· pdf file2 ise dr. naveen n.c. dr. sunanda dixit ... upgrade and tra nsform the methods and...

��

� � � �

DAYA

© 2015-201

��

� � � � ��

��

��Peer reviewed Bi-Annual Journal

Vol 1: Issue 1

AYANANDA SAGAR COLLEGE OF ENGIN

BENGALURU-560078

��

2016, DSI publications

��

��

��

ournal

NGINEERING

ISSN (OnlinOnline):2455 6068

I

International Journal of Electrical Sciences & Engineering (IJESE) Online ISSN: 2455 6068

CHIEF PATRONS

Dr. D. Hemachandra Sagar, Chairman, DSI Dr. D. Premachandra Sagar, Vice-Chairman, DSI

Sri. Galiswamy, Secretary, MGVP trust, DSI

PATRONS Dr. S.C. Sharma, Provost, DSU & Director, DSCE/DSATM

Dr .C.P.S. Prakash, Principal, DSCE

EDITORIAL BOARD

SL.No DEPARTMENT HEAD MEMBER 1 CSE Dr. D.R. Ramesh Babu Dr. S.Venkatesan 2 ISE Dr. Naveen N.C. Dr. Sunanda Dixit 3 E& IE Dr.V.G. Sangam Prof. Rajashekar J.S. 4 TC Dr. A.R. Aswath Dr. H.C. Srinivasaiah 5 EC Dr. A.R. Aswath Prof. Krishnananda, Dr. T.S. Rukmini 6 ML Prof. M.S. Nagananda Dr. Sairam Geethanath 7 EEE Dr. Shanmukha Sundar Dr. P. Usha 8 MCA Dr. Poornima Nataraja Prof. Samitha Khaiyum

INTERNATIONAL EDITORIAL MEMBERS

Dr.Chokalingam Aravind Vaithilingam, Taylor’s University Malaysia 475000.

Dr. Sattar J Aboud, Ph.D, Professor Department: Information Technology Iraqi Council of Representatives, Information Technology Advisor, Palace House, Baghdad-Iraq.

Prof. Ramesh Bansal, University of Pretoria, Pretoria 002, South Africa

Dr. Chow Chee Onn, Ph.D, Professor Department of Electrical Engineering Faculty of Engineering, University of Malaya, 50603, Kula Lumpur, Malaysia

Professor Venkatesh Saligram, Electrical & Computer Engineering, Boston University,

Dr. Xavier Fernando, Ph.D., Professor & Multimedia stream Coordinator, Department : Electrical and Computer Engineering, Ryerson University, 350 Victoria Street, Toronto, ON M5B2K3, Canada

Dr. Kashinath Basu, Senior Lecturer. Department of Computing and Communication Technologies, Oxford Brookes University , UK

Dr. Seyed Mohamed Buhari, Ph.D, Assistant Professor, Department of Information Technology, Faculty of Computing and Information Technology, King Abdulaziz University, PO Box 80221, Jeddah – 21589, Saudi Arabia.

Dr. Suresh Shanmugasundaram Professor & Dean, Faculty of Engineering & Applied Sciences Botho University - Gaborone

Dr. Amri B AB Rahman, Senior Lecturer Science Computer and Mathematics University Teknology Mara UITM Machang Kelantan Malaysia

I | Dayananda Sagar College of Engineering, Bengaluru-78

II | Dayananda Sagar College of Engineering, Bengaluru-78

Editorial

International Journal of Electrical Sciences and Engineering (IJESE)

On behalf of the International Journal of Electrical Sciences and Engineering, I am happy to

present this focused journal to the esteemed researchers and readers worldwide.

The avenues of research in Electrical Sciences and Engineering remain to multifold which can

promote the young mind to innovate, upgrade and transform the methods and practices suitable

for developmental needs of tomorrow. One of the key objectives of research should be its

usability and application. This journal attempts to document and spark a debate on the research

focused on technology in context of emerging subjects.

The first issue has been very carefully put together covering a range of technologies in the

domain Computer and Information Science & Engineering, Electronics and Communication,

Medical Electronics and Instrumentation Technology and computer applications but not limited

to The contributions have come in not only from India from industry and academics but also

from very renowned institutions and global industry groups as well.

In this issue of International Journal of Electrical Sciences and Engineering we present fourteen

topics of research for your closer study.

Editor in Chief

January 2016

For Subscriptions & Submissions [email protected]

III | Dayananda Sagar College of Engineering, Bengaluru-78

International Journal of Electrical Sciences & Engineering (IJESE)

Volume 1, Issue 1; January 2016

Contents

Editorial II

1. Improved CPU Utilization using Advanced Fuzzy Based CPU Scheduling algorithm (AFCS) 1

Xiao-Zhi Gao, Sasikumar Gurumurthy, S. Venkatesan

2. An Enhanced Approach in Cloud Computing to Reduce Security Risks and Minimize Data Loss in Railways 6

Ravi Gaurav, Shubham Kumar, S. Venkatesan, D.R. Ramesh Babu

3. Performance Measurement of Cryptographic Key using Biometric Images 13

Mohammed Tajuddin, C. Nandini

4. A Robustic Technique to Encrypt Medical Images using Bernstien Polynomial over Prime Field 19

Smitha Sasi, Santhosh B.

5. An Overview of Machine Learning and its Applications 22

Annina Simon, Mahima Singh Deo, S. Venkatesan, D.R. Ramesh Babu

6. Hadoop Framework for Flow Analysis and Congestion Control of the Big Data 25

Rakshitha Kiran P., Jeffrey Shafer

7. Intelligent Street Lighting System 29

Sanskar Gupta, Sonika Soni

8. A Secured Protocol for Efficient Discovery of Service in Vehicular Networks 33

Shalini K. B., Anil Sagar T., S. Venkatesan, Darshan H. Yogendra

9. Review on Emerging Techniques to Detect Oral Cancer 41

Santhosh B.

