Final Copy IJITS

Vol. 4, No.
1, Jan-June 2015 ISSN: 2277–9825
Refereed Journal
Global Impact Factor: 0.412
International Journal of Information Technology and Systems

(IJITS)
International Journal of Information
Technology & Systems
(IJITS)
Editor-In-Chief
Dr. Satinder Bal Gupta,

Professor, Vaish College of Engineering,
Rohtak, India
Associate Editor
Dr. Yudhveer Singh,
Associate Professor, UIET, MDU, Rohtak, Haryana, India
EDITORIAL BOARD MEMBERS
Dr. Hamid Saremi Dr. Santosh K. Pandey

Vice- Chancellor of Azad Islamic Department of Info. Technology
University of Iran (Quchan branch) The Institute of Chartered Accountants
Quchan-Iran of India , Noida, India
Dr. K.SURESH Dr. Dil Muhammad Akbar Hussain

Assistant Professor, Annamacharya Associate Professor, Aalborg
Institute Technology & Sciences, University, Denmark.
Rajampet, A.P, India
Dinesh Kadam
Dr. Sukumar Senthilkumar, Assistant Professor, Goverment Engg.
Center for Advanced Image and College, Bathinda, India
Information Technology,
Chon Buk National University, South Dr. Rajender Nath
Korea. Associate Professor, Kurukshetra
University, India
Dr.Sumit Mittal
Associate Professor, MMU, Mullana, Prof. (Dr.) Savita Gupta
India UIET, Punjab University, Chandigarh,
India.
Global Technocrats & Intellectual's Association (GTIA)

A Registered Society under societies Act 21 of 1860.
Published by Mr. Anshul Goel for GTIA
Mobile: +91-9812667898: E-mail: gtia@rediffmail.com
International Journal of Information Technology and Systems
(IJITS)
ISSN: 2277–9825
Volume 4 Number 1 Jan-June 2015
CONTENTS
1. Log mining in Query recommendation ……………………………………………….. 1-8

Sowmya Yalamanchili
2. Multipath Routing in Autonomous Mobile Mesh Networks …………………………9-12

Sangheethaa S, Arun Korath
3. Comparative Study of Enhanced Interior Gateway Routing Protocol (EIGRP)

and Open Shortest Path First (OSPF) for Data Transmission through A
Network System for Real Time Application ………………………………………..13-18
Philip O. Odion, David Bagudu
4. Review of SiGe Double Gate N-MOSFET ………………………………………… 19-24

Akshat Singh, Dhiraj Kapoor, Rajiv Sharma
5. Cross Site Scripting (XSS) Advanced Protection Concept ………………………… 25-28

Ajit S. Ghodke
6. A Review on Cloud Based Patient Monitoring …………………………………….. 29-34

Nabeena Ameen
7. Palm Recognition Using LBP and SVM …………………………………………… 35-41

Preeti Sehgal
International Journal of Information Technology & Systems, Vol. 4; No. 1: ISSN: 2277-9825 (Jan-June 2015
Log mining in Query recommendation

Sowmya Yalamanchili
IBM, Delhi.
Email: sowmya320@gmail.com
Abstract: In last decade lots of research was done on log mining. In fact log mining became an interesting area
of research. One of other interesting field is query recommendation. This paper analyzes log mining on query
recommendation. Log mining is a part of web mining and web mining in turn is a part of data mining which a
part is itself of Knowledge Discovery in Databases (KDD). Query recommendation is a method to improve
search results in web. In this paper it has been search that how log mining will be efficient in field of query
recommendation.
Keywords: Log mining, KDD, Web mining, Query recommendation
Accepted On: 28.01.2015
1. Introduction insufficient or imprecise to retrieve those. It

serves several purposes: correcting possible
Knowledge Discovery in Databases creates the spelling mistakes, guiding users through their
context for developing the tools needed to information seeking tasks, allowing them to
control the flood of data facing organizations locate information more easily, and helping
that depend on ever-growing databases of them to discover additional concepts related to
business, manufacturing, scientific, and personal what they are looking for. A key technology for
information. Data mining is the process of enabling query recommendations is query-log
discovering new patterns from large data sets. mining, which is used to leverage information
Data mining is the process of discovering new about how people use search engines, and how
patterns from large data sets. Web mining is the they rephrase their queries when they are
application of data mining techniques to looking for information.
discover patterns from the Web. Discovering With the increase of size [6] and popularity of
useful information from the World-Wide Web the World Wide Web, many users find it's
and its usage patterns. Web data is Web content difficult to obtain the desired information, even
i.e. text, image, records, and etc .Web structure though they use most efficient search engines
contains hyperlinks, tags, etc. Web usage uses e.g. Google, yahoo. In spite of the recent
http logs, app server logs, etc. Current software Advances in the Web search engine
application often some auxiliary text files known technologies; there are still many situations in
as log files. Such files are used during various which the user is presented with non relevant
stages of software development, mainly for search results. One of the major reasons for this
debugging and profiling purposes. Use of log difficulty [7, 8] is that Web search engines a lot
files helps testing by making debugging easier. have difficulties in forming a concise and
It allows to follow the logic of the program, at precise representation of the user’s information
high level, without having to run it in debug need. Most Web search engine users are not well
mode. Nowadays, log files are commonly used trained in organizing and formulating their input
also at customer’s installations for the purpose queries, which the search engine relies on to find
of permanent software monitoring and/or tuning. the desired search results. On the other hand,
Log files became a standard part of large users are often not clear about the exact terms
application and are essential in operating that best represent their specific information
systems, computer networks and distributed needs. In the worst case, users are still not clear
systems. of what exactly their specific information need
Query recommendations are a module of is. Log files are often the only way how to
modern search engines [1, 2, 3, 4, 5, 6]. It is a identify and locate an error in software, because
technique [7] that provides better queries to help log file analysis is not affected by any time-
users to get the needed documents when the based issues known as probe effect. Log files are
original query submitted by user may be often very large and can have complex structure.
www.gtia.co.in 1
Although the process of generating log files is the overall performance of web site. After
quite simple and straightforward, log analysis identifying the different web server log data files
could be a tremendous task that requires there is a need to merge the log files shown in
enormous computational resources, long time Fig. 1.
and sophisticated procedures. K. R. Suneetha and Dr. R. Krishnamoorthi [9]
The paper has been organized as follows: have analyzed NASA server log file of size
Section 2 we describes the approaches that are 195MB, various analysis has been carried out to
already in use on the basis of proposed work and identify the user behavior. The errors which
literature work that had been done. Section 3 arise in Web surfing were determined. Grace, L.
explains a novel architecture of proposed K. Joshila; Maheswari, V.; Nagamalai,
optimization system in detail along with Dhinahar[3]determine that Log files contain
algorithms. Section 4 concludes the paper with information about User Name, IP Address, Time
some discussion on future research. Stamp, Access Request, number of Bytes
Transferred, Result Status, URL that Referred
2. Related Work and User Agent. The log files are maintained by
the web servers. By analysing these log files
Web log mining is the process of applying data gives a neat idea about the user. This paper gives
mining technologies to discover usage patterns a detailed discussion about these log files, their
from the Web data. One important source to formats, their creation, access procedures, their
discover such patters is the Web log data that uses, various algorithms used and the additional
contains users Web browsing history. Web parameters that can be used in the log files
Usage Mining addresses the problem of which in turn gives way to an effective mining.
extracting behavioral patterns from one or more It also provides the idea of creating an extended
web access log. The entire process can be log file and learning the user behavior. Query
divided into three major steps. The first step, recommendation research is mainly based on
pre-processing, is the task of accurately previous query log of the search engine, which
identifying pages accessed by web visitors. This contains the history of submitted query and the
is a very difficult task because of page caching user selected URLs. Beeferman and Berger [1]
and accesses by web crawlers. The second step, exploited “click through data” in clustering
pattern discovery, involves applications of data URLs and queries using graph-based iterative
mining algorithms to the pre-processed data to cluster-ing technique. Wen et al. [2] used a
discover patterns. The last step, pattern analysis, similar method to cluster queries according to
involves analysis of patterns discovered to judge user logs. Both of their algorithms are difficult
their interestingness. to deal with in practice due to query log
sparseness. That is to say, only a part of popular
queries have sufficient log information for
mining their common clicked URLs while
distance matrices between most queries from
real query logs are very sparse. As a result,
many queries with semantic similarity might
appear orthogonal in such matrices.
Fonseca et al [4] showed a method to discover
related queries based on association rules. The
query log is viewed as a set of transactions.
However, the fact that similar queries are
Fig. 1. log files submit-ted by different users in most of case,
will also lead to sparseness problem. This is
Web server records all users’ activities of the because the support of a rule increases only if its
web site as web servers Logs. Most log files queries appear in the same query session, and
have text format and each log entry is saved as a thus they must be submitted by the same user.
line of text. There are many types of web logs The primary problem is the difficulty in
such as NCSA format, W3C format and IIS determining which sessions of these queries
format, but they share the same basic belongs to the similar search process. The next
information. These log data can be used in web problem the most interesting related queries
site designing, modifying and also to improve which are submitted by different users can't be
www.gtia.co.in 2
presently discovered. For the reason that the the query similarity between the two queries,
support of a rule increases merely if its queries based on user query keywords and clicked
are in the similar session and they must be URLs. After that, clusters are generated with the
submitted by the similar user. M.Hosseini and help of query clustering tool. This tool is used to
H. Abolhassni have described a method used for cluster user queries using query logs built by
recommending associated queries according to search engines which in result produce query
clustering process over web queries from search clusters. Once query clusters are formed, next
engines query log. Zaiane and Strilets [10] step is to find a set of favored queries from each
presented a method for recommending queries cluster. Favored query are those that occupy a
according to seven aspects of query similarity, major portion of the whole search request in a
three of them are moderated variations of the cluster. Once favored queries from their query
first and second ideas. In addition our method clusters are identified, next step is to optimize
recommends the related queries in the direction the user search by recommending him with most
of the input query although my search for favored query related to his search and returning
different issues similar to the previous the desired relevant pages in the [10] top of the
information from query log file. There is one search result list. In this paper I have added a
more approach to recommend related queries by new step that is that if we remove smallest
means of query expansion. In other research by cluster than the search done by query log
Nikita Taneja they have done a good work on it. become smaller and ultimately time complexity
I just want to add some more concepts in the decreases.
Nikita Taneja’s research. The proposed architecture of optimization
system is shown in Fig. 2 which consists of
3. Proposed Work following functional components.
1. Query Log
The proposed optimization system (Fig. 2) lying 2. Query Similarity
on learning from historical query logs is 3. Query Clustering Tool
proposed to calculate user's information 4. Remove unwanted cluster
requirements in a better way. The proposed sys- 5. Sequential Pattern Generator
tem works as follow. The prime feature of the 6. Rank Updater
system is to per-form query clustering by finding
Fig. 2. Purposed Architecture
www.gtia.co.in 3
3.1 Query log some drawbacks, due to the fact that keywords
Query log has been a popular data source for and meanings do not strictly correspond. The
query recommendation. The view of Web Log same keyword does not always represent the
Mining has been an area of interest as many same information need and different keywords
years. In the context of search engine the typical may refer to the same concept. As a result, the
logs [3] of search engines include the following calculated similarity between two similar queries
entries: (A) User IDs, (B) Query q issued by the may be small, while two unrelated queries may
user, (C) URL u selected by the user (D) remove be considered similar as shown in Fig. 3.
unwanted cluster (E) Rank r of the URL u
clicked for the query q and (F) Time t at which
the query has been submitted for search. A
sample query log is shown in Table 1 to better
understand this format.
In various studies, researchers and search engine
operators have used information from query logs
to study about the search process and to get
better search engines from early studies of the
logs created by users. Our method considers
only queries that appear in the query-log. A
single query may be submitted to the search
engine several times, and every submission of
the query induces a similar query session. A
simple notion of query session which consists of
a query, along with the URLs clicked
Table 1. Example of query log
Fig. 3. Similarity (KW:keyword)
The approach taken by this module is based on

two criteria: one is on the queries keywords, and
the other on clicked URLs. These approaches
are formulated below:
3.2.1 Similarity based on Query Keywords

If two queries contain the similar terms, they
3.2 Query Similarity denote the same information needs. The below
The next step in proposed system is computing mentioned formula is used to measure the
the query similarity. It is an important crisis and similarity between two queries.
has a wide range of applications in Information sim୩ୣ୷୵୭୰ୢ(୶,୷)
Retrieval in query recommendation. Traditional KW(x, y)
approaches make the use of keywords extracted = (1)
from documents. If two documents share some max൫‫)ݔ(ܹܭ‬, ‫)ݕ(ܹܭ‬൯
keywords, then they are thought to be similar to Where kW(x) and kW(y) are the number of
some extent. The more they share common keywords in the queries x and y respectively,
keywords, and the more these common KW(x, y) is the number of common keywords in
keywords are important, the higher their two queries. For example : if I put the query for
similarity is. This similar approach may also searching is data mining then the similarity with
apply to query clustering, when a query may the first query of Table 1 according to formula
also be represented as a set of keywords in the given in eq. 1 can be calculated as :
same way as a document. On the other hand, it is Put query = data mining
well known that clustering using keywords has Let q1 is the first query given in Table 1:
www.gtia.co.in 4
q1= data mining examined against all other queries whether

Sim (put query, q1) = 2/4=0.5 classified or unclassified by using (3). If the
similarity value turns out to be greater than the
3.2.2 Similarity based on Clicked URLs pre-specified threshold value (T), then the
Two queries are considered similar if they lead queries are grouped into the same cluster. The
to the selection of same documents. Document similar process is repeated until all queries get
selections are comparable to user relevance classified to any one of the clusters. The
feedback in the traditional IR environment, algorithm returns overlapped clusters i.e. a
except that document clicks indicate implicit single query may span multiple clusters. Every
relevance and not always valid relevance returned cluster is stored in the query cluster
judgments. The following formula dictates this database along with the related query keywords
similarity function. and the clicked URLs. The clustering algorithm
Simୡ୪୧ୡ୩ ୙ୖ୐(୶,୷) takes O (n2) worst [2] case time to find all the
RD(x, y) query clusters, where n is the total number of
= (2) queries.
max൫‫)ݕ(݀ݎ)ݔ(݀ݎ‬൯
Where rd(x) and rd(y) are the number of referred
3.4 Remove Unwanted Cluster
documents for two queries x and y respectively,
Now remove unwanted cluster that are not in use
RD(x, y) is the number of document clicks in
and we get these from above clustering tools
common.
Now, previous work on query similarity aims to
Algorithm for Clustering Queries
give a single similarity measure without
Algorithm: Query Clustering (Q, α, β, τ,th)
knowing the information that queries are
Given: A set of n queries and corresponding
indefinite and generally have several search
clicked URL’s stored in an array
intents.
Same example we take here
Q [q1, URL1…..URL m] 1<=I <=n
If we put url = http://datamining .com
It will come one 1.
Similarity Threshold τ
Output: A set C= {C1, C2….Ck} of k query
3.2.3 Combined Similarity Measure
clusters
The two criteria [1] have their own advantages.
//Start Algorithm
In using the first criterion, queries of similar
compositions can be grouped together. In using
K=0; // k is the number of
the second criterion, advantage can be taken
clusters
from user's judgments. Both query keywords
and the corresponding clicks URLs can partially
For (each query x in Q)
capture the users' interests when considered
individually. For that reason, it is better to
Set Cluster Id (x) = Null; //Initially No query is
combine them in a particular measure. An easy
clustered
way to perform is to combine both measures
linearly as follows:
For (each x € Q)
Simୡ୭୫ୠ୧୬ୣୱ(୶,୷) = Sim୏ୣ୷୵୭୰ୢ(୶,୷) +
{
Simୡ୪୧ୡ୩୙ୖ୐(୶,୷) (3)
Simୡ୭୫ୠ୧୬ୣୢ = (0.5) + (1) = 1.5 Cluster Id (x) = Ck;
3.3 Query Clustering Tool Ck = {x};

