vectors m
from d1 to dnalong with their corresponding encrypted data
and P2.During SMINn,any information relating to diis not
revealedto a third party .Only relevant information is
revealed even to P1 and P2.
6.1.2. Stage II
In stage II, C1 and C2 both compute the classifier label by
using majority ranking among the k value of query q which
only Alice knows. This stage computes class labels by using
SMAXn protocol which is used to classify the query record
to one of the w classes based on the query inputs.
Secure multiplication (SM)
Secure Minimum considers P1 with an input Es(e) and Es(f)
and provides output Es(e*f) to P1, where e and f are
unknown to P1 and P2. During this process, any detailsabout
the variables is not revealed to P1 andP2.
Secure minimum(SMIN)
In Secure Minimum, P1 contains sensitive input(y,z)and P2
holds the secret key S. The main aim of this protocol is for
P1 and P2 to calculate the encryption values of the
individual bits of the minimum number of y and z jointly. In
addition, they compute the encrypted Sminvalue between y
and z which will be known only to P1.
The two stages can be explained with an example as
follows,
Client id
Client
name
image
Father
name
Address phone
Heart
disease
smoker Risk class
#$%!@& +_=@#$ ------ *&%#”:@ +=1#@$% *&^%# &^%_+ @#$%_ -=#$$@!
-=/*$% $$%## 5&^ +_##1@# /+*^2%$# (%^##&*# (&$#)#$ &^%!@~#$ ~!#$^&^##
User Query=client id :LICA010,Heart disease=7,Smoker=3.
Y classification has w=2 classes(ATTRIBUTE VALUES).
1.Assume the value of parameter k which is equal to the
number of nearest neighbors in the database.
Let us consider k as 3.
2. Compute the distance between values in the query
instance and all the data records i.e training samples.
Using the query instance values (3,7),we calculate the
shortest square Euclidean distance which is more quick to
evaluate.
X1=heart
disease
X2=smoker SQUARE
DISTANCE TO
QUERY
INSTANCE.
7 7 (7-3)2
+(7-7)2
=16
7 4 (7-3)2
+(4-7)2
=25
3 5 (3-3)2
+(5-7)2
=4
1 3 (1-3)2
+(3-7)2
=20
3.Evaluate the distance and calculate the values of the
nearest neighbors which depends on the k parameter value.
X
1
X
2
SQUARE
DISTANCE
TO QUERY
INSTANCE
RANK
THE MIN
DISTANC
E
INCLUSIO
N IN THE 3-
NEAREST
NEIGHBOR
S?
7 7 16 3 Yes
7 4 25 4 No
3 5 4 1 Yes
1 3 20 2 Yes
4.Obtain the classification Ythat the evaluated nearest
neighbors belong to.
5.Use the average value of the majority category to which
the nearest neighbours belong to as a prediction strategy for
the query instance record.
Here the values that are within the k neighbor value are
classified to sub-standard life insurance whereas the ones
that are outside the neighbor value are credited with a
standard life insurance scheme.

_______________________________________________________________________________________
X
1
X
2
SQUARE
DISTAN
CE TO
QUERY
INSTAN
CE(3,7)
RANK
MIN
DISTAN
CE
IS IT
INCLUDE
D IN 3-
NEAREST
NEIGHBO
R
Y=CATEGOR
Y
OF NEAREST
NEIGHBOR
7 7 16 3 yes substandard
7 4 25 4 no standard
7. SIMULATION RESULTS
We discuss the application developed using the PPkNN
protocol for a life insurance scheme and record the obtained
results. We used the Private symmetric key
cryptosystem(AES) for creating the underlying fully
homomorphic encryption scheme and implemented the
proposed PPkNN protocol for a Life Insurance Policy
scheme with the dataset containing over1000 records and
more than 8 attributes and use of 4 classifiers.
The life insurance scheme was applied such that the client
attributes containing the medical details such as height,
weight, cholesterol,bloodgroup, heart disease and smoking
were analyzed on a mark based scale and the clients were
classified based on their medical details as to which class of
insurance they belong to.
The 4 classes were the risk levels available such as super
preferred, preferred, standard and substandard.
The clients with excellent health were given super preferred
class containing highest insurance benefits, the ones with
above average health were placed in the preferred class,
average health clients were in standard class and the ones
with very poor health were classified into sub-standard class
or were not able to avail insurance at all.
We evaluated and analyzed the efficiency the PPkNN
separately in two stages using J2EE and My SQL 5.0. The
application was developed using Windows 07 having
PENTIUM IV 2.6 GHz processor and 512 MB DD RAM
with 40GB hard disk.
7.1 Cost-performance Analysis of PPkNN
The experiment was conducted in two stages. Stage I
consists of executing PPkNN using SMINn where k value is
varied from 5 to 25 and key size K=32 bytes. The cost of the
stage I is directly proportional to the value of k .The cost
increases 5%-10% for the increase in k value.
In stage II, the computation cost to generate the final class
label w varies from 0.789sec to 1.89 sec when k value is
varied between 5 and 25.The low computation time for stage
II is due to the use of SMAXn.
The computation cost of stage I is relatively higher
compared to stage II in PPkNN, where stage I takes up
almost 99% of the computation time.
7.2 Improvement of PPkNN Performance
Efficiency
In order to improve the efficiency of our PPkNN protocol,
an offline phase is allocated for pre-computed encryptions
of random numbers denoted as PPkNNo.It is technically seen
that PPkNNo is 33% faster that PkNN without the offline
computation phase.
As operations on each record take place individually,
another method to increase the efficiency is to consider
parallelizing the stage I computations(PPkNNp). Computing
the stage I operations for each record parallel reduces
computation time effectively.
In case, the user operates a resource constraint device such
as a cell phone.Then the cost ofencrypting the user query
takes up to 17 milliseconds which is found to be very ideal.
8. CONCLUSIONS AND FUTURE WORK
Organizations can confidently outsourcetheir data over
tocloud without fearof security threats such as illegal access
or information theft. The existing techniques for secure data
storage promise safety and integrity but are not applicable
for data that is outsourced to third party environments where
data has to remain secure and encrypted on an external
server. This paper showed a new method to perform PPk-
NN classification protocol over semantically encrypted data
that is present in the cloud environment. Our protocol safe
guards the input query confidentiality and also covers the
patterns followed during data access.Thereby, eliminating
the remnant crumbsand traces left behind during
dataretrievaland cloud operations. We also evaluated the
performance of our protocol using a real time application for
both data ad images using secret image sharing the
technique. Since the efficiency of the proposed system
mainly is based on theSMINnprotocol, we strategize to
further analyze different and more effective answers to the
SMINn problem in detail in further studies and work along
with ways for improving encrypted image sharing and
classification techniques by efficiently managing
computation cost during pixel evaluation.
REFERENCES
[1] k-Nearest Neighbor Classification over Semantically
Secure Encrypted Relational Data Bharath K. Samanthula,
Member, IEEE, Yousef Elmehdwi , and Wei Jiang,
Member, IEEE.
[2]Homomorphic Symmetric key Encryption, Network of
the future,2013, Fourth National Conference, Pohang.
[3] Secure k-Nearest Neighbor Query over Encrypted Data
in Outsourced Environments, Yousef Elmehdwi, Bharath K.
Samanthula and Wei Jiang July 19, 2013, Technical Report
Department of Computer Science, Missouri S&T