10. Types of Cognitive Agents Based on Learning Models 47

Devaraj Verma C., M.V. Vijayakumar, Darryl N. Davis

11. Nonintrusive Method for Liquid Level and Volume Measurement 53

V.G. Sangam, Vinyl Ho Oquino

12. Analysis of Routing Protocols in Ad-hoc Networks Using NS2 Simulator 58

Bindu Bhargavi S.M., Rekha Jayaram, Prathima Mabel J., Manasa Manjunath

13. Smart Transportation System Using Big Data Analytics 64

Ajay Kumawat, Hardik Singhi, Anil T. Sagar

14. Design of Reconfigurable Rectangular Patch Antenna using PIN Diode 68

Banuprakash R., Dr. Hariprasad S.A., Sai Raghuvamshi

International Journal of Electrical Sciences & Engineering (IJESE) Online ISSN: 2455 6068; Volume 1, Issue 1; January 2016 pp. 1-5 © Dayananda Sagar College of Engineering, Bengaluru-78

1 | Dayananda Sagar College of Engineering, Bengaluru-78

Improved CPU Utilization using Advanced Fuzzy Based CPU Scheduling algorithm (AFCS)

Xiao-Zhi Gao1, Sasikumar Gurumurthy2, S. Venkatesan3 1Professor, Department of Electrical Engineering and Automation,

Aalto University School of Electrical Engineering, 00076 Aalto, Finland [email protected]

2,Associate Professor, Department of Computer Science and Engineering, Dayananda Sagar College of Engineering, Bangalore, India

3Professor, Department of Computer Science and Engineering, Dayananda Sagar College of Engineering, Bangalore, India

[email protected]; [email protected]

Absract: The operating system in our computer machines have changed a lot during the course of time, where in the initial stage of their development they were used to process a single task (process) at a time but now, in the era of supercomputers we have multiprogramming operating system running in our machines. At present we have a number of scheduling algorithms which are used to decide the order in which the processes loaded into the memory are to be executed. But none of the conventional scheduling algorithms is ideal, they have their own drawbacks. In this paper, an advanced fuzzy-based logic has been proposed for soft real time system toovercome the drawbacks of other algorithms for better CPU utilization and to minimize waiting, turn-around and response time. The proposed algorithm is preemptive in nature with minimum context switching and work to complete process within its deadline.

Keywords: Fuzzy logic, CPU scheduling, soft real time system deadline, preemptive process deadline, Dynamic priority

1. INTRODUCTION

As of now scheduling real time system involve allocation of resources and CPU time to task in such a way that certain performance requirements are happened. In real-time system scheduling has played a more acute role than non-real time system because in this system having the right answer too late is as bad as not having it at all[10].

Such a system reacts to the request within a fixed amount of time which is called deadline. In general, real time system can be categorized into two important groups: Hard real time system and Soft real time system. In hard real time systems, when task occurs it strictly completed at a given deadlines. While in soft real time system missing some deadlines is acceptable. In both cases, the scheduler is to be schedule in such a way that guarantees the deadline to meet when a new task is arrived.

Scheduling algorithm is necessary and important task when more than one jobsare present in ready queue. Criteria forchoosing best scheduling is depend upon following basic featuressuch as:

• Waiting Time

• Turnaround Time

• Response Time

• Utilization of CPU

• Throughput

There are various type of scheduling algorithms such as first come first serve, priority based scheduling, shortest job first etc. The main constrain of real time task is that it should be completed within deadline time. The above scheduling algorithms are inefficient for real time operating system task. Hence we have proposed a new scheduling algorithm to find out the dynamic priority of process using fuzzy logic.

2. RELATED WORK

New era of possibilities were open when Lotfi A. Zadeh introduced the term “fuzzy logic” with proposal of fuzzy set theory. To make concept of approximation[1] reality this fuzzy logic can be used. Process scheduling with fuzzy logic has also been thought by many researchers[2][3][4]. Scheduling with deadline concept is basic requirement for real time system [10]. This schedule can be preemptive or non preemptive. Soft real time system with optimal time slice and dead line [5] is considered in this paper.

3. SCHEDULING ALGORITHMS

3. 1 FCFS Scheduling Algorithm

Even with all evolution in scheduling algorithm the FCFS serves as base algorithm. It is as simple as it sounds. The task

Xiao-Zhi Gao, Sasikumar Gurumurthy, S. Venkatesan


is executed as it comes to ready queue in arrival time order. There are some disadvantages of FCFS such as follows: a. This does not support preemption.

b. Throughput decreases as CPU holding time of a task increases.

c. There is no concept of priority. Turnaround time, waiting time and response time is very high which can reduces the performance.

3. 2 Priority Based Scheduling Algorithm

In this algorithm priority is associated to each process and depends upon the highest priority the process is assigned to the CPU. If process has equal priority then it scheduled in FCFS. We know that priority is assigned by operating system. The disadvantages of this algorithm are as follows

a. The major disadvantage of this algorithm is indefinite blocking it also called as starvation. We know that Low priority process gets interrupted by highest priority process. But if there is large number of highest priority process are present then each time it interrupted to low priority process then starvation occurred.

b. Another disadvantage is that the waiting time and turnaround time depend upon the priority of process.

3. 3 Shortest Job First Scheduling Algorithm

In this scheduling algorithm we are select the process with smallest burst time to execute the process. This is one of the best scheduling algorithm in which we get minimum waiting and turnaround time as compare to other scheduling algorithm. But there are some disadvantages of this algorithm are as follows:

a. It is very difficult to know the burst time for next CPU request.

b. Again this algorithm is not implemented for the shortest level CPU scheduling.

c. One major drawback is that process starvation for the process whose burst time is long if smallest burst time process is continuously arrived.

4. FUZZY LOGIC

Fuzzy logic is the superset of Boolean logic which deals with the truth values that is 0’s and 1’s. It is the nonlinear mapping form input data to the output data. The fuzzy logic system first collect the crisp set of inputs and convert it’s to the fuzzy set using fuzzy linguistic variable, terms and membership

function, this process is called as Fuzzification. This fuzzy set is use for making inference. Finally, we used the defuzzification step in which the resulting output is mapped with crisp output using membership function.

There are twokinds of Fuzzy Inference System such as (i) Mamdani’s fuzzy inference method and (ii) Sugeno fuzzy inference method.

5. PROCESS DEADLINE

In any of the real time system the tasks are assigned some deadline, failure to meet the deadline is not tolerable in hard real time system but the soft real time system does not lead to system failure only performance degradation happens. In this paper an algorithm is proposed to avoid process starvation with deadline concept using some optimal time slice to execute process. Preemptive process deadline is used to denote the maximum time till which the process can be preempted.