In support of the clustering process, the Query If (Ck <th )
Clustering module uses the algorithm, where Remove Ck;
each run of the algorithm computes k clusters. For each b € Q such that x ≠ y
As query logs are dynamic in nature, query {
clustering [8] algorithm should be incremental in KW(x, y)
nature. sim୩ୣ୷୵୭୰ୢ(୶,୷) =
max൫‫)ݔ(ܹܭ‬, ‫)ݕ(ܹܭ‬൯
The algorithm is based on the easy [8]
RD(x, y)
perspective: firstly, all queries are considered to Simୡ୪୧ୡ୩ ୙ୖ୐(୶,୷) =
be unassigned to any cluster. Each query is max൫‫)ݕ(݀ݎ)ݔ(݀ݎ‬൯
www.gtia.co.in 5
Simୡ୭୫ୠ୧୬ୣୱ(୶,୷) found during the pass over the data. At the end
= Sim୏ୣ୷୵୭୰ୢ(୶,୷) of the pass, the algorithm determines which of
+ Simୡ୪୧ୡ୩୙ୖ୐(୶,୷) the candidate sequences are really frequent.
If () These frequent candidates become the seed for
Set Cluster Id (y) = Ck; the next pass. The algorithm terminates when
Ck = Ck U {y}; there are no frequent sequences at the end of a
Else pass or when there are no candidate sequences
Continue; generated.
} // End For Algorithm for Generalized Sequential Pattern

K=K+1; Pattern list 1: = Frequent atoms;
For k=2;
} //End Outer For Pattern list k! = empty; k = k + 1
Ck = set of candidate
Return Query Cluster Set C; For all sessions in S do
Increment count of X ∈ Ck contained in S
3.4 Sequential Pattern Generator End For
This component takes as input the query clusters Pattern list k = {X ∈ Ck / support(X) >
and finds the sequential patterns in each cluster. threshold}
Sequential pattern generator which discovers End For
frequent [4] subsequences as this component Set of all frequent Patterns: = ∪k Pattern list k
takes as input the query clusters and finds the Generalized Sequential Pattern [14] finds every
sequential patterns in each cluster. Sequential item sets with minimum support, transforms the
pattern generator which discovers frequent [4] database therefore each transaction is replaced
subsequences as patterns in a sequence database by the set of all frequent item sets contained in
is an important data mining problem with broad the transaction and then finds the sequential
applications. patterns as shown in Fig. 4.
It is based on their study of customer purchase
sequences as follows: Consider a database of
sequences, where each sequence is a list of
transactions ordered by transaction-time, and
every transaction is a set of items. The difficulty
is to discover all sequential patterns with a user-
specified minimum support, where the support
of a pattern is the number of data-sequences that
contain the pattern. Generalized Sequential
Patterns is an algorithm that discovers all
sequential patterns. Fig. 4. Pictorial Represtation of Sequential Pattern
Generalized Sequential Pattern algorithm [4]
makes multiple passes over the data. The first Candidate Generation-It defines that
pass determines the support of each item, i.e. the candidate’s sequences are generated before the
number of data-sequences that include the item. pass begins. We want to generate as few
At the end of the first pass, the algorithm knows candidates as possible while maintaining
which items are frequent, that is, have minimum completeness.
support. Each such item yields a 1-element Candidate Pruning- A pruning phase delete
frequent sequence consisting of that item. Each candidate sequences that have a contiguous k-
subsequent pass starts with a seed set, the 1subsequence whose support counts is less than
frequent sequences found in the earlier pass. the minimum support. If there is no max-gap
The seed set is used to generate new potentially constraint, we also delete candidate sequences
frequent sequences, called candidate sequences. that have any subsequence without minimum
Each candidate sequence has one more item than support.
a seed sequence so all the candidate sequences
in a pass will have the same number of items. Algorithm for Modified Generate Sequential
The support for these candidate sequences is Pattern
www.gtia.co.in 6
Algorithm: Modified_GSP (QC, min_sup) The methods to find the weight and improved
Given: A Query Cluster with a set of URLs rank of a page are as follows:
Threshold Value min sups //for pruning the
candidates A. Weight Calculation
Output: Set of frequent Sequential Patterns The weight of the URL determined its popularity
//Start of Algorithm Result = ф // Result Set and order of access related to the user query.
Contains all Sequential Patterns Suppose a sequential pattern
P1 = Set of Frequent 1-Length Page Sequence: < {AB} {C} {E}> belongs to a query cluster
//Pages whose support>=min_ sup matched with the user query. This pattern can be
K=2; obtainable as shown in Fig. 5.
While (P (k-1)! =Null)
{
Generate candidate sets Ck (set of candidate k-
sequences);
For (all page sequences S in the cluster QC)
{
Increment count of all c € Ck if S supports c;
Pk = {c € Ck if S supports c;
K=k+1;
Result = Result U Pk
}//end While
Fig. 5. Graphical Represtation of Sequential Pattern
Return Result;
The page A as well as B lies at level 1, page C is
3.5 Rank Updater
at level 2, while D being at level 3. The weight
Rank Updater module takes its input from the
of a page X is inversely proportional to its
query processor i.e. the matched documents of a
position in the sequential pattern and is
user query and an update is applied to modify
calculated as:
the rank score of the returned pages. The module
operates online at the query time and applies the weight(X) = ln൫len୮ୟ୲୦(ଡ଼) ൯ /level(X)
needed updates on the concerned documents. Where Len pat(X) is the effective depth of the
The updated documents are those which are sequential pattern in which X occurs and
most often accessed by the users and are level(X) is the depth of X in the pattern.
detected by the Sequential Pattern Generator.
The updater works in the following steps: B. Rank Improvement
Step 1: Given an input user query q and The rank of a page can be enhanced with the
matched documents D collected from the query help of its assigned weight. The original rank
processor, the cluster C is found to which the now becomes:
query q belongs. New Rank(X) =Previous Rank(X) +Weight(X)
Step 2: The sequential patterns of the concerned Where rank(X) is the existing rank value (Page
cluster are retrieved from the local repository Rank) of page X and weight(X) is the popularity
maintained by the Sequential Pattern Generator. given to X.
Step 3: The level weights are calculated for
every page X present in the sequential patterns. Weight Calculation and Rank Updation
Step 4: Final rank of a page d is computed if it From the above result (as shown in Fig. 4), a
happens to be present in the patterns of cluster Sequential Pattern is generated as AB C D
C. The improved rank is calculated as the weight of each URL can be determined as:
summing of previous rank and assigned weight Weight (A) =1.09
value. Weight (B) =1.09
By improving the ranks on the basis of user Weight (C) =0.54
feedbacks, the results of a search engine can be Weight (D) =0.36
optimized so as to better serve the user Table 2 shows the optimized the rank values and
requirements. At this time user can find the the new rank is used for the result presentation.
popular and relevant pages upwards in the result
list. Table 2. Rank Optimization with weight of pages
www.gtia.co.in 7
[5] D. Beeferman and A. Berger. Agglomerative

Clustering of a Search Engine Query log. In
KDD, pages 407-416, Boston, MA USA,
2000.
[6] Nitika Taneja and Rachna Chaudary, Query
Recommendation for Optimizing the Search
Engine Results in International Journal of
Computer Applications, Volume 50 –
No.13, July 2012
[7] O.R Zaine and A. Strilets., Finding similar
queries to satisfy searches based on query
traces. In proceedings of the international
4. Conclusion workshops on efficient web based
information system, France Sept. 2002.
In this paper, I try to find out method for log [8] Neelam Duhan, A.K Sharma.”Rank
mining. In today’s world million of people Optimization and Query Recommendation
accessing the web at same time. In billion of in Search Engine using Web Log Mining
entry are save in the log on web. These logs Technique. Journal of computing. Vol 2,
become wider day by day and this effect the Issue 12, Dec. 2010.
accessing of web. Many of the log information [9] A. Arasu, J Cho, H. Garcia-Molina, A.
are not of our need like when we access a web Paepcke, and S.Raghavan, Searching the
page then due to some error they not open. Lots web, ACM Transactions on Internet
of research has on this log. I find that in each Technology, Vol. No. 1, pp 97-101, 2001.
paper they firstly make crystal of this
information. Most of author use k mean
algorithm for that. After that different author use
different algorithm for find frequent set like
apriory, SOM etc. These algorithms are very
effective log mining.
References:
[1] Sana Siddiqui, Imran Qadri, IJARCSSE ,

Volume 3, Issue 12, December 2013.
Suneetha and Dr. R. Krishnamoorthi
IJCSNS International Journal of Computer
Science and Network Security, VOL.9 No.4,
April 2009
[2] L.K. Joshila Grace, V. Maheswari,
Dhinaharan Nagamalai International Journal
of Network Security & Its Applications
(IJNSA), Vol.3, No.1, January 2011.
[3] Jan Valdman, Technical Report No.
DCSE/TR-2001-04 July, 2001.
[4] J. Vellingiri and S. Chenthur Pandian,
Journal of Computer Science 7 (5): 683-689,
2011, Science Publications.
www.gtia.co.in 8
International Journal of Information Technology & Systems, Vol. 4; No. 1: ISSN: 2277-9825 (Jan-June 2015)
Multipath Routing in Autonomous Mobile Mesh Networks

Sangheethaa S*, Arun Korath**
*A.P., Dept. of Electrical and Computer Engg., School of Engineering and Tech., Dilla University, Ethiopia.
**Associate Prof, Dept. of Mgt. Studies, College of Business and Economics, Dilla University, Ethiopia.
Email: sangheethaa@gmail.com, arunkorath@gmail.com
Abstract: Autonomous mobile mesh networks are a combination of mobile ad hoc networks and mesh
networks. Mobile ad hoc networks are temporarily formed for a specific purpose, without any fixed
infrastructure. Mesh networks are fully connected networks with the support of fixed infrastructure.
Autonomous mobile mesh networks (AMMNET in short), is the framework designed for supporting mesh
networks, which are mobile still expects connectivity as opposed to ad hoc networks. This paper proposes a
multipath routing technology for AMMNETs which provides better packet delivery ratio in AMMNETs. The
simulations are done in ns2. The results prove that, multipath routing technology works well with AMMNET
framework.
Keywords: AMMNET, Routing, Multipath, NS2
1. Introduction organized in to following sections. Next section

briefs some of the existing approaches related to
Ad hoc networks [1, 2] are temporary wireless this paper. Section 3 discusses the AMMNET
networks normally formed at emergency model in detail. Section 4 explains the proposed
situations where there are no centralized multipath routing in AMMNET. Section 5 gives
authorities like base station or access points. Ad simulation results and finally, section 6
hoc networks have special characteristics. The concludes the paper along with some future
routers are free to move randomly and organize research scopes.
themselves arbitrarily; thus, the network's
wireless topology may change rapidly and 1. Related Work
unpredictably. Ad Hoc networks are used for
creating temporary networks. The main Some literatures [4, 5] study on improving the
application areas of Ad-hoc networks are; network topology in wireless mesh networks.
Creation of instant infrastructure, Disaster relief, But our paper deals with mobile mesh networks
network connection to remote areas where where routers also moving. Wireless sensor
installation of base station is not feasible. An ad covering is another area related to our paper.
hoc network consists of nodes which are free to But in sensor networks, tracking of sensors
move. The nodes may be anywhere, in airplane, become easy because the deployment of sensors
ships, trucks, cars even on every people. is fixed, and location of them is already known.
One important challenge in MANETs is to There is no much dynamicity to be considered
manage the network partitions due to dynamic while tracking sensors. But mobile mesh nodes
topology. This problem is addressed in the paper will move dynamically and tracking them is
[2] , and they have proposed a framework called more challenging than sensor nodes.
as AMMNET- autonomous mobile mesh Localization techniques [6,7,8,9] are another
networks. In a standard mesh network, mesh area where mobile clients’ location is tracked.
nodes may be mobile, but the mesh routers are But these are techniques to exactly know where
stationary. In AMMNET, the routers will move a client is. But our paper need just whether
along with mobile mesh clients to serve them in mobile client is there are not, it does not need
communication. We have conducted simulations the exact location of clients
and understood that, when AMMNET is used
with multipath routing, better throughput and 2. AMMNET
packet delivery ratio can be achieved. In [2] the
authors have used AODV [3] . This paper is
www.gtia.co.in 9
Autonomous mobile mesh network is the

framework for forwarding data between clients
in a mobile mesh networks. In normal mesh
networks, routers will be stationary. But in
AMMNET, routers will have to move along
with the mobile mesh nodes in order to provide
them seamless connectivity. Mesh nodes are
free to move, they broadcast beacon messages
periodically. On receiving this beacon
messages, the mesh router can detect presence
of a mobile mesh node. This framework
assumes group mobility patterns among the
mobile clients. When a group leader moves in a Fig. 2. Network with 3 groups of nodes.
particular direction, all the clients in that group
also starts moving in that direction. Three types The Fig. 1 shows an example for network with
of mobile mesh nodes are defined in [1] based some number of nodes. This is the initial
on their rolls. Intra group routers, intergroup network, where all nodes start from the same
routers and free routers. A mesh node is an intra location and one intra group router is enough to
group router, if it could serve at least one mobile support the entire network. The intra group
client by detecting its beacon message. These router is shown as a rectangle. The radio range
are responsible for communication and of this router is shown as the outer big circle.
monitoring inside a particular group. Inter group This means that entire network is under one
router as the name suggest, are responsible for group.
communication between the groups. A node is Fig. 2 shows the network with three groups.
called a free router if it not intra group or inter Each group has a intra group router to help
group router. communication within a group. Communication
Three algorithms are proposed in [1] to decide between groups is possible with the help of inter
which node will become intra, which will be group routers.
inter and which one will be a free router. Based [1] deals intensively in how to track a mobile
on these algorithms, a free router may become a client. It is not concentrating on increasing the
intra group router, or inter and vice versa. The throughput or increasing the packet delivery
main function of these routers will be to keep ratio. It clearly mentions that the main objective
track of the mobile clients in its range. When a is to keep track of the clients, rather than
mobile clients’ beacon is missing at a particular thinking whether data is delivered to correct
time, the router has to take necessary action to destination or not. But the ultimate aim of
detect that moving client and make tracking should be communication. AMMNET
arrangements for that client to be in the framework does not guarantee delivery of
connected area. [1] also proposes some packets to the correct destination. So this paper
optimization techniques to decide the number of utilizes the multipath routing concept to ensure
intra and inter group routers active at a time. that data is delivered to the destination and thus
Interested readers can know more about these increasing the throughput and packet delivery
algorithms from [1]. ratio of AMMNET. Next section explains the
concept of multipath routing and its use in
AMMNET.
4. Multipath Routing in AMMNET

Multipath routing [2] means routing the packets
through different available routes to the
destination. Existing on-demand protocols like
Fig. 1. Initial network with one intra group router AODV, DSR are giving shortest path to the
destination. They use route request message to
find a route. Route request is sent as a broadcast
message by the source which wants to send
data. Intermediate nodes and/or destination will
www.gtia.co.in 10
give route reply. On receiving the first route Fig. 5 shows the packet delivery ratio achieved
reply, or best route reply, the source starts for AMMNET and AMMNET with multi-path
transmission of data. The source is actually routing. From the graph it is clear that, using
receiving multiple route replies from the multipath routing in AMMNET increases the
intermediate nodes as well as from the packet delivery ratio by 20 %. When the traffic
destination. So it would be better, if more than increases, AMMNET is not able to achieve
one route is used for transmission of data to the better packet delivery ratio, where as in
destination. AMMNET with multi path routing, PDR
AOMDV [10] is a multipath routing version of increases. This is due to the fact that data is
the Adhoc on-demand routing protocol transmitted over multiple available paths. Not
(AODV). It is proved in [10] that AOMDV is that AMMNET uses AODV for updating its
superior to AODV and other existing routing routing table. But multipath routing uses
protocols. It utilizes all alternative paths, thus AOMDV.
not monopolizing any routes. It ensures delivery
of packets to the correct destination. Fig. 4 gives
an example for two alternative paths to reach
destination in group 2 from a source in group 3
through different intra group routers.
Group1 Group2
Intrarouter Intrarouter
Fig. 5. Packet Delivery ratio Vs. Simulation Time