_______________________________________________________________________________________
[4] Fully Homomorphic Encryption Scheme with
Symmetric Keys ,Itisharma, university college of
engineering
[5] Safe Secret Image Sharing with Fault Tolerance Key
IJCSNS International Journal of Computer Science and
Network Security, VOL.11 No.9, September 2011 , Wen-
Pinn Fang Yuanpei University, Taiwan, R.O.C.
[6] Practical Techniques for Searches on Encrypted Data,
Dawn Xiaodong Song David Wagner Adrian Perrig,
Dawnsong, daw, perrig, University of California, Berkeley
[7] Fully Homomorphic Encryption scheme with Symmetric
Keys with Application to Private Data Processing in Clouds,
C P Gupta, and Iti Sharma , Department of Computer
Sciences and Engineering.
[8] S. De Capitani di Vimercati, S. Foresti, and P. Samarati.
Managing and accessing data in the cloud: Privacy risks and
approaches. In 2012 7th International Conference on Risk
and Security of Intersnet and Systems (CRiSIS), pages 1–9.
IEEE, 2012.
[9] Finding Minimum Weight Connected Dominating Set in
Stochastic Graph basedOn Learning Automata, Peter
Mell,Timothy Grance,2008.
[10] A Private Database Search with Sub linear Query Time,
Keith B. Frikken and Boyang Li, 2012
[11] R. Rivest, L. Adleman, and M. Dertouzos. On data
banks and privacy homomorphisms: In Foundations of
Secure Computation, pages 169-180, 1978.
[12] C. Gentry. Fully homomorphic encryption using ideal
lattices. In Proc. of STOC, pages 169-178,
[13] M. van Dijk, C. Gentry, S. Halevi, and V.
Vaikuntanathan.Fully homomorphic encryption over the
integers. In Proc. of Eurocrypt, volume 6110 of LNCS,
pages 24-43. Springer, 2010.
[14] P. Paillier, “Public-key cryptosystems based on
composite degree residuosity classes”, Proc of
EUROCRYPT-99, Springer, pp 223–238, 1999.
[15] J.-S. Coron, A. Mandal, D. Naccache, and M.
Tibouchi.Fully homomorphic encryption over the integers
with shorter public-keys. In Advances in Cryptology - Proc.
CRYPTO 2011, volume 6841 of Lecture Notes in Computer
Science. Springer, 2011
[16] D. Stehle and R. Steinfeld. Faster Fully Homomorphic
Encryption. Cryptology ePrint Archive: Report 2010.
[17] P. Paillier, “Public-key cryptosystems based on
composite degree residuosity classes”, Proc of
EUROCRYPT-99, Springer, pp 223–238, 1999.
[18] A. Shamir, “How to share a secret,” Communication of
the ACM, Vol. 22, no. 11, pp. 612-613, 1979.
[19] Secure Two-Party Computation of Squared Euclidean
Distances in the Presence
ofMaliciousAdversaries,Marc Mouffron, Frederic Rousseau,
Huafei Zhu.

Classification of secure encrypted relationaldata in cloud computing

Recommended

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Classification of secure encrypted relationaldata in cloud computing (20)

More from eSAT Journals (20)

Recently uploaded (20)

Classification of secure encrypted relationaldata in cloud computing