Proposed Algorithm

1. Check weather new process is arrived than add to ready queue else continue

2. While (ready queue != NULL)

3. Set dynamic priority to output FIS

Calculate dynamic priority (DPi):- 1. For each process Pi in ready queue fetch its parameters

burst time (BTi), static priority (PTi), and arrival time (ATi) and give them as input to FIS.

2. For each process (Pi), Evaluate membership function of priority (�p) �p=PTi/(max (PTi) +1); where 1<=i<=n

3. For each process (Pi), Evaluate membership function of burst time ( �b) �b=1-(BTi/(max (BTi) +1)); where 1<=i<=n

4. For each process (Pi) in ready queue find minimum priority process. To calculate dynamic priority (DPi)

5. If process Pi has minimum priority then DPi= (�p+�b) Else DPi= max {�p, �b} where 1<=i<=n

4. Calculate optimal time slice (OTS) only once for each process

Improved CPU Utilization using Advanced Fuzzy Based CPU Scheduling algorithm (AFCS)


X=half of the highest burst time in ready queue (upper bound) Y=average burst time in ready queue (consider upper bound) Z=highest burst time-(OTS of 1stprocess in queue) calculate Zevery time new process gets in ready queue

For 1st process in ready queue: If(X <= Y) OTS(Pi) =X [i=priority no 1 to 0] Else OTS(Pi) =Y [i=priority no 1 to 0] From 2nd process in ready queue: OTS(Pi) = Z [i=priority no 1 to 0]

5. Calculate deadline for each process in ready queue only once

For the process with highest priority: [i =highest priority] [i-1 =second highest priority] [D(Pi) =deadline of process Pi] [BT(Pi) =burst time of process Pi] [RBT(Pi-1) = remaining burst time of next process in ready queue] [AT(Pi) =arrival time of process Pi] D(Pi)= AT(Pi) +BT(Pi)+RBT(Pi-1) if(BT(Pi)<=OTS(Pi)) complete total task else switch on OTS(Pi) Gotostep(7) For next process except last process: [SPBT(Pi)= sum of previous burst time completion in CPU of process Pi] [PPD(Pi)=preemptive process deadline of process Pi+1] D(Pi)=AT(Pi) + D(Pi+1)+RBT(Pi-1) If(BT(Pi)<=OTS(Pi)&&(SPBT(Pi)+BT(Pi)<=PPD(Pi-1)) complete task Else switch on PPD(Pi) Otherwise switch on OTS(Pi) Gotostep(7)

For last process: D(Pi)=D(Pi-1)

If(BT(Pi)<=OTS(Pi)&&(SPBT(Pi)+BT(Pi)<=PPD(Pi-1)) complete task

Else

switch on PPD(Pi) Otherwise switch on OTS(Pi) Gotostep(7)

6. Calculate remaining burst time(RBT):

[RBT(Pi)=remaining burst time of process Pi, i denote priority no 1 to 0] [EBT(Pi)=total execution burst time of Pi, i denote priority no 1 to 0] RBT(Pi)=BT(Pi)-EBT(Pi) Gotostep(7)

7. Calculate deadline of preemptive process for every

process with priority lower than equal to latest executed process : for arrival in CPU: PPD(Pi)=D(Pi)-RBT(Pi) If(PPD(Pi-1)<=PPD(Pi)) ¦¦(PPD(Pi-1)<D(Pi)) Then

PPD(Pi)=PPD(Pi-1)-RBT(Pi) Gotostep(1)

8. Removing process from ready queue a) Remove from queue when next lower process

completes 2) Else remove if no lower priority process

9. If new process coming then Goto step (1).

6. RESULT AND PERFORMANCE EVALUATION

To demonstrate proposed algorithm some case studies have been considered with comparison to other algorithms on same cases. The results are denoted in terms of Gant chart and some statistical representation.

Case Study 1:

TABLE 1: Case Study 1 Data Set

Process ID

Arrival Time (ATi)

Burst Time (BTi)

Static Priority

(PTi)

Dynamic

Priority (DPi)

Deadline (D)

P1 0 3 2 0. 25 3

P2 2 6 7 0. 875 9

P3 4 4 5 0. 43 17

P4 6 5 6 0. 62 17

P5 8 2 1 0. 72 22

Xiao-Zhi G

4 | Dayanand

Gant Chart for priority Scheduling

PID P1 P2 P4 P3

0 3 9 14 18 20

Gant Chart for Improved Fuzzy based CPU

PID P1 P2 P5 P4

0 3 9 11 16 20

Gant Chart for Advanced Fuzzy based CPU

PID P1 P1 P2 P2 P5 P4

0 2 3 7 9 11 13 1620

RBT 1 0 2 0 0 3 0 0

PPD 2 3 7 9 22 10 13 13

Comparison Table

TABLE 3:Comparison between various algoristudy1

Algorithm Average Waiting

Time

Average Turnaround

Time

Priority Algorithm

4. 8 8. 8

IFCS 3. 8 7. 8 AFCS 3. 8 7. 8

Case Study 2:

TABLE 4: Case Study 2 Data Set

Process ID

Arrival Time (ATi)

Burst Time (BTi)

Static Priority

(PTi)

DynamPrior

(DP

P1 0 18 1 0. 13

P2 0 2 3 0. 89

P3 0 1 2 0. 9

P4 0 4 6 0. 7

P5 0 3 5 0. 8

P6 0 12 11 0. 91

P7 0 13 7 0. 5

Gant Chart for priority Scheduling

PID P6 P7 P4 P5 P2

0 12 25 29 32 34 35 53

hi Gao, Sasikumar Gurumurthy, S. Venkatesan

anda Sagar College of Engineering, Bengaluru-78

P5

PU Scheduling

P4 P3

PU Scheduling

P4 P4 P3

orithms for case

d Average Response

Time

4. 8

3. 8 3. 8

Set

namic iority

Pi)

Deadline (D)