Group3
Intrarouter
Source
Fig. 4. Two alternative paths to reach destination
Even though multiple paths are not available

outside the group, i.e. the routes are through the
same intergroup routers, still inside a group,
multiple paths are available.
5. Simulation Results
Simulations are done in ns2 [11] version 2.35.
The parameters set for simulation are given in Fig. 6. Throughput Vs. Simulation Time
Table 1. The parameters used for performance
comparison are, Packet delivery ratio, Fig. 6 shows that the throughput is increased by
Throughput, Average energy consumption of 40% in AMMNET with multipath routing,
nodes and End to end delay. compared to AMMNET. It is also due to data
transmission over multiple available paths. Fig.
Table 1. Simulation Parameters 7 shows the average energy consumption of all
nodes in the network. Surprisingly, it is also less
compared to AMMNET.
www.gtia.co.in 11
the number of nodes can be increased and the

effect of multipath routing in AMMNET can be
analyzed.
References:
[1] Wei-Liang Shen, Chung-Shiuan Chen,Kate

Ching-Ju Lin, Autonomous Mobile Mesh
Networks, IEEE Transactions On Mobile
Computing, Vol. 13, No. 2, 2014
[2] http://en.wikipedia.org/wiki/Wireless_ad_h
oc_network.
[3] C. E. Perkins and E. M. Royer, “Ad hoc On-
Fig. 7. Average energy consumption Vs. Simulation Demand Distance Vector (AODV) routing,”
time RFC 3561, Jul. 2003.
http://www.ietf.org/rfc/rfc3561.txt
Fig. 8 gives the average end to end delay for [4] J. Tang, G. Xue, and W. Zhang,
transmission of the packet from source to Interference-Aware Topology Control and
reception at the end. End to end delay is also QoS Routing in Multi-Channel Wireless
less in multipath routing compared to Mesh Networks, Proc. ACM MobiHoc,
AMMNET without multipath routing. Thus 2005.
from the graphs it is clear that, AMMNET when [5] A. Naveed, S. Kanhere, and S. Jha,
used with multi path routing technology, Topology Control and Channel Assignment
improves PDR, throughput and efficiency. in Multi-Radio Multi-Channel Wireless
Mesh Networks, Proc. IEEE Int’l Conf.
Mobile Adhoc and Sensor Systems
(MASS), 2007.
[6] A. Savvides, C.-C. Han, and M.B.
Strivastava, Dynamic Fine-Grained
Localization in Ad-Hoc Networks of
Sensors, Proc. ACM MobiCom, 2001.
[7] C. Savarese, J.M. Rabaey, and K.
Langendoen, Robust Positioning
Algorithms for Distributed Ad-Hoc
Wireless Sensor Networks, Proc. USENIX
Ann. Technical Conf., 2002.
[8] J. Liu, Y. Zhang, and F. Zhao, Robust
Distributed Node Localization with Error
Management, Proc. ACM MobiHoc, 2006.
[9] H.T. Kung, C.-K. Lin, T.-H. Lin, and D.
Vlah, Localization with Snap-Inducing
Shaped Residuals (SISR): Coping with
Fig. 8. End to end delay Vs. simulation time Errors in Measurement, Proc. ACM
MobiCom, 2009.
6. Conclusion [10] Mahesh K. Marina,Samir R. Das, Ad hoc
on-demand multipath distance vector
This paper has added multipath routing concept routing in Wireless Communications And
along with AMMNET and proved that, using Mobile Computing Wirel. Commun. Mob.
multipath routing in AMMNET will increase the Comput.2006;6 :969–988 Published online
throughput as well as packet delivery ratio. The in Wiley InterScience.
simulations are done for just 39 nodes, In future,
www.gtia.co.in 12
Comparative Study of Enhanced Interior Gateway Routing

Protocol (EIGRP) and Open Shortest Path First (OSPF) for Data
Transmission through A Network System for Real Time
Application
Philip O. Odion*, David Bagudu**
*Department of Computer Science, Nigerian Defence Academy, Kaduna, Nigeria.

** Department of IT, National Open University of Nigeria, Kaduna Study Centre, Kaduna.
Email: *podion2012@gmail.com, **davebagudu@gmail.com
Phone no: *07062777976, **08027411996
Abstract: Routing protocol is taking a vital role in the modern communication networks. The performance of
each routing protocol is different from each other. In the context of routing protocol performance, each of them
has different architecture, adaptability, route processing delays and convergence capabilities. Among different
routing protocols, Enhanced Interior Gateway Routing Protocol (EIGRP) and Open Shortest Path First (OSPF)
have been considered as the pre-eminent routing protocols for the real-time application. EIGRP and OSPF are
dynamic routing protocols used in practical networks to disseminate network topology towards the adjacent
routers. This paper is based on a comparative performance analysis between EIGRP and OSPF for real time
applications during data transfer. In order to evaluate the performance of EIGRP and OSPF, two network
models that are configured with EIGRP and OSPF, respectively were assessed. The comparative evaluation
shows that EIGRP routing protocol provides a better performance than OSPF routing protocol for real time
applications.
Keywords: Comparative Study, Enhanced Interior Gateway Routing Protocol (EIGRP), Open Shortest Path
First (OSPF), Data Transmission, Network System, Real Time Applications.
1. Introduction EIGRP is a Cisco propriety distance-vector

protocol based on Diffusing Update Algorithm
In modern era, computer communication (DUAL). On the other hand, OSPF is a link-state
networks are growing rapidly day by day. interior gateway protocol based on Dijkstra
Communication technology facilitates provision algorithm (Shortest Path First Algorithm) [2]. A
of such services as file transferring, print performance comparison between router
sharing, video streaming and voice protocols has been research topics for many
conferencing. Internet is a global system of years. Since OSPF routing process is less
interconnected computer networks. Today complicated than one in Routing Information
internet is playing a vital role in communication Protocol (RIP), the routing and convergence
networks. Computer communication network are times are shorter by using OSPF than RIP in the
based on a technology that provides across the same network but different area. Also, OSPF is
technical infrastructure, where routing protocols a standard protocol in all routers until CISCO
are used to transmit packets across the internet. has developed additional EIGRP protocol in
Routing protocols specify how routes their router. EIGRP uses less memory and CPU
communicate with each other by disseminating times than OSPF, and EIGRP also uses distance
information [1]. The router has prior knowledge vector instead of link state technique which is
about the adjacent networks, which can assist in used in OSPF.
selecting the routes between two nodes. There Distance-vector protocols are based on
are various types of routing protocols being calculating the direction and distance to any link
widely used. Among difference routing in a network. "Direction" usually means the next
protocols, Enhanced Interior Gateway Routing hop address and the exit interface. "Distance" is
Protocols (EIGRP) and Open Shortest Path First a measure of the cost to reach a certain node.
(OSPF) has been considered as the pre-eminent The least cost route between any two nodes is
routing protocols for real-time applications. the route with minimum distance. Each node
www.gtia.co.in 13
maintains a vector (table) of minimum distance network in the form of a graph showing which
to every node. The cost of reaching a destination nodes (i.e., routers, switches, computers and
is calculated using various route metrics. other network devices) are connected to which
Routing Information Protocol (RIP) uses the hop other nodes. Each router then independently
count of the destination whereas Interior calculates the best next hop for every possible
Gateway Routing Protocol (IGRP) takes into destination in the network using only its local
account other information such as node delay copy of the map and without communicating
and available bandwidth. Updates are performed with any other node. The collection of best next
periodically in a distance-vector protocol where hops forms the routing table for the router. Link
all or part of a router's routing table is sent to all state routers use Dijkstra's algorithm to calculate
its neighbors that are configured to use the same the lowest cost paths, and normally update other
distance-vector routing protocol. RIP supports routers with which they are connected only
cross-platform distance vector routing whereas when their own routing tables change. Invented
IGRP is a Cisco System proprietary distance by the Dutch computer scientist Edger Dijkstra,
vector routing protocol. Once a router has this this algorithm is a technique for finding the
information, it is able to amend its own routing shortest path between two nodes in a weighted
table to reflect the changes and then inform its graph. Link state routing is an improvement
neighbors of the changes. This process has been over distance vector routing protocols, the other
described as ‘routing by rumour because routers main class of routing protocols for packet
are relying on the information they receive from switched networks. These protocols, such as RIP
other routers and cannot determine if the (routing information protocol), normally use
information is actually valid and true. There are only a single metric (such as hop count) and
a number of features which can be used to help which exchange all of their table information
with instability and inaccurate routing with all other routers on a regular basis. Link
information. state routing normally requires more memory
Enhanced Gateway Protocol (EGP) and Border and processing than distance vector routing, but
Gateway Protocol (BGP) are not pure distance- there is less transmission overhead and it reacts
vector routing protocols because a distance- more quickly to changes in network
vector protocol calculates routes based only on connectivity. Despite their differences, both
link costs whereas in BGP, for example, the algorithm types perform well in most
local route preference value takes priority over circumstances [4].
the link cost [3]. OSPF rerouting time is an experiment objective
Link state routing is a complex routing in the network when there is a failure connection
technique in which each router shares link with the Border Gateway Protocol (BGP) in
information with other routers about the reach two separate areas. Optimized Network
ability of other networks and the metric to reach Engineering Tool (OPNET) is used to do
the other networks in order to determine the best simulation for comparison three protocols
path. Routing is the process of moving packets (OSPF, RIP, and EIGRP) and the results show
across a network from one host to another. It is that EIGRP has the smallest routing time,
usually performed by dedicated devices called memory time and CPU time, and the least delay
routers. Packets are the fundamental unit of time, but OSPF can transmit the most data.
information transport in all modern computer OSPF and EIGRP are also compared by OPNET
networks, and increasingly in other simulation for routing time, delay time and
communications networks as well. They are packet loss in video streaming and results show
transmitted over packet switched networks, that EIGRP has less packet loss but OSPF has
which are networks on which each message (i.e., less delay and more throughput. The rerouting
data that is transmitted) is cut up into a set of time after there is a failure link is experiment
small segments prior to transmission. The metric with many methods such as equal cost
is based on number of hops, link speeds, traffic multipath, loop free alternate, and U-turn
congestion and other factors as determined by alternate which are compatible in OSPF protocol
the network designers. A hop is the trip that a [5].
packet takes from one router to another as it EIGRP and OSPF are dynamic routing protocols
traverses a network on the way to its destination. used in practical networks to disseminate
In link state routing, every router on the network network topology to the adjacent routers. There
receives a map of the connectivity of the are various numbers of static and dynamic
www.gtia.co.in 14
routing protocols available but the selection of The formula that EIGRP uses to calculate scale
appropriate routing protocol is most important bandwidth is:
for routing performance. The right choice of
routing protocol is dependent on several Delay= Dn*256 ------------------------- (4)
parameters [6].
where: Dn represents in microseconds
2. Enhanced Interior Gateway and sum of the delays configured on the
interface to the destination [8].
Routing Protocol (EIGRP) versus
Open Shortest Path First (OSPF) While the Open Shortest Path First (OSPF) is a
routing protocol that was developed by the
Enhanced Interior Gateway Routing Protocol Interior Gateway Protocol (IGP) working group
(EIGRP) is a CISCO proprietary protocol, which of the Internet Engineering Task Force for
is an improved version of the interior gateway Internet Protocol (IP) networks. OSPF is a link
routing protocol (IGRP). EIGRP is being used as state routing protocol that is used to distribute
a more scalable protocol in both medium and information within a single Autonomous
large scale networks since 1992. EIGRP is said System. In 1989, the first version of OSPF was
to be an extensively used IGRP where route defined as OSPFv1, which was published in
computation is done through Diffusion Update Request for Comment (RFC) 1131. The second
Algorithm (DUAL). However, EIGRP can also version of OSPFv2 was introduced in 1998,
be considered as hybrid protocol because of which was defined in RFC 2328. In 1999, the
having link state protocol properties [7]. third version of OSPFv3 for IPv6 was released
With the use of total delay and the minimum in RFC 2740 [7].
link bandwidth, it is possible to determine the The path cost of an interface in OSPF is called
routing metrics in EIGRP. Composite metrics, metric that indicates standard value such as
which consists of bandwidth, reliability, delay, speed. The cost of an interface is calculated on
and load are considered to be used for the the basis of bandwidth. Cost is inversely
purpose of calculating the preferred path to the proportional to the bandwidth. Higher
networks. The EIGRP routing update takes the bandwidth is attained with a lower cost.
hop count into account though EIGRP does not
include hop count as a component of composite Cost = 108/bandwidth in bit per second
metrics. The total delay and the minimum (bps) ---------------------------------(5)
bandwidth metrics can be achieved from values
which are put together on interfaces and the where the value of 108 is 100000000 in bps is
formula used to compute the metric is followed
called reference bandwidth based on by default.
by:
256*(K1*BW+K2*BW/256-Load+K3*Delay)
*K5/K4+Reliability ------------------------------ (1) 2.1 Advantages and Disadvantages of
For weights, the default values are: EIGRP
‫ܭ‬1=1, ‫ܭ‬2=0, ‫ܭ‬3=1, ‫ܭ‬4=0, ‫ܭ‬5=0. There are some advantages provides by EIGRP
where: K1, K2, K3 ,K4, and K5 are EIGRP Metric as follows:
Weight. I. Easy to configure.
These default values efficiently trimming down II. Loop free routes are offered.
the above formula to: III. It keeps a back up path in the network to
256∗ (BW + ‫ )ݕ݈ܽ݁ܦ‬------------------------------ (2) get the destination.
A significant and totally non-obvious fact is that, IV. Multiple network layer protocols are
if ‫ܭ‬5=0, the term K5/K4+Reliability is not used included.
(i.e. take as 1). V. EIGRP convergence time is low and it is
responsible for the reduction of the
EIGRP uses to calculate scale bandwidth (Bw)
bandwidth utilization.
as:
VI. It can work with Variable Length
Bw = (10000000/Bn)*256 ----------------------- (3)
Subnet Mask (VLSM) and Class Less
Where: (݊) is in kilobits per seconds and Inter Domain Routing (CIDR).
represents the minimum bandwidth on VII. EIGRP also supports the routing update
the interface to destination. authentication.
www.gtia.co.in 15
The disadvantage of EIGRP is that Routers from

other vendors are not able to utilize EIGRP
2.2 Advantages and Disadvantages of OSPF
The advantages of OSPF are:
I. OSPF always determines the loop free
routes.
II. If any changes occur in the network it
updates fast.
III. OSPF minimizes the routes and reduces
the size of routing table by configuring
area.
IV. Low bandwidth utilization.
V. Multiple routes are supported.
VI. Support variable length subnet masking.
It is suitable for large network. Fig. 1. DUAL placed in the network
The disadvantages of OSPF are:

I. Difficult to configure.
II. Link state scaling problem.
III. More memory requirements.
3. EIGRP and OSPF Algorithms

In EIGRP, all route calculations are done by the
DUAL. One of the tasks of DUAL comprises
keeping a table, known as topology table, which
includes all the entries found from the loop-free
paths advertised by all routers. DUAL selects
the best loop-free path (known as successor Fig. 2. Failure link of network topology
path) and second best loop free path (known as
feasible path) from the topology table by using
OSPF is a link state routing protocol that uses a
the distance information. It then saves this into
shortest path first algorithm to calculate the least
the routing table. The neighbour of the least cost
cost path to all known destinations. Dijkstra
route to the destination is called a successor. In
algorithm is used for calculating the shortest
case of having no loop-free path in the topology
path. Several procedures of this algorithm are
table, the re-calculation of the route must be
given below:
done and then the DUAL inquire its neighbors.
I. For any change in the routing information, a
This occurs during re-calculation for searching a
link state advertisement is generated by a
new successor. Although, the re-calculation of
router. This advertisement provides all link
the route does not seem to be processor-
states on that particular router with
intensive, it may have an effect on the
information.
convergence time and consequently, it is useful
II. All routers exchange Link-State
for avoiding needless computations. In case of
Advertisement (LSA) by flooding. The link
having any Feasible Successor (FS); an adjacent
state update is received by each router and
router which is used to offer a loop free backup
preserves a copy of link state database in it.
path to the destination by fulfilling the
This link state update propagates to all other
conditions of Feasible Condition (FS), DUAL is
routers.
used for avoiding any needless re-computation.
III. After creation of database of each router,
Fig.1 shows the understanding on how the
DUAL converges. This example aims at router routers start calculating shortest path tree to
K for the destination only. The cost of K (hops) the destinations. In order to find the least
coming from each router is presented below: cost path, the router uses Dijkstra algorithm.
IV. If any changes occurred in the OSPF
network such as link cost, new network
www.gtia.co.in 16
being added or deleted, Dijkstra algorithm is

recalculated to find the least cost path.
Every router uses this algorithm at the root of
the tree in order to find the shortest path on the
basis of cost to reach the destinations. Fig. 3
shows a network diagram that is indicated with
the interface cost to find out the shortest path.
Consider the network in Fig. 3 below routing
table, the network 128.213.0.0 is learned via 10
equal-cost paths (192.123.111.0 and
222.211.1.0.0), the traffic to this network will be Fig. 4: OSPF Network (Cisco, 2002)
load-balancing.
Router RTA will forward packed from 10 to 5
via 192.213.11.0 and 10 to 10 via 222.211.10.0 4. Design and Configuration of
because they are shortest path. EIGRP and OSPF Network
Consider the network in Fig. 4 running OSPF.
Assume the link between R3 and R5 fails. R3 Topology
detects link failure and sends LSA to R2 and R4.
Since a change in the network is detected traffic 4.1 EIGRP Scenario
forwarding is suspended. R2 and R4 updates
In this scenario, EIGRP routing protocol is
their topology database, copies the LSA and
flood their neighbors. By flooding LSA all enabled first for all routers on the network. After
devices in the network have topological configuring routing protocols, individual
awareness. A new routing table is generated by Discrete Event Simulation (DES) statistics is
all routers by running Dijkstra algorithm. The chosen to select performance metrics and to
traffic is now forwarded via R4. evaluate the behavior of this routing protocol.
Then simulation run time is set for 900 seconds.
The network model for EIGRP is shown in the
Fig. 5.
Fig. 5: Network model for EIGRP (Cisco, 2002)
Fig. 3: Shortest path tree (Cisco, 2002) 4.2 OSPF Scenario

First task is to set routing configuration OSPF as
a routing protocol for this network topology.
Then individual DES statistics is chosen that
would be viewed in the results from the DES
www.gtia.co.in 17
menu. Finally, time duration to run the References:

simulation is set for 900 seconds. The network
model for OSPF is shown in the Fig. 6. [1] Thorenoor S.G., Dynamic Routing Protocol
Implementation Decision between EIGRP
and RIP Based on Technical Background
Using OPNET Modeler, A Proceeding of
2010 Second International Conference on
Computer and Network Technology
(ICCNT 2010), Bangalore, India, pp 191-
195, 2010.
[2] Cisco, OSPF Design Guide
http://www.cisco.com/en/US/tech/tk365/tec
hnologies_White_paper09186a0080094e9e.
shtml, 2006. (Accessed: 4th November,
2014).
[3] Hedrick C., Routing Information Protocol:
A perspective of Internet Engineering Task
Force, Journal of International Federation of
Automatic Control (IFAC), 50(1), p.3-43,
2014.
[4] http://www.techtarget.com/contributor/
Margaret-Rouse/2014 (Accessed: 10th
Fig. 6: Network model for OSPF (Cisco, 2002)
November, 2014).
[5] Wendell D.B., A collection of materials for
5. Conclusion the study of intranet and internet
networking, GTCC, 2010. http://www.gtcc-
A comparative analysis of EIGRP and OSPF it.net/billings/eigrp.html
was studied. There are two network models, [6] Cisco, Internet Technology Handbook, 2002.
which are configured and run as first scenario http://www.cisco.com (Accessed 4th
with EIGRP alone and second one with OSPF November, 2014).
alone. Background utilization in each network [7] Todd, Cisco Certified Network Associate,
by varying the link utilization and analyze the Network Associate Study Guide: Exam 640-
variation of each defined parameters were 802, Deluxe, 5th edition, 2007.
considered. The data transmission time before [8] Brent D.S., CCNP BSCI Official
and after there is a failure link using EIGRP Certification Guide, Fourth Edition, pp 64-
protocol is less than OSPF protocol. Also the 65, 2008.
rerouting time for finding new transmission path
in EIGRP is shorter than OSPF. Based on
EIGRP networking topology, it shows that the
network convergence time is faster than OSPF
networks, because EIGRP network can learn the
topology information and updates more rapidly.
Data packets in EIGRP network reach faster to
the destination compared to OSPF network.
www.gtia.co.in 18
Review of SiGe Double Gate N-MOSFET

Akshat Singh*, Dhiraj Kapoor*, Rajiv Sharma**
* Department of Electronics and Communication Engineering, Vaish College of Engineering, Rohtak, India
** Professor, Department of Electronics and Communication Engineering, SGTIET, Gurgaon, India
Email: singhakshat001.as@gmail.com, kapoordhiraj79@gmail.com, rsap70@rediffmail.com
Abstract: The current SOI MOSFET technology supports the scaling down to nanometer regime but it is
unable to keep the performance level high as it suffers from the restrictions like effective interface coupling,
channel orientation, channel mobility, off-state leakage, switching delay and latch up. The additional parameters
such as injection velocity, short channel length effects, narrow channel effects, body effects, hot carrier effect,
depletion width effect, substrate doping effect and Band-to-Band Tunneling (BTBT) also results in decrease of
inversion charge, increase in leakage current and decrease in the drive current. In order to minimize these
effects, double gate came into existence, which relies on the exploration of novel higher mobility channel
materials such as SiGe, which might perform even better than Si. This paper reviews the double-gate N-
MOSFET using SiGe as channel material.
Keywords: Scaling, GAA, SOI, Fin-FET, Double-gate MOSFETs
1. Introduction Although the SOI technology is highly attractive

in terms of speed, power consumption and
The Si-based Single Gate MOSFETs have been scaling. Most prominent of these is voltage
the most important building blocks of the scaling which is highly important component of
Integrated Circuits (ICs), since they were first MOSFET scaling because it maintains constant
presented in 1960. Today's Integrated Circuits electric field, which is important for reliability,
(ICs) makes use of MOSFETs as the basic and it lowers transistor power, which is required
switching element [1]. Even though basic to maintain constant power density [4].
structure of the device is same since constructed. Nevertheless, it is well defined that, thin film
Since development of Moore’s Law, physical SOI MOSFETs do not scale well to short channel
size has been continuously decreased by factor of lengths because of lack of a conducting plane
two per year with increase in complexity of under the silicon region. If we place a conducting
components as shown in Fig.1, because of which, film beneath the device, it requires only a thin
faster and highly complex chips with lowered insulator for making the silicon film effective.
cost per transistor have been constructed [2].
With development of concept of device scaling
over the past few decades, needs such as
performance and power consumption in VLSI
circuits can be meet [3]. However, device
structures, such as bulk MOS transistors, are
moving towards the basic physical limits.
Moreover, as device dimensions reduces to
micrometers and below, the limits of MOS
structures increasing because of strong short-
channel effect which causes the decrease in
performance which makes it quite necessary to
move towards new device structures for
sustaining the growth of the VLSI industry in the
nano-scale generations. For this purpose, Fig. 1. Moore’s Law [5]
Double-Gate Fully-Depleted (DGFD) Silicon-on-
Insulator (SOI) transistors can be considered as a Thus, a continuous effort is carried out for
good technology choice [3]. increasing current drive and to have better
control over short-channel effects, silicon-on-
www.gtia.co.in 19
insulator (SOI) MOS transistors have been Moreover, various SiGe-channel n-MOSFETs
generated from classical, planar, single-gate with high-K metal gate stacks together for
devices into three-dimensional devices with a providing enhanced value of drive current and
multi-gate structure (double-, triple- or transconductance and performance advantage
quadruple- gate devices). It must be noted that, in depends mainly on the ability to scale a shorter
most cases, the term "double gate" refers to a channel length and its structure with the high-K
single gate electrode, which is present on two dielectric gate stack is shown in Fig. 4 [10].
opposite sides of the device [6]. Ideally, double-
gate SOI and GAA can be viewed as important
structures for minimizing short channel effects
like DIBL, but they are found to be incompatible
with SOI CMOS standard [6]. The different gate
configurations of multi-gate are shown in Fig. 2
[7].
Fig. 4. Structure of a Ge channel DG MOSFET
with the high-k dielectric gate stack
2. Motivation
In a continuous effort to improve the
Fig. 2. Different gate configurations for SOI devices: semiconductor device features, down scaling of
1) single gate; 2) double gate; 3) triple gate; 4)
the devices is the major demand of VLSI
quadruple gate; 5) new proposed Pi-gate MOSFET [7]
industries. However, as on entering in sub-
micrometer regime the short channel effects
For DGMOS transistors, back and front gates are
hinder the device performance. Double gate
interconnected for achieving more ON current
MOSFETs are replacing the conventional MOS
(Ion) and to minimize short-channel effects. It
devices due to their capability to attain higher
must be noted that, larger is the current, more it
speeds and reduced short channel effects (SCE’s).
will be helpful in decreasing circuit delay in
The work done so far completely stress on
critical paths [8].
providing the explanation regarding the transport
If we plot the I-V Characteristics of the double
in Si MOSFETs through simple band structure
gate, for Si and consider the cases of connected
and mobility calculations. However, as we carry
front and back gates (VF gate=VB gate) and
out scaling on MOSFETs with very short channel
independent gate control (with back gate bias
lengths, it is found that the effect of the high-field
voltage of zero) it is found to be one as shown in
transport, density of states (DOS), band structure,
Fig. 3 [9]. When these gates are connected
mobility and effective mass, plays out important
together, two inversion channels are formed: one
role in carrying out the calculations of the
at the top of the silicon film near the top gate
eventual drive current, we require expanded
insulator, and the other at the bottom near the
investigation [11].
bottom insulator [9].
In addition, based on the analysis carried for 1-3
nm thick SOI MOSFET, in double-gate (DG)
mode and single mode, strong interface coupling,
ideal subthreshold swing, improved
transconductance and volume inversion
demonstrates the superiority of DG-MOSFETs
over SG-MOSFETs[12].The scaling of MOSFETs
is defined based on the criteria of International
Technology Road Map for Semiconductors
(ITRS), which enables us to take part in the
worldwide semiconductor market, which is
growing at a phenomenal rate. If the device
dimensions are continuously scaled down, then we
Fig. 3. I-V Characteristics [9] can achieve high performance and low power
www.gtia.co.in 20
operation, it is found that thinner gate oxides are 1. Body Effect

needed now for nanoscale devices [13]. Whenever reverse biasing from well-to-source
Since, MOS technology scaling is continued over junction of a MOSFET transistor is done, the
past decades, short-channel effects (SCEs) and bulk depletion region width increases, which
variation in transistor performance creates further increases the threshold voltage. The effect
difficulties for the planar bulk MOSFET design of body bias can be considered in the threshold
so, alternative MOSFET designs are eventually voltage equation 1 and equation 2 given below
required in order to extend transistor scaling into [16].
the 20-nm gate length region. For which suitable Vth=Vfb+2ψB+√2εSiqNa(2ψB+vbs)/Cox (1)
example is DG, that utilizes the double-gate where
structure (which increases capacitive coupling Vfb is the flat-band voltage;
between the gate and the channel) to minimize Na is the doping density in the substrate [16];
SCEs and variation because of random dopant ψB = (KT/q)ln(Na/ni) (2)
fluctuation (as long as the channel doping level is is the difference between the Fermi Potential and
low) [14]. the Intrinsic Potential in the substrate [16]
In comparison with the bulk FinFET design, the
double-gate bulk MOSFET design can be used to 2. Short Channel Effects
achieve comparable intrinsic delay. Even though The device becomes short-channel device, if
the optimized FinFET design has 12% higher effective length is approximately equal to source
[14], ON-state current compared with the and drain junction depth. In general, these effects
optimized double-gate bulk MOSFET, the double- are generated whenever electric field lines from
gate bulk MOSFET has lower parasitic source and drain affect control of channel region
capacitance. The channel stripe height/width by gate.
aspect ratio is lower for the double-gate bulk Various short channel effects includes a) DIBL
MOSFET (0.8 versus 2.17) than for the bulk b) GIDL
FinFET, and the retrograde channel doping profile
can be made less steep for the double-gate bulk a) DIBL
MOSFET (4 nm/dec versus 1 nm/dec) [14] in It stands for Drain Induced Barrier Lowering.
comparison with the bulk FinFET to minimize DIBL occurs when the depletion regions of the
delay [14]. drain and the source interact with each other near
In addition, SiGe alloys are found to be interesting the channel surface to lower the source potential
candidates in thermoelectric applications because barrier. When a high drain voltage is applied to a
of different impact of the material composition on short-channel device, it lowers the barrier height,
mobility and thermal conductivity. Moreover, resulting in further decrease of the threshold
various SiGe-channel n-MOSFETs provides voltage [16].
enhanced value of drive current and
transconductance and performance advantage b) GIDL
depends mainly on the ability to scale a shorter It stands for Gate Induced Drain Leakage. The
channel length. GIDL was first demonstrated in MOSFET in the
late 1980s by the Berkeley group, which is
3. Limitations of Scaling defined as the parasitic current, which arises
from interband/Band-to-Band Tunneling
For achieving high performance and low power mechanisms. This leakage mechanism mainly
consumption, MOS devices have been influences the device performance by limiting the
continuously scaled down since past 30 years. off state current to a minimum value [17].
However, the scaling down of threshold voltage Due to these scaling effect MOSFETs
results in the substantial increase in sub threshold experienced various challenges such as reduction
leakage current. High leakage current in deep- in gate controllability over the channel and
sub micrometer regions is becoming a significant increase leakage current. To address these
contributor to power dissipation of MOSFET scaling challenges evolutions from conventional
circuits as threshold voltage, channel length, and MOSFET to multigate MOSFET have been
gate oxide thickness are reduced. This gives rise conceded. Mostly multigate devices are grown
to over the silicon substrate but now, the multigate
1. Body Effect devices such as DG MOSFETs using SiGe [18]
2. Short Channel Effects [15] are becoming extensively attractive to increase
www.gtia.co.in 21
current drive and better control on short-channel Its operation starts with off state condition, where
effects. electrons can’t be transferred from source to the
body without threshold voltage so gate potential
4. Double Gate SiGe MOSFET is applied on the both terminals as it start to
increase the potential of the surface start to
The important characteristic for SiGe is lattice increase which reduces the source limit and
thermal conductivity, which decreases notably forward biases the source-body junction. As the
with increase in Germanium content until 50% source limit is reduced, there is exponential
[18]. Above, 50% this reverses to give increase in injecting electron from source to body
conclusion that the value is moving towards pure is noticeably seen because of increase in the
germanium [18].Using this property, construction energy level of electrons [19].
of device with property of has been done [18].
Table 1. Comparison between Si DG NMOS and SiGe DG NMOS [22]
Under this concept, channel is under the control