136 53

894 18

. 95 13

. 79 35

. 84 22

917 15

. 58 53

P3 P1

Gant Chart for Improved Fuzzy b

PID P3 P6 P2 P5

0 1 13 15 18 22 35 53

Gant Chart for Advanced CPU Sc

PID P3 P6 P2 P6 P5

0 1 11 13 15 18 22 32 35 45 5

TABLE 5: Gantt chart f

RBT 0 2 0 0 0 0 3 0 8 0

PPD 3 13 18 13 22 35 32 3545

Comparison Table

TABLE 6: Comparison between vastudy

Algorithm Average Waiting

Time

ATur

Priority Algorithm

23. 86

IFCS 14. 86

AFCS 14. 71

Statistical analysis of the proposed

1. Waiting time vs. No. of Proce

0

5

10

15

20

25

Case Study 1

Priority IFCS

y based CPU Scheduling

P5 P4 P7 P1

Scheduling

P4 P7 P7 P1 P1

53

rt for case study 2

5 53

various algorithms for case

Average urnaround

Time

Average Response

Time

31. 43 23. 86

22. 43 14. 85

22. 43 14. 57

sed and existing algorithm

ocess

Case Study 2

CS AFCS

Improved CPU Utilization us

5 | Dayanand

Turnaround Time vs. No. of process

3. Response Time vs. No. of process

7. CONCLUSION AND FUTURE WORK

The proposed algorithm reduces the response ttime by minimal difference but with different difference of time may occur. This algorithmgiven processes within deadline. Preemptiooccurs. The time slice value is kept optim

0

5

10

15

20

25

30

35

Case Study 1 Case Study 2

Priority IFCS AF

0

5

10

15

20

25

Case Study 1 Case Study 2

Priority IFCS AFC

using Advanced Fuzzy Based CPU Scheduling algorit


K

e time and waiting nt case studies the hm completes the tion of processes imal to minimize

context switches and increase the rThis algorithm can be further immembership function in fuzzyficativalue can also be calculated with dfurther reduce the context switches.

REFERENCES

[1] Sankar K. Pal and Deba Prasaapproximate reasoning: an oCommunication Science Unit, Calcutta 700 035, India

[2] Shatha J. Kadhim and Kasim Evaluation of a Fuzzy-Based Springer-Verlag Berlin Heidelber

[3] H. S. Behera, RatikantaPattanImproved Fuzzy-Based CPU SchReal Time Systems”, Internationand Engineering (IJSCE) ISSN: 1, March 2012

[4] PrernaAjmaniand ManojSethi,Scheduling Algorithm (PFCS) Systems”, BIJIT - BVICAMInformation Technology BharComputer Applications and MaDelhi (INDIA)

[5] Vikash Chandra Sharma, PriyManindarsinghnehra,” CPU SDeadline and Optimize Time Slicinternational journal of enhancecomputer applications

[6] RajaniKumari, Vivek Kumar Design and Implementation ofScheduling Algorithm”, InternaApplications (0975 – 8887) Vol2013

[7] M. M. M. Fahmy,” A fuzzy alperiodic jobs on soft real-time sShams Engineering Journal (2010

[8] Patricia Balbastre, Ismael RiMinimum deadline calculation fdynamic priority systems”, SpOffice (CICYT)

[9] Mamdani E. H., Assilian S, “synthesis with a fuzzy logic conof Man-Machine Studies, Vol. 7,

Book References:

[10] William Stallings : Operating SEducation India, 2006.

dy 2

AFCS

dy 2

AFCS

rithm (AFCS)

e response time of processes. improved by choosing good ation process. The time slice different way of thinking to

es.

asad Mandal,” fuzzy logic and overview”, Electronics and it, Indian statistical institute,

M. Al-Aubidy,” Design and ed CPScheduling Algorithm”, berg 2010 tanayak, PriyabrataMallick,”An cheduling (IFCS) Algorithm for

ional Journal of Soft Computing N: 2231-2307, Volume-2, Issue-

i,” Proposed Fuzzy CPU S) for Real Time Operating M’s International Journal of aratiVidyapeeth’s Institute of Management (BVICAM), New

riyadarshini, Ajay Chaudhary, Scheduling Algorithm with

Slice for soft real time systems”, ced research in management &

r Sharma, Sandeep Kumar,” of Modified Fuzzy basedCPU rnational Journal of Computer olume 77 – No. 17, September

algorithm for scheduling non-e single processor system”, Ain 10) Ripoll and Alfons Crespo,” n for periodic real-time tasksin Spanish Government Research

, “An experiment in linguistic controller”, InternationalJournal 7, No. 1, 1975.

Systems 5th Edition. Pearson



An Enhanced Approach in Cloud Computing to Reduce Security Risks and Minimize Data Loss in Railways

Ravi Gaurav1, Shubham Kumar2, S. Venkatesan3, D.R. Ramesh Babu4 1,2Department of computer science & Enginnering, Dayanand Sagar College of Enginnering, Bangalore

3,4Professor, Department of Computer Science&Engineering, DayanandaSagar College of Engineering, Bangalore

Abstract: The distributed computing is another registering model which originates from framework processing, appropriated figuring parallel processing, virtualization innovation, utility figuring and other PC advances and has more point of interest characters, for example, huge scale calculation and information storage, virtualization, highexpansibility, high unwavering quality and low value service. Cloud processing has conveyed new changes and chances to IT industry. It is the consequence of the advancement of an assortment of techniques. And the railroad office will utilize the distributed computing innovation to accomplish the sharing of the rail route data assets furthermore, to enhance the limit of data handling. In any case, with the advancement of the distributed computing, it additionally confronted with numerous troubles, distributed computing security has turned into the main source of obstructing its improvement. Distributed computing security has turned into an intriguing issue in industry and scholarly research. This paper will investigate the status of the advancement of distributed computing security, break down the information privacy, security examining, information checking and different difficulties that the distributed computing security confronted with. We will depict the arrangements which the industry and the educated community proposed for some key issues of distributed computing security, for example, virtualization security and movement observing between virtual machines et cetera. Also, we broke down the security of distributed computing in railroad environment. We proposed a distributed computing security reference system. The motivation behind this paper is endeavored to bring more noteworthy clarity scene about distributed computing security.