of two gates, i.e. the front and back gates. This
helps in reducing the effect of the drain field in
reaching the source and thus results in reduced
short channel effects (SCE) [20].
Currently, germanium (Ge) has got considerable
attention largely because of its higher carrier
mobility, specifically for the hole mobility which
is nearly four times larger than that in Si.
Now, we plot the I-V Characteristics of the
double gate, for SiGe at VDS equal to 0.5V and
channel length variation from 7nm to 20nm, it is Fig. 5. Drain current versus gate voltage for
found that the reduction in channel length leads conventional DG MOSFET SiGe when channel length
to shifting of characteristics to the left and it is varies from 7nm to 20 nm at VDS=0.4V [21]
clear that when the channel length is smaller
than 15nm, the subthreshold current rises 5. Conclusion
drastically. As the drain current depends on the
channel length, the threshold voltage of the So far, all efforts have been made through the Si
devices should also be channel length dependent as channel material because Si posses quite large
as shown in Fig. 5 [21]. advantages than Ge. If we study n analytical
However, if keep on looking into future model for double gate (DG) MOSFET it is found
nanoscale MOSFETs, it becomes important to that if we use SiGe, it is found that DG MOSFET
inspect novel channel materials such as SixGe(1-x). using SiGe has higher performance over DG
Thus, for value of x=0.2, it becomes Si0.2Ge0.8, MOSFET using Si because much lower leakage
which is similar to Si=20% and Ge=80% [22] as current is obtained in case of DG MOSFET using
described by Table 1. SiGe because of its effective electron mobility
and hole mobility. Under this concept, channel is
under the control of two gates, i.e. the front and
back gates. This helps in enhancing the
www.gtia.co.in 22
performance in nanoscale MOSFETs; it is found (100nA/µm) and mainly due to the enhancement
that SiGe shows the best performance in terms of / delay reduction (~3X compared to Si) [11] and
driver current and enhancement/delay reduction very low off-state leakage (<1nA/µm) [11]. Thus,
(~2X compared to Si) [11]. However, minimum SiGe can be utilized as channel material in
off-state leakage due to BTBT is rather high future.
References: [11] T.Krishnamohan C.Jungemann, D. Kim, E.

Ungersboeck, S. Selberherr, A.T. Pham, B.
[1] D. Kim, Doctorate of Philosophy Meinerzhagen, P. Wong, Y. Nishi, K.C.
Dissertation on Theoretical Performance Saraswat, High Performance, Uniaxially-
Evaluations of NMOS Double Gate FETs Strained Si & Ge, Double-Gate P-MOSFETs,
with High Mobility Material: Strained III-V, Microelectronic Engineering, vol. 84, no. 9,
Ge and Si, Department of Electrical and pp.2063–2066, 2007
Engineering and the Committee, Stanford [12] T. Ernst, S. Cristoloveanu, G. Ghibaudo, T.
University, 2009. Ouisse, S. Horiguchi, Y. Ono, Y. Takahashi,
[2] G.E. Moore, Cramming more components K. Murase, Ultimately Thin Double-Gate SOI
onto the integrated circuits (Reprinted from MOSFETs, IEEE Transactions on Electron
Electronics, pp. 114-117, April 19, 1965), Devices, vol. 50, no. 3, pp. 830-837, 2003
Proceedings of the IEEE, vol. 86, no. 1, pp. [13] M.I. Alamgir, A.U.H. Gulib, K. M. U.
82-85, 1998. Ahmed, Performance Analysis of Dg Mosfets
[3] R. Zhang, K. Roy and D.B. James “Low- with High-K Stack on Top & Bottom Gate”,
Power High-Performance Double-Gate Fully International Journal of Scientific &
Depleted SOI Circuit Design”,IEEE Technology Research, vol. 1, no. 5, pp. 96-
Transactions on Electron Devices, vol. 49, 102, 2012.
no. 5, pp. 213-218, 2002. [14] B. Ho, X. Sun, C. Shin T.J.K. Liu, “Design
[4] M. Bohr, A 30 Year Retrospective on Optimization of Multigate Bulk MOSFETs”,
Dennard’s MOSFET Scaling Paper, IEEE IEEE Transaction on Electron Devices, vol.
Solid-State Circuits Society, vol. 9, no. 17, 60, no.1, pp. 28-33, 2013.
pp. 11-15, 2007. [15]S G.Narendra and A. Chandrakasan,
[5] D.A. Pucknell, K. Eshraghian, Basic VLSI “Leakage in Nanometer CMOS
Design, Third Edition, Prentice Hall of India Technologies”, First Edition, Springer Science
Publication, 1994 . + Business Media Publications, 2006.
[6] J.P. Colinge, Multi-gate SOI MOSFETs, [16] K. Roy, S. Mukhopadhyay, H.M. Meimand,
Microelectronic Engineering, vol. 84, no. 9, Leakage Current Mechanisms and Reduction
pp. 2071- 2076, 2007 in Techniques Deep- Submicrometer CMOS
[7] J.T. Park, J.P. Colinge, Multiple-Gate SOI Circuits, Proceedings of the IEEE, vol. 91, no.
MOSFETs: Device Design Guidelines, IEEE 2, pp. 305-327, 2003.
Transactions on Electron Devices, vol. 49, [17]Q. Rafhay, C. Xu, P. Batude, M. Mouis, M.
no. 12, pp. 2222-2229, 2002. Vinet, G. Ghibaudo, Revisited approach for
[8] A. Chandrakasan, W.J. Bowhill, F. Fox, Characterization of Gate Induced Drain
Design of High-Performance Microprocessor Leakage, Solid-State Electronics, vol. 71, no.
Circuit, IEEE Press, First Edition, Wiley 1, pp. 37–41, , 2012.
Publications, 2001. [18]M. Wagner, Doctorate of Philosophy
[9] H. Mohammadi, H. Abdullah, C. Fu Dee, A Dissertation on Simulation of Thermoelectric
Review on Modeling the Channel Potential Devices, Department of Electrical
in Multi-Gate MOSFETs, Sains Malaysiana Engineering and Committee, Vienna
Publications, vol. 43, no. 6, pp. 861-866, University of Technology, 2007.
2014. [19]D. Sahu, A. Tirkey, Bachelors of Technology
[10] S. Bhattacherjee, A. Biswas, Performance Project on Design and Simulation of Double
Analysis of Long Ge Channel Double Gate Gate FETs using ATLAS, Department Of
(DG) P-MOSFETs with High-K Gate Electronics And Communication Engineering,
Dielectrics Based on Carrier Concentration National Institute of Technology, Rourkela,
Formulation, Microelectronics Reliability, 2014.
vol. 51, no. 6, pp. 1105-1112, 2011. [20]S. Subramaniam, R.N.A. Wale, S.M. Joshi,
Drain Current Models for Single-Gate
www.gtia.co.in 23
MOSFETs & Undoped Symmetric &

Asymmetric Double-Gate SOI MOSFETs and
Quantum Mechanical Effects: A Review,
International Journal of Engineering Science
and Technology, vol. 5, no. 1, pp. 96-105 ,
2013.
[21] H. Valinajad, R. Hosseini, M.E. Akbari,
Electrical Characteristics of Strained Double
Gate MOSFET, International Journal of
Research and Reviews in Applied Sciences,
vol. 13, no. 2, pp. 436-442, 2012.
[22] T. Krishnamohan, Doctorate of Philosophy
Dissertation on Physics and Technology of
High Mobility, Strained Germanium
Channel, Heterostructure MOSFETS,
Department of Electrical Engineering and the
Committee, Stanford University, 2006.
www.gtia.co.in 24
Cross Site Scripting (XSS) Advanced Protection Concept

Ajit S. Ghodke
SIBACA, Lonavala, Pune University, Pune, India
Email: ghodke.a@gmail.com
Abstract: Cross-site scripting (XSS) is an attack against web applications in which scripting code is injected
into the output of an application that is then sent to a user’s web browser. In the browser, this scripting code is
executed and used to transfer sensitive data to a third party (i.e., the attacker) The trusted website is used to
store, transport, or deliver malicious content to the victim. The target is to trick the client browser to execute
malicious scripting commands. JavaScript, VBScript, ActiveX, HTML, or Flash. In today’s world more than
80% of the web applications are vulnerable to XSS threats. User friendly web applications are developed to
increase the customer base and hackers utilize the features provided by the web applications. Cross Site
Scripting (XSS) Attacks are currently the most popular security problems in modern web applications. This
paper was through light on one conceptual solution for all applications. This concept was used to protect all
types of web application from XSS attack and from different attack also.
Keywords: XSS vulnerability, protection mechanisms, new concept of protection.
1. Introduction at the victim (client) browser, with the “access

privileges” of the web site.
Cross-Site Scripting, commonly known as XSS,
is a type of attack that gathers important 2. Review of Literature
information about a user’s typically in the form
of a specially crafted hyperlink that will save the Web application security is a major concern in
users credentials. Cross-site scripting or XSS is today’s era for various organizations and
a web security vulnerability where the attacker banking sectors. Most of the organizations and
injects malicious client-side script into a web banking sectors who use internet to provide web
page. When a user visits a web page, the script based services that protect their sensitive data
code is downloaded and transparently run by the using firewalls and few access control
web browser. The malicious script inherits the mechanisms. However, still the organizations
user’s rights, authentication, and so on. The data are revealed by internet hackers by some
Cross Site Scripting is one of the most common means of purposefully designed java script.
application level attacks that hackers use to Therefore, the current security mechanisms are
sneak into web applications. One reason for the not sufficient to provide effective security to the
popularity of XSS vulnerabilities is that web databases. In such a scenario, it is necessary
developers of web-based applications often have to provide additional protection mechanisms for
little or no security background. That’s why securing the critical information that are
developer unable to protect the web-application retrieved by SQL queries or java script designed
from the attacker. carefully by hackers. This analysis has been
The goal of the XSS attack is to steal the client used for formulating new methods to prevent
cookies, or another sensitive information, which various types of attacks in web applications [1].
can identify the client with the web site. With Several works have been proposed and
the token of the legitimate user at hand, the implemented in the past by various researchers
attacker can proceed to act as the user in his/her for the protection of web applications from cross
interaction with the site specifically, site scripting attacks.
impersonate the user. For example, in one audit There are many works that have been proposed
conducted for a large company it was possible to in the literature for securing web applications
peek at the user’s credit card number and private from SQL injection and cross site scripting
information using a CSS attack. This was attacks. In most works, only static analysis
achieved by running malicious JavaScript code methods were proposed and implemented by
www.gtia.co.in 25
many researchers. Among the important static b) Stored (Persistent) - Forum, bulletin
analysis methods the static analysis method board, feedback form
proposed by Nenad Jovanovic et al., addressed c) DOM based XSS - PDF Adobe Reader ,
the problem of vulnerable web applications FLASH player
using static source code analysis. They used a a) Reflected (Non-Persistent): In this attack
flow sensitive analysis and context sensitive data the malicious code will not been stored on
flow analysis to discover vulnerable points in a web server. The attacker were simply send a
program [2]. malicious link to the victim and when a
In the past, many woks pertaining to dynamic victim click on that link the code attach with
analysis have been proposed and implemented that link will be executed and attack will be
by various researchers to secure web perform on victim machine as shown in Fig.
applications from SQL injection and cross site 1.
scripting attacks. Yao Wen Huang et al.,
proposed techniques for web application security
assessment mechanisms in order to identify poor
coding practices that render web applications
vulnerable to attacks such as SQL injection and
cross site scripting attacks. They used software
testing techniques (including dynamic analysis,
black box testing, fault injection, and behavior
monitoring), for testing the security in web
applications [3].
3.Objectives
The main objective of this research paper is to Fig. 1. Reflected (Non-Persistent)
protect web application from XSS attack in a b) Stored (Persistent): In this attack if any
effective manner with a simple technique to web server will vulnerable from XSS attack
develop an addons or extention for a web brower then attacker will stored the malicious code
who protect the client from this XSS attack. The to that server and when any victim will
main objective of this technique is to provide a request to that web server so in respond to
security to each and every web appllication by that request web server download the
not to concentrate on creation of secure code for webpage with the malicious code to the
web application but to protect it by an tools of victim computer then this malicious code
browser means addons or extension of web will be automatically executed by the victim
browser. This tools will be updated according to browser and attack will be performed on the
the discovery of new vulnerablity that’s why its victim machine [4, 5] as shown in Fig. 2.
an easy and secure way to protect any web
application from XSS attack.it’s an easy and
user friendly way to secure from XSS attack. By
this technique web developer and a cyber-
security team both perform their task in different
but co-ordinate and in effective manner. After
using this security concept web application will
be more secure as previous one.
4. Conceptual Study
4.1. Types of XSS
According to their attacking technique Cross Fig. 2. Stored (Persistent)
Site Scripting (XSS) is classified in three types c) DOM based XSS: As their name, Document
a) Reflected (Non-Persistent) - Link in Object Modeling (DOM) the attack with
other website or email document that means some site provides a
facility for upload and download of PDf or
www.gtia.co.in 26
any document file and if this document will guide line which guides a program to prevent
not filter before stored to database. Then their web application from each and every attack
attacker take an advantage for that and he can which was present in cyber world. This cheat
store any malicious code file in place of that sheet provide a guide to how a secure web
document file so if any user can click on that application built and which function and
file so this malicious code will be executed at techniques will be used to prevent their web
the client side by client browser without any application from different attacks.
command of that victim as shown in Fig. 3.
5. New Concept of XSS Security
In all the previous research all attention has been
given on the vulnerable code of the web
application that means where the loop hole is
present on the web application and how to
protect this loop hole or a vulnerability of the
web application. But the web application
developer will not been aware of each and every
security concept of the web application at every
level. Because now a day’s web attack will be
increase in very fast manner that’s why now web
security is not just a part of web application
development. These both are different and much
Fig. 3. DOM based XSS wide fields [6].
In the web application development process a
4.2. Prevention Techniques web application developer try to develop secure
This attack will be performed by attacker web application and after that a web security
because developers of web-based applications team were appointed to find vulnerability and
often have little or no security background. secure the web application from XSS attack. But
That’s why they were unable to create a secure the attacker will take an advantage of the gap
code. But due to the importance of security point between the development and the security of
of view there are some prevention techniques web application. That means after the
which help in securing web application and development of an web application an attacker
programmer use this to secure their web can scan or find the vulnerability in a web
application. Those prevention techniques are as application and then perform the attack on that
follows: vulnerability to harm company web server,
Validate Input client etc [7].
o Letters in a number field? Many different types of vulnerability will be
o 10 digits for 4 digit year field? find in web application so that attackers will use
o Whitelist (e.g. /[a-zA-Z0-9]{0,20}/) different techniques to attack on them
o Reject, don’t try and sanitize application that’s why it’s a big task for cyber
Use of function like:-There are many secrete to prevent all the application from all
function available in web development that attack [8,9].
language’s which help to secure the web My research say that cyber security team have to
application from XSS attack by filter the develop one conceptual solution for all web
input provided by the user. Some example of applications that means providing security for
PHP function are as follows: every XSS vulnerablity to each web application
o Stripslashes(); its a most diffcult task. So cyber security team
o mysql_real_escape_string(); have to develop an new addons or new extension
o str_replace(); for security purpose. That means develop of that
o addslashes(); tool which was added to the browser and this
tool will scan all the request and reponses from
Cross site scripting (XSS) prevention cheat
browser to server and vice versa and if it find
sheet:-It is a most effective way to secure any
any XSS attack malicious code and it denined
web application. XSS prevention cheat sheet is a
this request at that time and stop futher process.
list of tools and technique or we can say it is
www.gtia.co.in 27
This tool will be update as new type of context sensitive string evaluation. Recent
vulnerablity will be discover [10]. Advances in Intrusion Detection (RAID
2005), 2005.
6. Conclusion [6] Su, Z. and Wasserman, G. The essence of
command injections attacks in web
Cross site scripting is a most popular and a applications. 33rd ACM Symposium on
dangerous attack in today’s world. Cross site Principles of Programming Languages,
scripting (XSS) vulnerability is found in most of 2006.
the web application so it’s a big task for cyber [7] Livshits, B. and Erlingsson, U. Using web
security team to protect this all web application application construction frameworks to
of the world. So use of one conceptual solution protect against code injection attacks. 2007
for all web applications that protect web workshop on Programming languages and
application and clients from XSS attack. So by analysis for security, 2007.
creating a addons or extension tools of XSS [8] Christina Warrender, Stephanie Forrest, and
security for web browser will provide the Barak Pearlmutter. Detecting intrusions
security from XSS attack by scan each and every using system calls: Alternative data models.
request and reponses between client and a web In In IEEE Symposium on Security and
server that‘s why if there is any attacking Privacy, IEEE Computer Society, 1999.
malicious code is present there then this code [9] P. Vogt, F. Nentwich, N. Jovanovic, E.
will be remove by that addons and remaining Kirda, C. Kruegel, and G. Vigna. Cross-site
usefull code will be executed by the web scripting prevention with dynamic data
browser and the attack will not be perform on tainting and static analysis. In Proceeding of
client machine. the Network and Distributed System
This is the new concept which work as a middle Security Symposium (NDSS), San Diego,
man between web server and a client to protect CA, February 2007.
the client and the web application from the cross [10] C. Reis, J. Dunagan, H. Wang, O.
site scripting attack.this concept is different from Dubrovsky, and S. Esmeir. Browser-shield:
all the previous security research and it is most Vulnerability-driven _ltering of dynamic
effective and easy way to protect from cross site html. ACM Transactions on the Web, 1(3),
scripting attack. 2007.
References:
[1] Kirda, E., Kruegel, C., Vigna, G., and

Jovanovic, N. Noxes: A client-side solution
for mitigating cross-site scripting attacks.
21st ACM Symposium on Applied
Computing, 2006.
[2] Howard, M. and LeBlanc, D. Writing secure
code. Microsoft Press, Redmond, 2nd ed.,
2003.
[3] Cary, C., Wen, H. J., and Mahatanankoon,
P. A viable solution to enterprise
development and systems integration: a case
study of web services implementation.
International Journal of Management and
Enterprise Development, 1(2):164–175,
Inderscience, 2004.
[4] Hallaraker, O. and Vigna, G. Detecting
Malicious JavaScript Code in Mozilla. 10th
IEEE International Conference on
Engineering of Complex Computer Systems
(ICECCS’05), 2005.
[5] Pietraszeck, T. and Vanden-Berghe, C.
Defending against injection attacks through
www.gtia.co.in 28
A Review on Cloud Based Patient Monitoring

Nabeena Ameen
Assistant Professor, Department of Information Technology, B. S. Abdur Rahman University, Chennai, India
Email: nabjunaid@gmail.com
Abstract: Detection of bradycardia, and tachycardia is the common fatality in the current scenario of the health
care society. Thus, the proposal presents an innovative solution that addresses problems of integration, such as
medical staff from one institution being able to monitor patients located at another. Many hospitals are sharing
patient data on cloud as service. Hence in the present study, a new approach of Cloud based Wireless Remote
Elderly Patient Monitoring system during emergency was proposed as a prototype to monitor the vital health data
and needed drugs are prescribed. The heart rate, pulse, oxygen saturation, body temperature, lungs air volume and
blood glucose level are measured. The ECG arrhythmia classification scheme is composed of data acquisition,
feature extraction and normalization, feature reduction, and classification. A multi-class support vector machine
(SVM)-based classifier is proposed. A particle swarm optimization algorithm is propose to search for the best value
of the SVM parameters and upstream by looking for the best subset of features that feed the classifier. Different
types of arthymia are classified. Sensitivity, Specificity Accuracy is calculated.
In this paper, I present an extensive review of the significant researches associated with wireless patient monitoring
using wireless sensor networks based on network issues and cloud based patient monitoring using wireless sensor
networks and provided description about the emergency department.
Keywords: Wireless sensor network, Cloud computing, QOS, RPM, SVM.
1. Introduction down to the size of a grain of dust, although

functioning "motes" of genuine microscopic
A wireless sensor network (WSN) consists of dimensions have yet to be created. The topology
spatially distributed autonomous sensors to of the WSNs can vary from a simple network to an
monitor physical or environmental conditions such advanced multi-hop wireless mesh network. The
as temperature, sound, pressure and to main characteristics of a WSN include:
cooperatively pass their data through the network 1) Power consumption constraints for nodes
to a main location. using batteries or energy harvesting.
A wireless body area network is a category of 2) Ability to cope with node failures.
wireless sensor networks. At the patient's bedside, 3) Mobility of nodes.
there are sensor nodes which are loaded with 4) Communication failures.
software to collects, encode, and transmit data 5) Heterogeneity of nodes.
through wireless communication channels to be 6) Scalability to large scale of deployment.
stored. The WSN is built of "nodes" – from a few Sensor nodes are usually consist of a processing
to several hundreds or even thousands, where each unit with limited computational power and limited
node is connected to one sensor. Each such sensor memory, sensors or MEMS a communication
network node has typically several parts: a radio device (usually radio transceivers or alternatively
transceiver with an internal antenna or connection optical), and a power source usually in the form of
to an external antenna, a microcontroller, a battery. The base stations are one or more
electronic circuit for interfacing with the sensors components of the WSN with much more
and an energy source, usually a battery or an computational, energy and communication
embedded form of energy harvesting. A sensor resources. They act as a gateway between sensor
node might vary in size from that of a shoebox nodes and the end user as they typically forward
data from the WSN on to a server. Other special
www.gtia.co.in 29
components in routing based networks are routers, their internal IT infrastructure to a cloud service
designed to compute, calculate and distribute the provider.
routing tables. The National Institute of Standards and
Technology (NIST) has formalized the cloud
services being offered to consumers into three
different types : Infrastructure-as-a- Service
(IaaS), Platform-as-a-Service (PaaS), and Software
as- a-Service (SaaS). At the most basic level, IaaS
is the model in which a service provider makes a
set of virtualized computer components (e.g.,
virtual machines or networked storage) available
to consumers as a fully outsourced service that can
be used to build and run applications without
purchasing the actual computer components.
Fig. 1. The proposed architecture Each of these hardware components is visible to
the consumer. An example of an IaaS service is
In the Fig. 1. at the patient's bedside, there are Amazon which allows customers to dynamically
sensor nodes which are loaded with software to provision new virtual machines on the Amazon
collects, encode, and transmit data through cloud using configuration templates with their own
wireless communication channels to be stored. custom settings.
The Deploy Service acts like a broker between Moving to the next highest level, PaaS is the
local and remote services. It is responsible to model in which a service provider or vendor offers
receive collect data from sensors and to dispatch it to the client an entire runtime environment for an
to appropriate to storage service hosted on cloud. application (i.e., design, deployment, and test).
It also receives requests from content service to This would include the basic hardware in addition
retrieve data from the Cloud Service, whose to critical libraries that the consumer’s application
functionality is two folded: will rely on. An example of a PaaS service is
i. It is responsible to provide services to store Google AppEngine which allows developers to
collected data. make use of some of the same tools and libraries
ii. It provides a platform for development, testing used in Google Web applications to build their
and Deployment of applications needed by own custom applications by building on a
medical staff. Mobile and stationary devices common application platform. At the highest level,
interacts with applications using Content SaaS provides consumers with the use of on-
Service. demand software (e.g., business applications)
iii. This service acts like a “door” where medical running on the provider’s servers and the
staff devices can access all available associated data over the Internet.
information.
2.1 Hardware Setup
2. Cloud computing In the Fig. 2 three electrodes for sensing the
potential NI ELVIS for hardware platform is used
Cloud computing has become a compelling and it has an instrumentation amplifier and a band
business model for companies that own large data pass filter built on it.
centers to essentially rent out different layers of
their computing resources.
This phenomenon has emerged in large part
because many companies now internally rely both
on technologies that must be able to scale
dynamically and on large sets of computing
hardware for their processing. Therefore,
organizations have found that they can reduce IT
costs by merely outsourcing one or more types of Fig. 2. Hardware setup
www.gtia.co.in 30
emphasis has been on improving access to care

It is coupled it to the system and used the values and the last few years have seen a significant
for further exploitation and processing of the increase in the number of elderly patients being
signals as shown in Fig. 3. admitted for cataract surgery, hip or knee
replacements and interventional cardiac surgery.
In addition to traditional geratologists and
consultants in care of the elderly, many hospitals
have set up specialists ('modern matrons' or nurse
consultants) focusing on the needs of the elderly
whilst in hospital and on discharge.
Fig. 3. Signal Acquisition
3.2 Stroke Care
3. The Emergency Department Evidence suggests that stroke patients fare best
when admitted to specialised stroke units. The aim
It is one of the most information-intensive is to provide rapid access to diagnostic services,
environments in a hospital. Ever increasing patient care provided in stroke units led by specialized
volumes, an aging population, high occupancy physicians and multidisciplinary intervention to
levels, decreased availability of inpatient beds, enable early discharge, rehabilitation and
nursing shortages, and increased patient secondary prevention.
complexity all require ED staff members to
assimilate information from diverse sources to 3.3 Management of Falls
facilitate efficient information management and Falls are the leading cause of mortality in the
communication as shown in Fig. 4. elderly patients. All patients who have had a fall
should be offered a multi-factorial risk assessment
and multi-factorial interventions. The following
risks occur
(i) Multi-Factorial Risk Assessment

Older people who present for medical attention
because of a fall, or report recurrent falls in the
past year, or demonstrate abnormalities of gait
and/or balance should be offered a multi-factorial
fall risk assessment. This assessment should be
Fig. 4. Emergency department
performed by healthcare professionals with
appropriate skills and experience, normally in the
The medical applications can be of two types: setting of a specialist falls service. This assessment
wearable and implanted. Wearable devices are should be part of an individualized, multi-factorial
used on the body surface of a human or just at intervention Multi-factorial assessment may
close proximity of the user. The implantable include the following:
medical devices are those that are inserted inside a) Identification of the history of the falls.
human body. There are many other applications Assessment of gait, balance and mobility
too e.g. body position measurement and location and muscle weakness.
of the person, overall monitoring of ill patients in b) Assessment of osteoporosis risk.
hospitals and at homes. Body-area networks can c) Assessment of the older person's perceived
collect information about an individual's health, functional ability and fear relating to falling.
fitness, and energy expenditure. d) Assessment of visual impairment.
e) Assessment of cognitive impairment and
3.1. Specialist Care in Hospital neurological examination.
With the change in demography in the UK, a f) Assessment of urinary incontinence.
significant proportion of elderly people in g) Assessment of home hazards.
hospitals require secondary care needs. The
www.gtia.co.in 31
(ii) Multi-Factorial Interventions their data to a central processing node for data
All elderly people with recurrent falls or assessed fusion [8].
as being at increased risk of falling should be Agent Based Health Monitoring of Elderly People
considered for an individual. in Indoor Environments Using Wireless Sensor
Networks, [9] has proposed a design of a health
4. Extensive Review of Significant care monitoring system based on wireless sensor
Researches on Wireless Patient networks (WSN) which is capable of collecting,
retrieving, storing and analysing the vital signs of
Monitoring the patient. Vital signs such as body temperature,
blood pressure, pulse rate and respiratory rate are
The significant researches can be classified based monitored.
on network issues, cloud based approaches. Technological monitoring in elderly patients has
A new approach of wireless patient monitoring proposed WSN based home monitoring system for
system was proposed [1] using Zigbee to monitor elderly activity [10].
the patient. Real time monitoring of ECG signal using PIC and
A Prioritization Based Congestion Control web server has proposed the design of a wireless
Protocol for Healthcare Monitoring Application in sensor network to monitor patients suffering from
Wireless Sensor Networks [2] was proposed to chronic diseases in their homes without regularly
have congestion control and service prioritization going to health center [11].
protocol for real time monitoring of patients’ vital Wireless Sensor Networks for Ambient Assisted
signs using wireless biomedical sensor networks. Living was proposed to develop arrhythmia
Detecting mobility for monitoring patients with detection algorithm [12].
Parkinson’s disease at home using RSSI in a Patient Health Monitoring Based on ZigBee
Wireless Sensor Network was proposed [3] and Module was proposed to monitor the health of
wireless monitoring of the patients with patients continuously [13].
Parkinson's disease, for preventing potential falls A Robust Wearable Health Monitoring System
and injuries and recording the number of based on WSN was proposed for wearable health
occurrences of Freezing of Gait (FoG) over single monitoring system [14].
or multiple days. New Frontier to Monitor, Track and Assist Older
A new patient monitoring framework and Energy- Adults Suffering from Type-2 Diabetes was
aware Peering Routing Protocol (EPR) for Body proposed in wireless sensor network system within
Area Network [4] was proposed in Body Area the home environment that captures activity of
Networks (BANs). daily living [15].
Efficient location tracking system for patient
Monitoring focused on an efficient location
tracking system design for monitoring elder
5. Review on Significant Research on
patients in home environments [5]. Cloud Based Approach on Patient
Using ZigBee low-power wireless standard for Monitoring
monitoring patients’ signs presents Wireless Body
Area Network [6] was proposed. Highly Personalized Health Services using Cloud
An improved localization algorithm with Wireless and Sensors has proposed new medical services
heartbeat monitoring system for patients safety such as telemedicine, monitoring of chronic
was proposed in hospitals to maintain patient patients, personalized health services, creating
safety and ensure that patients do not leave to new services for dependants [16].
hospital without system can ease the workload of Novel Cloud and SOA Based Framework for E-
nurses, help locate patients, and monitor patient’s health Monitoring using Wireless biosensors, has
heartbeats [7]. proposed Chronic Diseases (CD) in governmental
Biometric Monitoring System for Multi Sensor and private healthcare systems [17]
Multi-modal Node has proposed a general network Applying Cloud computing technologies to
topology used for wireless body area networks is Gerontology and Geriatrics Health Care System
the star topology with the sensor nodes sending (GGHCS) has proposed overall healthcare IT
www.gtia.co.in 32
market with great potential impact on Gerontology Issue 4, pp 2605-2631, 2013.