Keywords-cloud computing; cloud computing security; cloud security framework; data leakage; encryption;

1. INTRODUCTION

Since 2007, cloud computing has been able to be hot issue, various associations began to attempt to use appropriated processing organizations. The common dispersed registering organization are Amazon's EC2 and's Google App Engine, they use the Internet to unite with outside customers, and take the broad number of programming and IT establishment as an organization provided for customers. With the solace,

economy, high versatility and diverse purposes of interest, appropriated registering enables the endeavor opportunity from the generous weight of the IT establishment organization and backing. Circulated registering change the Internet into another figuring stage, is an arrangement of activity that fulfill purchase on-interest and pay-per-use in framework, has a wide progression prospects. Railroad is the zones' one that proposed to offer need to make in national "Eleventh Five-Year Plan"; the change example of quick, overpowering and thick in rail line, makes an extensive variety of data including highlight and sound data in considerable scale growing, so it passes on enormous challenges to the information method of the railroad, including significant - scale appropriated figuring, data examination and taking care of, data sharing and the joining of enrolling resources and so forth; the conveyed registering as the improvement of distinctive progressions, has the key particular characteristics of dealing with the issues abovefl].

Yet, the change of disseminated processing is standing up to various separating issues, the most perceptible is the security issue, with the creating reputation of dispersed figuring, the importance of security show relentless upward example, transform into a basic segment in the headway of appropriated registering. 2009 Gartner diagram exhibited that more than 70% of respondents said they don't plan to use the circulated processing at later, the essential reason fears the data security and insurance. Moreover, the burst of different security events continue extending more people pushed over the cloud. Case in point, in March 2009, the event that a far reaching number of customer's archives were discharged happened in Google. Henceforth, remembering the final objective to affiliations and associations can make usage of tremendous scale cloud organizations, circulated processing advancement and stages, rest ensured that their data were migrated to the cloud, we must clarify the issues that dispersed registering security faced with. The inspiration driving this paper is tried to bring more paramount clarity scene about circulated figuring security.



2. II. THE CONCEPT OF CLOUD COMPUTING AND CHALLENGES

A. The concept of Cloud Computing

The thought of Cloud Computing Distributed registering is in being taken a shot at, there are no for the most part recognized bound together definition. In unmistakable periods of progression or from a substitute perspective has a substitute cognizance on the cloud. U. S. National Institute of Standards and Technology (NIST) portrays 5 key segments, 3 organization model and 4 course of action model of cloud[2], This definition is wide industry gathering.

B. Challenges

In 2008, the U. S. data innovation research and counselling firm gartner issued a “distributed computing security hazard appraisal” report, fundamentally from the seller perspective about security abilities investigated security dangers confronted by the cloud, posting seven noteworthy security chances that the distributed computing innovation exists [3], as appeared in table 1

TABLE 1: SEVEN TOP SECURITY RISKS GARTNER

RISK Description

Privileged user access

Delicate information handled outside the undertaking carries with it an inborn level of danger

Regulatory compliance

Distributed computing suppliers who decline to outside reviews and security affirmations

Data location When you utilize the cloud, you most likely won't know precisely where your information is facilitated

Data segregation Information in the cloud is ordinarily in a common situation nearby information from different clients

Recovery Investigative support

Regardless of the fact that you don't know where your information is, a cloud supplier ought to let you know what will happen to your information and administration in the event of a catastrophe

Long-term viability

Researching unseemly or illicit movement may be incomprehensible in distributed computing. You must make certain your information will stay accessible even after such an occasion

• Data protection: As a client, we lose control over physical security, by what means would we be able to guarantee

that information won't spillage and privacy can be protected

• Key administration: If the data is encoded, then who controls the encryption/unscrambling key? Client or Service supplier?

• Data Integrity: It is not exist that a typical standard to guarantee information hone.

3. SAFETY STATUS OF THE CLOUD.

A. The government concern about the safety of the cloud

November 2010, the U. S. government and agency CIO

CSA appropriated in 2009 "in key locales of the cloud Safety Guide" and moved up to frame 2. 1 [4], generally from the perspective of the aggressor packed the huge threats that conveyed figuring environment may be stood up to, proposed 12 key fields that security concerns, then issued a cloud reduced reports security perils, the Security Guide was concentrated to 7 of the most broadly perceived, the best hazard to pernicious levels, as showed in Table II.

TABLE 2: SEVEN TOP SECURITY RISKS CSA

Risk Description

Abuse and Nefarious Use of Cloud Computing

By abusing the relative anonymity behind these registration and usage models, spammers, malicious code authors, and other criminals have been able to conduct their activities with relative impunity

Insecure Interfaces and APIs

It increases risk as organizations may be required to relinquish their credentials to third parties in order to enable their agency.

Malicious Insiders A provider may not reveal how it grants employees access to physical and virtual assets, how it monitors these employees, or how it analyzes and reports on policy compliance.

Shared Technology Issues

The underlying components that make up this infrastructure (e. g., CPU caches.) were not designed to offer strong isolation properties for a multi-tenant architecture.

Data Loss or Leakage

The threat of data compromise increases in the cloud, due to the number of and interactions between risks and challenges which are either unique to cloud, or more dangerous because of the architectural or operational characteristics of the



cloud environment. Account or Service Hijacking

Your account or service instances may become a new base for the attacker. From here, they may leverage the power of your reputation to launch subsequent attacks.

Unknown Risk Profile

Versions of software, code updates, security practices, vulnerability profiles, intrusion attempts, and security design, are all important factors for estimating your company's security posture.

Asked the government to assess the security risks about the cloud computing, described the challenges of cloud computing and the security for cloud computing.

September 2010, Westone held a conference about cloud security and cloud storage in Beijing, responded enthusiastically.

May 2010, in the second session of the China Cloud Computing Conference, the Ministry of Industry Vice Minister Lou said, we should strengthen the information security of cloud computing to solve common technical problems. March 2010, the European Network and Information Security Agency (ENISA) announced they will promote the management department to request cloud service provider not conceal attacks on cloud.

B. The Cloud Security Standards Organization and Progress on Cloud Many standards organizations have begun to develop standards of cloud computing and cloud security, in order to enhance interoperability and security, to promote the healthy development of cloud computing industry. Such as: Open Cloud Manifesto (OCM), National Institute of Standards and Technology (NIST), Cloud Security Alliance (CSA) and Distributed Management Task Force (DMFT).

• Open Cloud Manifesto (OCM) There are more than 300 units join in the organization currently, the main results of the organization is open cloud manifesto [6], the open cloud manifesto describes the challenges that cloud computing faced with, including governance and management, security, data and application interoperability and portability, measuring and monitoring.