and Geriatrics [18]. [3] Suryadip Chakraborty, Saibal K Ghosh,
Cloud computing for u-Health and Automotive Anagha Jamthe, Dharma P Agrawal, Detecting
Environment has proposed elderly and the chronic mobility for monitoring patients with
disease can have their own personal healthcare parkinson’s disease at home using RSSI in a
devices and they can monitor themselves easily wireless sensor networks’, Procedia Computer
[19]. Science 19, 956 – 961, 2013.
Cloud-Enabled Wireless Body Area Networks for [4] Zahoor A. Khan , Shyamala Sivakumar,
pervasive was proposed methodologies for William Phillips, Nauman Aslam, A new
transmitting vital sign data to the cloud by using patient monitoring framework and Energy-
energy-efficient routing ,cloud resource allocation, aware Peering Routing Protocol (EPR) for
semantic interactions, and data security [20] . Body Area Network communication,
Enabling Smart Personalized Healthcare: a Hybrid Springer J Ambient Intell Human Comput .
Mobile-Cloud Approach for ECG Telemonitoring [5] Daniel Rosner, Rzvan Ttroiu, Constantin Ilas,
was proposed personalized medical diagnosis and Efficient location tracking system for patient
treatment [21]. monitoring , 2013.
SeaMo+ is a Virtual Real-time Multimedia Service [6] Olivera Kotevska, Elena Vlahu-Gjorgievska,
Framework on Handhelds to enable Remote Real- and Saso Koceski, Using ZigBee low-power
time Patient Monitoring for Mobile Doctors was wireless standard for monitoring patients’signs
proposed the key requirements for enabling real- ,ICT Innovations 2013 Web Proceedings,
time remote healthcare service [22]. 2013.
[7] Chen-Yang Cheng, Tsuo-Hung Lan, Chin-
6. Conclusion Hong Chan,”An improved localization
algorithm with wireless heartbeat monitoring
Wireless Remote Elderly Patient Monitoring system for patient safety in psychiatric wards”,
during emergency helped in monitoring the health /journal of Engineering Applications of
data by the doctors. These parameters like heart Artificial Intelligence 26, 905–912, 2013.
rate, pulse, oxygen saturation, body temperature, [8] J. Mohana, V. Thulasibai, Biometric
lungs air volume and blood glucose level are Monitoring System for Multi Sensor Multi-
measured. In addition, a two-way audio/video modal Node Architecture, Volume III, Special
cares greatly valuable for doctors to monitor the Issue, August 2013 ,International Journal of
patients. In this paper, I have presented a Computer Applications in Engineering
comprehensive survey of the significant and recent Sciences, 2013.
researches that are concerned with wireless patient [9] V. Vaidehi, M.Vardhini, H.Yogeshwaran, G.
monitoring using cloud and Patient monitoring Inbasagar, R. Bhargavi, C. Sweetlin
using wireless sensor networks Along with this, an Hemalatha, Agent Based Health Monitoring
introduction to Wireless Sensor Networks, Cloud of Elderly People in Indoor Environments
computing and the Emergency department tasks Using Wireless Sensor Networks, Procedia
are provided. Computer Science 19, 64–71, 2013.
[10] S.D. Grigorescu, O.M. Ghita, C.K. Banica, S.
References: Potlog, A.M. Paraschiv, Health Monitoring
Solution Using Dedicated ZigBee Sensor
[1] N. M. Z. Hashim, M. S. Sizali ,Wireless Network, The 8th International Symposium
patient monitoring system, International On Advanced Topics In Electrical
Journal of Science and Research (IJSR), India Engineering Bucharest, Romania.
online Vol. 2 ,Issue 8. [11] N. Beaulah, A L. Kumarappan, Technology
[2] Mohammad Hossein Yaghmaee, Donald Improvements In Elderly Monitoring System,
Adjeroh, A Prioritization Based Congestion International Journal of Advanced Trends in
Control Protocol for Healthcare Monitoring Computer Science and Engineering, Vol.2 ,
Application in Wireless Sensor Networks” No.2, Pages : 29-32, 2013.
Wireless Personal Communications, Vol. 72,
www.gtia.co.in 33
[12] Saurabh Prakash, V.Venkatesh ,”Real time [17] Diego Gachet Páez, Fernando Aparicio,
monitoring of ECG signal using PIC and web Manuel de Buenaga, Margarita Rubio, Highly
server “, International Journal of Engineering Personalized Health Services using Cloud and
and Technology (IJET). Sensors ,IEEE, 2013.
[13] Raúl Aquino-Santos Diego Martinez-Castro [18] Abdelghani Benharref, Novel Cloud and
, Arthur Edwards-Block and Andrés Felipe SOA Based Framework for E- health
Murillo-Piedrahita , Wireless Sensor Monitoring Using Wireless Biosensors, IEEE
Networks for Ambient Assisted Living , 2013.
Sensors, 13, 16384-16405, 2013. [19] Tse-Chuan Hsu, Chih Hung Chang, Applying
[14] A. Deepa ,P. Nirmal Kumar, Patient Health Cloud computing technologies to
Monitoring Based on ZigBee Module, IEEE , Gerontology and Geriatrics Health Care
2013 . System(GGHCS), IEEE, 2013.
[15] Young Hwan Kim, Kuk-Jin Jang, Seung-chul [20] Minwoojung, Kabsu Han, Cloud Computing
Lee, and HeeYong Youn, A Robust Wearable for u-Health and Automotive, IEEE, 2013.
Health Monitoring System based on WSN, the [21] Jiafu Wan and Caifeng Zou, Cloud-Enabled
10th Annual IEEE- CCNC Smart Spaces and Wireless Bod Area Networks for Pervasive
Sensor Networks. Healthcare”, IEEE, 2013.
[16] Samir Chatterjee, Kaushik Dutta, Harry (Qi) [22] Xiaoliang Wang, Qiong Gui, Bingwei Liu,
Xie Jongbok, A New Frontier to Monitor, Zhanpeng Jin, and Yu Chen, Enabling Smart
Track and Assist Older Adults Suffering from Personalized Healthcare: a Hybrid Mobile-
Type- 2 Diabetes, 1530-1605, IEEE, 2012. Cloud Approach for ECG Telemonitoring.
www.gtia.co.in 34
Palm Recognition Using LBP and SVM

Preeti Sehgal
M.Tech CSE, Rawal Institute of Engineering & Technology, Faridabad, Haryana, 121004, India
Email: preeti.sehgal694@gmail.com
Abstract: Biometric system is becoming more popular and increasingly important, since they provide more
reliable and accurate means of identification. These days biometric systems are used almost everywhere for
identification. In Japan, banks have used the palm vein authentication technology for customer identification
since July 2004. In addition, Fujitsu has integrated the technology into the access control of electronic door lock
systems. Their system has a false acceptance rate of less than 0.00008% and a false rejection rate of 0.01%. In
this paper, biometric user recognition system based on Palm Print has been proposed. The proposed system
attempts to reduce the complexity of Palm Vein Authentication system. The proposed system is implemented
using MATLAB and LBP method and uses SVM technique which reduces the complexity there by a increase
the performance. The system has an average 97.5% recognition rate. This Research work is an attempt to
propose such a system that can bring out a reduced complexity for Palm vein recognition. Palm vein recognition
is easily implemented using a feature extraction algorithm (LBP) and SVM for classification which reduces
complexity thereby increases performance. The performance analysis is done using CASIA database.
Keywords: Authentication System, Acquisition, Median Filter, Segmentation, Clustering, SVM, LBP,
Recognition
1. Introduction (VPTA). Palm print recognition is one of the

most important approaches of biometric
Biometrics means to identify humans by their technology and is one of the most reliable and
characteristics or by features. Biometrics is used successful human being identification and
as a form of identification and authentication authentication methods [4].
which is also used to identify individuals in In this paper, the focus is on feature extraction as
groups that are under surveillance [1]. The well as classification part of recognition. Feature
distinctive, measurable characteristics used to tag Extraction is the process of using the most
and describe each unique personality are important information of the cropped palm vein
Biometric identifiers which can be categorized as images for classification purpose. The three main
physiological versus behavioral. categories of feature extraction are used in
Physiological traits are related to the shape of the literature. They are, Line based approach,
body. Behavioral characteristics are related to the Appearance based and texture based approach. I
pattern of behavior of a person, including gait, have used texture based approach for extracting
and voice. The latter part of the category, i.e., the features. I have used Linear Binary Pattern
behavioral characteristics is also termed as for that purpose.
behaviometrics by some researchers. This paper is divided into five sections. First
There are numerous distinguishing traits used for section explains motivation for the research. The
personal identification, but this research will second section follows Literature Review. Third
focus on using palm prints to more correctly and section is helpful to understand the intended
efficiently identify different personnel through work or the proposed system and
classification at a low cost. Palm print is implementation. Fourth section shows the
preferred since it is distinctive, easily captured by experimental results. At last fifth section
low resolution devices when compared to other concludes the paper and followed by references.
methods such as fingerprint or iris [2]. Various
algorithms for authentication had already been 1.1 Motivation for Research
proposed by researchers [3]. These are: Vascular Why we need Palm vein technology?
Pattern Pointer Algorithm (VPPA), Vascular Palm print is preferred compared to other
Pattern G-B Conversion Algorithm (VPGBCA), methods such as fingerprint or iris because it is
and Vascular Pattern Thinner Algorithm distinctive, easily captured by low resolution
www.gtia.co.in 35
devices. Iris input devices are expensive and the image with a training image the line matching
method is intrusive as people might fear of technique has been used which is based on pixel-
adverse effects on their eyes. to-area algorithm.
Palm print is therefore suitable for everyone and Online Palm print Identification system (Zhang,
it is also non-intrusive as it does not require any Kong, You and Wong [9]) takes online palm
personal information of the user. prints, and uses low resolution images. Low pass
• Vein patterns are unique to individuals and filter and boundary tracking algorithm is used in
contain detailed characteristics for pre-processing phase. Circular Gabor filter used
formulation of algorithm template. Contact for feature extraction and 2-D Gabor phase
less authentication is hygienic and non- coding is used for feature representation. A
invasive, thus promoting high-level of user normalized hamming distance is applied for
acceptance. matching.
• Extremely difficult to forge, thereby enabling A dynamic selection scheme by introducing
a high level of security. global texture feature measurement and the
• Palm print is hardly affected by age and detection of local interesting points had also been
accessories in comparison to face proposed by J. You, W. Kong, D. Zhang, and K.
recognition. Cheung [10].
• Palm print images contain more information An effective indexing and searching scheme for
and needs only low resolution image an image database to facilitate fast retrieval when
capturing devices when compared with the size of a palmprint database is large search
fingerprint recognition. [11].
• Compared with iris recognition the palm Palmprint Authentication Using Fusion of
print images can be captured without Wavelet Based Representations had been
intrusiveness as people might fear of adverse proposed by Prasad, Govindan and Sathidevi[12].
effects on their eyes and cost effective. Features extracted are Texture feature and line
Hence it has become an important and rapidly features. In proposed system pre-processing
developing biometrics technology over the last includes low pass filtering, segmentation,
decade. location of invariant points, and alignment and
extraction of ROI.
Bhattacharyya et al. [13] proposed three different
2. Literature Review algorithms for processing Palm Vein Pattern
Many approaches have been proposed for palm Image of an individual. These three different
print matching in the recent years. processes are: Vascular Pattern Marker
An indexing technique that can employ two Algorithm (VPMA); Vascular Pattern Extractor
biometric matcher was proposed by A. Algorithm (VPEA); and Vascular Pattern
Gyaourova and A. Ross [5]. Thinning Algorithm (VPTA).
A method for palm print matching that deals with Verification system which utilizes these hand
skin distortion, discrimination power and geometry features for user authentication was
computational complications was given by Jifeng proposed by Jobin J [14].
Dai et. al. [6]. They used the method of cascade Mohit Soni et. al. presents a technique which
filtering in matching process. The matching extracts the forking from the skeleton image by
algorithms proposed in this paper deals with examining the local neighborhood of each ridge
noise and distortion efficiently. pixel using a 3X3 window. It can be seen from
Palm print matching using high resolution the preprocessing image that an ROI contains
images and uses multiple features for matching. some thinned lines/ridges. These ridges
[7]. Here multiple features such as density, representing vein patterns can be used to extract
orientation, minutiae, and principle lines were features [15].
taken to significantly improve the matching Sunita Aeri et. al. proposed a new approach for
performance. biometric authentication system using infrared
Palm print verification based on robust line thermal hand vein patterns. The proposed work
orientation code was proposed by Jiaa, Huanga presents a Euclidean distance based vein’s
and Zhang [8]. Modified finite Radon transform pattern based biometric authentication that can be
has been used for feature extraction, which used for matching the biometric identity of
extracts orientation feature. For matching of test person under scanner [16].
www.gtia.co.in 36
The palm print recognition system includes

3. Proposed System preprocessing followed by Image segmentation.
After Segmentation, hand features are extracted.
Vein recognition is a type of biometrics that can The palm print is then accepted or rejected on the
be used to identify individuals based on the vein basis of classifying the image and matching it
patterns in the human palm [17]. In this paper, I with image stored in database. All these
have proposed a method for recognition of palm processes have been studied and reviewed. Each
using LBP and SVM. step is further described below in detail.
Palm Recognition system contains following
modules: 3.2 Image Preprocessing
• Image acquisition: capturing input image In this stage, the input hand image is
• Image Preprocessing: Filtering of image preprocessed to adjust the contrast and
• Segmentation: clustering image brightness. Then, the preprocessed image is
• Feature Extraction Using Linear Binary filtered using median filter to remove ‘salt and
Pattern pepper’ noises so that the blurring of edges can
be reduced. After filtering, the image is
• Classification and Matching with
segmented.
databaseusing Support Vector Machine
Median filtering is a common step in image
(SVM).
processing. Performing a high degree of noise
In Fig. 1, the proposed system is shown at block
reduction in an image is necessary in image
level.
preprocessing before performing higher-level
Image acquisition processing steps. It is particularly useful to
reduce speckle noise and ‘salt and pepper’ noise
and is also useful for removing blurring of edges
because of its edge preserving nature. The
median filter is a non-linear digital filtering
Image preprocessing technique, often used to remove noise from
images or other signals.
Image segmentation is the first step in most
image analysis and pattern recognition systems
Feature Extraction and one of the most difficult tasks, which
determines the quality of the final result of
analysis.
It is the process of dividing an image into
Classification and different regions such that each region is
Matching homogeneous. The segmentation problem can be
informally described as the task of partitioning an
Fig. 1. Proposed System image into homogeneous regions. By using a
. clustering algorithm, the pixels of an image can
3.1 Image Acquisition be labeled to form homogeneous regions.
There are various ways to capture palm print Different clustering algorithms are commonly
image. Some of them are CCD-based scanners, used in image segmentation algorithms. But I
digital scanners, video camera and tripod to have used K-means clustering algorithm.
collect palm print images. A CCD-based scanner
captures high resolution images and aligns palms 3.2.1 K-Mean Clustering Based Segmentation
accurately because it has pegs for guiding the Clustering is a classification technique.
placement of hand. Digital scanners produces Clustering is a way to form natural groupings or
low quality image and requires large time for clusters of patterns. Unlike Classification where
scanning, therefore it cannot be used for real time the task is to learn to assign instances to
applications. Digital and video cameras can also predefined classes, in clustering no predefined
captured palm images but can cause recognition classification is required. The task is to learn a
problems. classification from the data. Clustering methods
Initially the hand images are obtained from the are some of the earliest data segmentation
user group and stored in the database. techniques to be developed.
www.gtia.co.in 37
K-mean Clustering Algorithm • The examined window is divided into

K-means clustering is an algorithm to classify or cells (e.g. 16x16 pixels for each cell).
to group objects based on attributes/features into • Compare the pixel to each of its 8
K number of group. K is positive integer neighbors (on its left-top, left-middle,
number. The grouping is done by minimizing the left-bottom, right-top, etc.) for each pixel
sum of squares of distances between data and the in a cell. Follow the pixels along a
corresponding cluster centroid. circle, i.e. clockwise or counter-
The algorithm consists of three steps discussed as clockwise.
follows:
Take any random object as initial centroid, and
then repeat the three steps mentioned below until
converge.
• Determine centroid coordinates.
• Determine distance of each object to the
centroid.
• Group the object based on minimum
distance. Fig. 2. Obtaining a Binary Level for a Pixel
3.3 Feature Extraction • Write “1” where the center pixel's value
After segmentation stage, it is necessary to is greater than the neighbor's value.
extract the features of the Palm before matching. Otherwise, “0” is to be written which
Various algorithms have been used in literature gives an 8-digit binary number.
for Feature extraction such as Gabor filter, SIFT,
• Histogram is computed, over the cell, of
PCA. I have used Linear Binary Pattern (LBP)
the frequency of each "number"
which is the latest method and gives us the better
occurring ((i.e., each combination of
results and accuracy. It is a simple and fast for
which pixels are smaller and which are
implementation.
greater than the center).
The feature extraction module will output the
• Normalize the histogram optionally.
biometric template, which will be used in the
matching stage. The feature extraction technique • Now, Histogram of all cells are
used in the developed system is the Linear concatenated which gives the feature
Binary pattern (LBP) which is the latest method vector for the window.
and gives us the better results and accuracy.LBP The feature vector can now be processed using
is texture based approach. the Support vector machine (SVM) or confusion
A useful extension to the original operator is the matrix to classify images.
so-called uniform patterns, which can be used to Given a central pixel in the image, a pattern code
reduce the length of the feature vector and is computed by comparing it with its neighbors
implement a simple rotation invariant descriptor. [18] as given in equation 1 below:

This idea is motivated by the fact that some LBPP,R =∑ − 2
----(1)
binary patterns occur more commonly in texture
1, ≥ 0
images than others. A local binary pattern is S(x) =
called uniform if the binary pattern contains at 0, < 0
most two 0-1 or 1-0 transitions. For example, Where gc is the gray value of the central pixel, gp
00010000(2 transitions) is a uniform pattern, is the value of its neighbors, P is the total number
01010100(6 transitions) is not. In the of involved neighbors and R is the radius of the
computation of the LBP histogram, the histogram neighborhood.
has a separate bin for every uniform pattern, and
all non-uniform patterns are assigned to a single 3.4 Classification and Matching
bin. Using uniform patterns, the length of the Support Vector Machine (SVM) is mostly used
feature vector for a 3x3 window reduces from for splitting clustered data into two classes. We
256 to 59. It is shown in Fig. 2. have two sets training set and testing set which is
The LBP feature vector, in its simplest form, is used for matching. An SVM classifies data by
created as discussed below: finding the best hyper plane that separates all
data points of one class from those of the other
class. The best hyper plane for an SVM means
www.gtia.co.in 38
the one with the largest margin between the two preprocessed using median filter and
classes. Margin means the maximal width of the segmentation. Fig. 4 shows the filtered and
slab parallel to the hyper plane that has no segmented images hence completing
interior data points. Fig. 3 below shows the preprocessing stage.
hyper plane.
Fig. 4. Filtered and segmented images of palm
Fig. 5 shows the user interface showing the

user is presented with five options as given
below:
Fig. 3. Hyper Plane
As we can see from Fig. 3, H3 does not separate

the two classes while H1 separate the two class
with a small margin, only H2 gives a maximum
margin between two classes, therefore it’s the
right hyper plane used by support vector
machine. However, instead define a function for
the hyper plane itself; we define the margin in
between the two classes.
The points nearest to the separating hyper plane
are called Support Vectors. Only they determine
the position of the hyper plane. All other points
have no influence! Fig. 5. User Interface
Mathematically: the weighted sum of the Support
Vectors is the normal vector of the hyper plane. Create Database: When user click on create
Now after classification, Matching is to be done. database, it will create the database of all
We can check the performance of the system images which was stored earlier in previous
using confusion matrix. stages.
A confusion matrix contains information about Load Test Image: When user clicks on this
actual and predicted classifications done by a button, it will take the image from user, it
classification system. Performance of systems is depend on the user whether to take image
commonly evaluated using the data in the matrix. from database or any raw image is taken.
A confusion matrix, also known as a contingency It will automatically perform the tasks of
table or an error matrix, is a specific table layout preprocessing; if we take any colored image
that allows visualization of the performance of an it will show the image as grayscale.
algorithm, typically a supervised learning one. Clear All: It will clear the images as well as
Each column of the matrix represents the blank the text box shown.
instances in a predicted class, while each row Matching: This button displays the image
represents the instances in an actual class. after matching the loaded image from the
This is shown in the next section. database. If the image exists in the database
it will generate response that “image belongs
to user no..” but if image doesn’t exist in
4. Implementation and Results database it will shows “User doesn’t exist in
database”.
I have implemented and tested the system
Overall Recognition: This button will display
performance using CASIA database. In this,
the overall recognition rate of the system
image of hand is acquired from each user and
which is 95.7% as shown in figure 8. It will
used for acquisition process. After that, image is
also generate confusion matrix showing the
www.gtia.co.in 39
correct detected and incorrect detected diagonal values are correct detected and
images. whichever comes under other than diagonal are
The following Fig. 6 and 7 shows the incorrect.
matching stage. The very last two diagonal elements(yellow
color and green color) shows the overall
recognition rate, and the value left side of
that(green color) shows that 2 are correctly
recognized and one is false detected.
We can calculate overall recognition rate as
follows.
Firstly add total number of test samples (all
values in matrix diagonally and off-diagonally),
secondly add correctly recognized values
(diagonal elements) and then add off-diagonal
elements (false detected).
Fig. 6. Matching showing correct detection Recognition rate=correctly recognized/total
number * 100
It gives the overall recognition rate which is
97.5%. Table 1 shows the average percentage of
positive recognition using SVM
Table 1. Recognition Performance using SVM
Fig. 7. Matching showing incorrect detection
The system is evaluated for the performance

using 120 samples (users). Out of 120 samples,
115 are correctly recognized and 5 samples are
misidentified. So identification accuracy is
95.7% which is shown using confusion matrix in
Fig 8.
5. Conclusion
Biometrics is being used all over the globe and is
undergoing constant development. The Palm
Print has proved to be a reliable biometric.
Several existing methods have been reviewed for
palm print recognition.
Here in this Paper I proposed a hybrid LBP
method for extracting features which reduces the
complexity of the system thereby increasing
efficiency and is fast to compute. The input is
Fig. 8. Confusion Matrix full Palm Print and the output shows the
recognition performance and confusion matrix.
The confusion matrix actually tells about the Using SVM, I have classified images and match
target and the actual result got. For example, on x with database. Hence it shows increase in
axes we have targeted and on y axes we have performance.
actual then if the samples which come under
www.gtia.co.in 40
References: [10] J. You, W. Kong, D. Zhang, and K. Cheung,

On hierarchical palmprint coding with
[1] Allen George, G. Karthick, Palmprint multiple features for personal identification
Recognition Using Ridge Features, in large databases, IEEE Trans. Circuits
International Journal of Advanced Research Syst. Video Technol., vol. 14, no. 2,pp. 234–
in Computer and Communication 243, Feb. 2009.
Engineering Vol. 2, Issue 11, November [11] W. Li, J. You, and D. Zhang, Texture-based
2013. palmprint retrieval using a layered search
[2] Apurva Puranik, Rahul Patil, Sharayu Kirad, scheme for personal identification, IEEE
Varsha Patil, Milind Rane, A Survey of Palm Trans. Multimedia, vol. 7, no. 5, pp. 891–
Print Recognition, International Journal of 898, Oct. 2009
Innovative Research and Studies Vol.2, Issue [12] S. M. Prasad, V. K. Govindan , P. S.
4, April 2013. Sathidevi, Palmprint Authentication Using
[3] Papri Ghosh, Ritam Dutta, A new approach Fusion of Wavelet Based Representations,
towards Biometric Authentication System in IEEE, pp. 978-1-4244-5612-3, 2009.
Palm Vein Domain, IJAITI 2012 Volume 1 [13] Mona A. Ahmed et al, Analysis of Palm
Number 2 (Mar/Apr 2012) ISSN: 2277– Vein Pattern Recognition Algorithms and
1891. Systems International Journal of Bio-Medical
[4] [4] Ramandeep Kaur, Meenakshi Garg, A Informatics and e-Health, 1(1), June – July
Review On Palmprint Recognition Using 2013, 10-14.
Features From Principle Lines, International [14] Jobin J., Jiji Joseph, Sandhya Y.A, Soni P.
Journal Of Engineering And Computer Saji, Deepa P.L., Palm Biometrics
Science ISSN:2319-7242 Volume 4 Issue 1 Recognition and Verification System,
January 2015 Page No. 10112-10115. International Journal of Advanced Research
[5] Gyaourova and A. Ross, Index codes for in Electrical, Electronics and Instrumentation
multibiometric pattern retrieval, IEEE Trans. Engineering Vol. 1, Issue 2, August 2012
Inf. Forensics Security, vol. 7, no. 2, pp. [15] Mohit soni, Sandesh Gupta, M.S. Rao,
518–529, Apr. 2012. Phalguni Gupta, “A New Vein Pattern-based
[6] A.K Jain and J. Feng, Latent Palmprint Verification System”, (IJCSIS) International
Matching, IEEE Trans. Pattern Analysis and Journal of Computer Science and
Machine Intelligence, Vol.33, No.1, pg.88- Information Security, Vol. 8, No. 1, 2010.
100, Jan 2011. [16] Sunita Aeri and Sukhvinder Kaur,” Vein
[7] J.Dai and J.Zhou, Multi-feature-based High Patterns as Bio-Metric Identifier using
Resolution palm recognition, IEEE Trans. Euclidean Distance”, International Journal of
Pattern Analysis and Machine Intelligence, Computer Applications (0975 – 8887)
Vol.33, No.5, pg.945-957, May 2011. Volume 89 – No 20, March 2014.
[8] Huang, W. Jia, D. Zhang, Palmprint [17] Nirmal et al., Palm vein authentication,
verification based on robust line orientation International Journal of Advanced Research
code, Pattern Recognition, Science Direct, in Computer Science and Software
pp. 1504 – 1513, 2008. Engineering 4(6), June - 2014, pp. 41-44.
[9] D. Zhang, W. K. Kong, J. You, M. Wong, [18] Udhayakumar. M, Gayathri. S, Mary jeya
Online Palmprint Identification, IEEE priya. V, Thiyagu. M, Palmprint Recognition
Transaction on Pattern Analysis and Machine by using Modified local binary pattern,
Intelligence, Vol.25, No. 9, pp. 0162-8828, International Journal of Innovative Research
Sept 2003. in Science, Engineering and Technology,
Vol. 2, Issue 9, September 2013.
www.gtia.co.in 41
Dear Researchers,
This is to inform you that Call for Paper is open for 2015 in
1. INTERNATIONAL JOURNAL OF INFORMATION TECHNOLOGY AND SYSTEMS

2. INTERNATIONAL JOURNAL OF PHYSICAL, CHEMICAL AND
MATHEMATICAL SCIENCES
3. INTERNATIONAL JOURNAL OF PHARMACY, BIOLOGY AND MEDICAL
SCIENCES
4. INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING AND
MANAGEMENT
It is our honor to invite you to submit your research paper for the same.
Name of Journal: INTERNATIONAL JOURNAL OF INFORMATION TECHNOLOGY AND SYSTEMS (IJITS)

Issue: Vol. 4, Issue 2. July-Dec., 2015
ISSN: 2277-9825
Period of Processing: 15 to 20 days
Details are given on our website: www.gtia.co.in
Email: editorijits@gmail.com
Name of Journal: INTERNATIONAL JOURNAL OF PHYSICAL, CHEMICAL AND MATHEMATICAL SCIENCES (IJPCMS)
ISSN: 2278–683X
Email: editorijpcms@gmail.com
Name of Journal: INTERNATIONAL JOURNAL OF PHARMACY, BIOLOGY AND MEDICAL SCIENCES (IJPBMS)
ISSN: 2319 – 3026
Email: edditorijpbms@gmail.com
Name of Journal: INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING AND MANAGEMENT (IJIEM)

ISSN: 2319 – 3344
Period of Processing: 10 to12 days
Email: editorijiem@gmail.com
All the above journals are printed copy, open access, peer reviewed international journals with a primary
objective to provide the academic community and industry for the submission of original research and
applications. Submitted papers should not have been previously published nor be currently under
consideration for publication elsewhere. All papers are refereed through a double blind process.
No Publication Charges. The Printed Copy of the Journal will be available

in Jan 2016 but the Papers will be online as soon as these are accepted for
publication.
A prompt response will reflect your interest to research world. Contact us for any queries.
Published By
Global Technocrats and Intellectual’s Association

Final Copy IJITS

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Final Copy IJITS

Uploaded by

Copyright:

Available Formats

Vol. 4, No.

1, Jan-June 2015 ISSN: 2277–9825

Global Impact Factor: 0.412

International Journal of Information Technology and Systems

Dr. Satinder Bal Gupta,

EDITORIAL BOARD MEMBERS

Dr. Hamid Saremi Dr. Santosh K. Pandey

Dr. K.SURESH Dr. Dil Muhammad Akbar Hussain

Global Technocrats & Intellectual's Association (GTIA)

Volume 4 Number 1 Jan-June 2015

1. Log mining in Query recommendation ……………………………………………….. 1-8

2. Multipath Routing in Autonomous Mobile Mesh Networks …………………………9-12

3. Comparative Study of Enhanced Interior Gateway Routing Protocol (EIGRP)

4. Review of SiGe Double Gate N-MOSFET ………………………………………… 19-24

5. Cross Site Scripting (XSS) Advanced Protection Concept ………………………… 25-28

6. A Review on Cloud Based Patient Monitoring …………………………………….. 29-34

7. Palm Recognition Using LBP and SVM …………………………………………… 35-41

Log mining in Query recommendation

Keywords: Log mining, KDD, Web mining, Query recommendation

Accepted On: 28.01.2015

1. Introduction insufficient or imprecise to retrieve those. It

Fig. 2. Purposed Architecture

Table 1. Example of query log

Fig. 3. Similarity (KW:keyword)

The approach taken by this module is based on

3.2.1 Similarity based on Query Keywords

q1= data mining examined against all other queries whether

3.3 Query Clustering Tool Ck = {x};

} // End For Algorithm for Generalized Sequential Pattern

[5] D. Beeferman and A. Berger. Agglomerative

[1] Sana Siddiqui, Imran Qadri, IJARCSSE ,

Multipath Routing in Autonomous Mobile Mesh Networks

Keywords: AMMNET, Routing, Multipath, NS2

Accepted On: 11.02.2015

1. Introduction organized in to following sections. Next section

Autonomous mobile mesh network is the

4. Multipath Routing in AMMNET

Fig. 5. Packet Delivery ratio Vs. Simulation Time

Fig. 4. Two alternative paths to reach destination

Even though multiple paths are not available

the number of nodes can be increased and the

[1] Wei-Liang Shen, Chung-Shiuan Chen,Kate

Comparative Study of Enhanced Interior Gateway Routing

*Department of Computer Science, Nigerian Defence Academy, Kaduna, Nigeria.

Accepted On: 26.03.2015

1. Introduction EIGRP is a Cisco propriety distance-vector

The disadvantage of EIGRP is that Routers from

The disadvantages of OSPF are:

3. EIGRP and OSPF Algorithms

being added or deleted, Dijkstra algorithm is

Fig. 5: Network model for EIGRP (Cisco, 2002)

Fig. 3: Shortest path tree (Cisco, 2002) 4.2 OSPF Scenario

menu. Finally, time duration to run the References:

Review of SiGe Double Gate N-MOSFET

1. Introduction Although the SOI technology is highly attractive

operation, it is found that thinner gate oxides are 1. Body Effect

Under this concept, channel is under the control

References: [11] T.Krishnamohan C.Jungemann, D. Kim, E.

MOSFETs & Undoped Symmetric &

Cross Site Scripting (XSS) Advanced Protection Concept

Keywords: XSS vulnerability, protection mechanisms, new concept of protection.

Accepted On: 11.05.2015

1. Introduction at the victim (client) browser, with the “access

[1] Kirda, E., Kruegel, C., Vigna, G., and