Other challenges to be aware of: [5] • National Institute of Standards and Technology (NIST)

NIST fundamentally through specialized direction and elevate the institutionalization work to help government and industry

sheltered and compelling utilize the distributed computing innovation. In the May 2010, the NIST held symposium about the distributed computing, in October 2009, issued a "sheltered and compelling utilization of distributed computing" scholastic discourse. The primary aftereffects of NIST are:

NIST meaning of distributed computing VI5 [7], the archive gives the meaning of distributed computing, and portrays five attributes of distributed computing, three administration models and four arrangement examples, and reference by numerous Standards Alliance, for example, DMTF.

Sheltered and productive utilization of distributed computing report [8], the report presents the idea of distributed computing, elements, outline, and nitty gritty examination of the distributed computing security, relocation and other related advances and the institutionalization about cloud security.

• Cloud Security Alliance (CSA)

CSA is a non-advantage affiliation, establishment in the RSA Conference in 2009, rule focused on the security dangers that the endeavor went up against with when sending the conveyed registering structure and given the correct wellbeing urging. CSA suggestion a conveyed processing security basic arranging reference model [4], as showed in Figure

Fig. 1. CSA cloud computing security architecture reference model



CSA proposed 15 focus areas of cloud computing security from the point of cloud management and cloud computing run, as shown in Tablelll

TABLE 3. CSA 15 FOCUS AREAS OF CLOUD SECURITY

Cloud Cloud computing run

management Compliance

and Traditional Security Disaster Recovery

Audit Governance

and Emergency Notification and

Enterprise risk Response Repair management

Portability and Business Continuity Application

Security Interoperability

Legal and Encryption and Key Virtual ization Electronic Management discovery

Information Data Center Identity Access Lifecycle Security Management

Management

• Distributed Management Task Force (DMFT)

The association worried about the distributed computing administration principles, concentrate on enhancing the interoperability about cloud administration between cloud administration suppliers and clients, and between cloud administration suppliers and cloud administration designers, built up the interoperability measures through improvement the assention of cloud assets administration, embodiment organization and security components. The primary aftereffects of DMTF [9] are: the Open Virtualization Format Specification (OVF), cloud administration construction modeling, cloud interoperability white paper.

4. STEPS TO PREVENT DATA LEAKAGE AND DETECTION

Data Leakage Detection: With the snappy advancement of database business on the net, the data may be hazardous in the wake of experiencing the unsecure framework. The data purchasers may dither to buy the data organization for the going with suspicion. In any case, the data recipient may suspect that the data are upset by unapproved person. Second, they may suspect the data got are not conveyed and gave by the endorsed suppliers. Third, the suppliers and purchasers truly with differing interest should have particular parts of rights in the database organization or using. So how to secure and affirm the data ends up being fundamental here. The late

surge in the web's improvement results in offering of a broad assortment of electronic organizations, for instance, database as an organization, mechanized chronicles and libraries, e-exchange, online decision sincerely strong system et cetera. All through coordinating, here and there delicate information must be offered over to obviously trusted outsiders. For example, an expert's office may give patient records to specialists who will devise new arrangements. We call the information's proprietor the shipper and the to the degree anyone knows trusted outsiders the professionals. We will apparently perceive when the merchant's delicate information have been spilled by experts, and if conceivable to see the directors that released the information.

We consider applications where the first fragile data can't be irritated. Inconvenience is a particularly important system where the data are balanced and made "less tricky" before being given to masters. For example, one can add unpredictable clatter to particular properties, or one can supplant exact qualities by scopes [13]. On the other hand, on occasion, it is essential not to adjust the first shipper's information. For example, if an outsourcer is doing our cash, he must have the accurate compensation and client record numbers. On the off chance that supportive specialists will be treating patients (instead of fundamentally planning estimations), they may require definite information for the patients. For the most part, spillage watermarking in order to recognize evidence is managed, e.g., a novel code is embedded in each appropriated copy. If that copy is later found in the hands of an unapproved assembling, the leaker can be recognized.

Watermarks can be especially profitable from time to time, however again, incorporate some adjustment of the first data. In addition, watermarks can now and again be crushed if the data recipient is vindictive. In this paper, segment I gives the examination of systems to perceiving spillage of a game plan of things or records. In the wake of giving a course of action of things to experts, the wholesaler discovers some of those same articles in an unapproved spot. (For example, the data may be found on a site, or may be overcome a true blue revelation process.) At this point, the dealer can overview the likelihood that the spilled data began from one or more administrators, as opposed to having been self-rulingly aggregated by distinctive means. In case the dealer sees "enough verification" that an experts spilled data, he may stop working with him, or may begin honest to goodness strategies. In area II a blameworthy operators is present which is create for evaluating the "blame" of specialists furthermore display calculations for disseminating articles to operators, Sections III and IV, introduce a model for computing "blame" probabilities in instances of information spillage. At long last, in Section V, assessing the methodologies in distinctive information spillage situations, and check whether they without a doubt distinguish a leaker.



I HOW IS ACCESS TO THE DATA GAINED?

"Who brought on the gap?" quality. These qualities are not tradable, yet rather comparing and the diverse ways to deal with acquire access to sensitive data can be bundled into the going with social affairs. Physical spillage channel suggests that physical media (e.g., HDD, convenient PCs, workstations, CD/DVD, USB devices) containing unstable information or the record itself was moved outside the affiliation. This more every now and again infers that the control over data was lost even before it leaved the affiliations.

Difficulties are an) Encryption: and averting information spills in travel are hampered because of encryption and the high volume of electronic interchanges. While encryption gives intends to guarantee the privacy, credibility and uprightness of the information, it likewise makes it hard to recognize the information holes happening over encoded channels. Encoded messages and document exchange conventions, for example, SFTP suggest that correlative DLP mechanisms should be employed for greater coverage of leak channels. Employing data leak prevention at the endpoint – outside the encrypted channel can possibly identify the breaks before the correspondence is scrambled. b) Access Control: It gives the first line of guard in DLP. On the other hand, it doesn't have the best possible level of granularity and may be obsolete. While access control is suitable for information very still, it is hard to execute for information in travel and being used is not included in. C) Semantic Gap in DLP: DLP is a multifaceted issue. The meaning of an information hole is liable to differ between associations relying upon the delicate information to be ensured, the level of cooperation between the clients and the accessible correspondence channels. The present cutting edge principally concentrates on the utilization of abuse identification (marks) and after death examination (criminology). The regular deficiency of such methodologies is that they do not have the semantics of the occasions being observed. At the point when an information hole is characterized by the conveying gatherings and the information traded amid the correspondence, a straightforward example coordinating or get to control plan can't induce the way of the correspondence. Hence, information spill counteractive action components need to stay informed concerning who, what and where to have the capacity to guard against complex information spill situations. The characterization by spillage station isimportant with a specific end goal to know how the occurrences may be averted later on and can be named physical or logical. enerally, spillage disclosure is dealt with by watermarking, e. g. , a noteworthy code is introduced in each flowed copy. If that copy is later found in the hands of an unapproved assembling, the leaker can be recognized. Watermarks can be to a great degree accommodating once in a while, yet again, incorporate some change of the first data. Plus, watermarks can as a less than dependable rule be devastated if the data recipient is malicious. E. g. A facility

may give patient records to researchers who will devise new medications

II Guilty Agent

To perceive when the wholesaler's tricky data has been spilled by administrators, and if possible to recognize the authorities that discharged the data. Disturbance is an amazingly important methodology where the data is changed and made "less fragile" before being given to administrators. An unassuming system is made for distinguishing spillage of a game plan of articles or records. Accept that in the wake of offering articles to pros, the shipper finds that a set S _ T has spilled. This infers some untouchable, called the goal, has been gotten having S. For example, this target may be demonstrating S on its site, or perhaps as a noteworthy part of a genuine disclosure change, the goal turned over S to the wholesaler. Since the administrators U1; Un have a data's rate, it is sensible to suspect them discharging the data. Then again, the administrators can battle that they are guiltless, and that the S data were gotten by the target through diverse means. Case in point, say that one of the things in S identifies with a customer X. Possibly X is moreover a customer of some other association, and that association gave the data to the target. Then again perhaps X can be repeated from distinctive straightforwardly available sources on the web. We will probably gage the likelihood that the spilled data started from the masters rather than diverse sources. Intuitively, the more data in S, the harder it is for the experts to battle they didn't discharge anything. So additionally, the "rarer" the things, the harder it is to battle that the target procured them through diverse means. Not simply would we need to gage the likelihood the administrators spilled data, on the other hand we may moreover get a kick out of the chance to see whether one of them, particularly, was more disposed to be the leaker. For instance, if one of the S things was simply given to administrators U1, while substitute articles were given to all experts, we may suspect U1 more. The model we show next gets this sense. We say an administrators Ui is at risk and if it contributes one or more dissents the goal. We mean the event that administrators Ui is accountable by Gi and the event that masters Ui is subject for a given discharged set S by Gi/jS.



DATA LEAKAGE PREVENTION: Data spill revultion (DLP) is a course of action of information security gadgets that is proposed to keep customers from sending delicate or fundamental information outside of the corporate framework. Appointment of DLP, contrastingly called DATA mishap evasion, information incident expectation or removal neutralizing activity, is being driven by essential insider threats and by more exhaustive state security laws, a critical number of which have stringent data confirmation or access fragments. DLP things use business rules to take a gander at record substance and label private and separating information with the goal that customers can't reveal it. Marking is the technique of requesting which data on a system is private and checking it suitably. A customer who adventitiously or malevolently attempts to uncover mystery information that has been named will be denied. Case in point, marking may even keep a sensitive budgetary spreadsheet from being informed by one delegate to another within the same organization. DLP things generally have the going with fragments: Endpoint: Monitor and control activities Network: Filter data streams Storage: Protect data still. Actualizing an undertaking DLP item can be entangled.

Most expansive associations have several servers with a huge number of catalogs and records put away on them and particular sorts of information that should be labeled. The product can be helpful for recognizing very much characterized substance (like Social Security or Mastercards numbers) yet tends to miss the mark when a chairman is attempting to distinguish other delicate information, similar to protected innovation that may incorporate realistic parts, recipes or schematics. To actualize undertaking DLP effectively, work force from all levels of administration should be effectively included in making the business rules for labels. . Information spill aversion (DLP) is a suite of advances went for stemming the loss of touchy data that happens in endeavors over the globe. By concentrating on the area, grouping and observing of data very still, being used and in movement, this arrangement can go far in helping an undertaking understand what data it has, and in ceasing the various breaks of data that happen every day. DLP is not a fitting and-play arrangement. The fruitful usage of this innovation requires critical readiness and industrious progressing upkeep. Ventures trying to coordinate and actualize DLP ought to be arranged for a huge exertion that, if done accurately, can extraordinarily lessen danger to the association. Those executing the arrangement must take a key approach that addresses dangers, effects and alleviation ventures, alongside fitting administration and affirmation measures. New little and fair size endeavors can assimilate both the monetary and PR harm dispensed by genuine breaks focusing on delicate information. But, they're frequently under ensured in light of the fact that information spill counteractive action, or DLP, items are, generally speaking, essentially excessively costly. In the interim, there's been a critical rise in

cybercrime following a consistent five-year decrease, as indicated by the 2007 CSI Computer Crime and Security Survey. Insider misuse of system resources is the most common assault, ahead even of infections, with normal misfortunes of around $350, 000. Code Green Networks, which was dispatched by the originators of Sonic Wall, means to handle this issue. Code Green's most up to date offering, the CI-750 Content Inspection Appliance, is designed particularly for systems with 250 or less clients and offers the same components and usefulness as its higher-finished items, beginning at $10, 000. The CI-750 uses "fingerprints" to distinguish both organized information, for example, Social Security or charge card numbers, and unstructured information, for example, records, documents, source code, etc. Where numerous DLP items for littler organizations depend on sifting for certain document sorts or give just essential pivotal word or example matching, Code Green's technology creates hash values of the actual data to be protected and scans outgoing traffic for matches. We found Code Green's fingerprinting technology accurate, and a built-in mail transfer agent. However, without the help of third-party proxies, the appliance is blind to encrypted data, and it can't stop movement of internetwork and web-based traffic. This means the appliance represents only part of a robust.

DLP system. FINGERPRINT TRAIL: The CI-750 can be deployed in a variety of ways. Included a kit it was a network tap device, which let us passively monitor traffic flowing through our WAN connection, and a mail transfer agent. Customers can route outgoing messages from their mail servers through the mail transfer agent for additional mail-filtering abilities; questionable e-mail can be held until approved by an administrator. Admin also can create policies to encrypt e-mail carrying sensitive information. This functionality is provided via Code Green's partnership with the Voltage Security Network, which offers e-mail encryption as a service. After connecting the device to network, A selected sources of data that the appliance should protect. It has built-in functionality to fingerprint both structured and unstructured data such as that in CIFS. Setup for CIFS was simply a matter of providing the server and share name, along with appropriate access credentials. The device then scans the share at user-defined intervals. CIFS scanning was trouble-free and didn't cause performance issues on our Windows file server.

Be that as it may, it's officeholder on IT to guarantee that substance to be fingerprinted gets put into the fitting CIFS offer. This can be tricky. For instance, our organization depends vigorously on private wiki pages and not shared volumes for the majority of our interior data. Code Green's proposed workaround is to have a script that dumps the substance of our wikis to a CIFS offer all the time. Given the uptick in communitarian workspaces, for example, wikis in the business group, we'd like to see a completely computerized approach to get such information fingerprinted.



It in like manner would look good if the contraption could use Web pages as sources clearly; support for other data stores furthermore would grow the compartment's out convenience of this machine and discard the prerequisite for extra scripting. It should be noted, then again, that various battling offerings, some essentially more excessive, don't even offer database joining. Resulting to selecting data hotspots for fingerprinting, IT then describes development to screen and what moves should be made in the event a break is recognized. We organized some comprehensively scrutinized chooses and found that the CI-750 made a wonderful demonstrating alerted us to data discharges happening within email, Web, IM, and even stuffed report transmissions. We fused a two-sentence part from an assention in an email to a client. Following a moment, we had an email communicating that there had been an encroachment. The official interface on the machine exhibited that an email had been sent to our customer and had the full association of the email to show the encroachment. The interface can similarly appear past encroachment that may have been joined. Partial PREVENTION: While we were awed with the fingerprinting's exactness, the contraption did not have the limit truly seclude the message in light of the way that it was sent by method for Web mail. Associations that need healthy frustrating of Web and framework movement should place assets into a mediator device. The Code Green

machine can be outlined as an Internet Content Adaptation Protocol server when joined with an ICAP delegate, for instance, those from Blue Coat Systems or Squid. Exactly when so joined, Code Green can square HTTP, HTTPS, and FTP action. It in like manner can unscramble development for examination. Versatile workstations similarly will speak to an issue for Code Green customers. The association offers an endpoint experts that controls the usage of removable media, for instance, glimmer drives and CDs.

5. CONCLUSION

From the investigation of the information spillage, we can identify and keep the information from the break by utilizing a few calculations and methods and can apply in Indian railroads. Ideally there would be no compelling reason to hand over touchy information to specialists that may accidentally or perniciously spill it. Furthermore, regardless of the possibility that we needed to hand over delicate information, ideally we could watermark every article so we could follow its starting points with supreme sureness.

REFERENCES

[1] LIU Zhen, LIU Feng, ZHANC Baopeng, MA Fei, CA 0 Shiyu. Research on Cloud Computing and Its Application in Railway [J]. Journal of Beijing Jiaotong University, 2010, 34(5): 14-19. (inChinese)

[2] MELL P, GRANCE T. The NIST Definition of Cloud Computmg [EB/OL]. [2010-05-10]. http://csrc. nist. gov/groups/SNS/cloud-computing/.

[3] MATHER T, KUMARASWAMY S, LATIF S. Cloud Security and Privacy: An Enterprise Perspective on Risks and Compliance [M]. [s. l. ]: O'ReillyMedia, Inc., 2009.

[4] Cloud Security Alliance. Security Guidance for Critical Areas of Focus in Cloud Computing V2. 1. 2009.

[5] http://www. infosectoday. com/Articles/Cloud_Security_Challenges. htm

[6] http://www. opencloudmanifesto. org/ [7] National Institute of Standards and Technology. NIST

definition ofcloud computing VI5. 2010 [8] National Institute of Standards and Technology. Safeand

efficient useof cloud computing report. 20094362 [9] http://www. dmtf. org/ [10] L. Sweeney, “Achieving K-Anonymity Privacy Protection

UsingGeneralization and Suppression,” http://en. scientificcommons. org/43196131 2002.

[11] Ravi Gaurav Shubham Kumar et. al, Cloud Security in Southern Railways in India National Conference on Information and Technology organised by Dayananda Sagar College of Engineering 2015. PP 42-47



Performance Measurement of Cryptographic Key using Biometric Images

Mohammed Tajuddin1, C. Nandini2 1Associate Professor, Department of Computer Science and Engineering,

Dayananda Sagar College of Engineering, Bangalore [email protected]

2Professor, Department of Computer Science & Engineering, Dayananda Sagar Academy of Technology & Management, Bangalore

[email protected]

Abstract: The biometric field is one of the promising frontiers of scientific advancement in network security applications. Hence, the cryptographic key is generated using the retina biometric features. The Biometric features will improve the security of cryptographic system. In this paper, thinned vascular tree of retina biometric is used to generate the cryptographic key. This work emphasizes upon unification of features which enables to generate more secured cryptographic key. This work introduces a unique method to generate a secured cryptographic key for any network security applications. This technique of operations in network security creates more complexity for hackers to crack or guess the key. Thus, security is further enhanced using the above technique.

Keywords: Cryptography, Biometrics, Endpoints, Bifurcation, island, Morphological operation and vascular tree, Encryption & decryption

1. INTRODUCTION

With the rapid growth of internet and the advancements of network communication technologies, the communication channel must be secure and protect the message send across the communication channel at the same time user expecting secure data transmission.

Many cryptographic algorithms are used which are simple and efficient to implement on high performance to convert the message into unintelligible message. Some cryptographic algorithms operation which takes few milliseconds in securing the messages, perform authentication and the integrity check on machine. Hence, it is critical to user to select the specific algorithm to provide maximum security. The primary objective of cryptography is to ensure provisioning of confidentiality, integrity and availability. The goal of confidentiality is data exchange between two users must be on trusted network, the information while exchange remains unchanged and secret. Integrity the information is always exchange between two users, but changes should be made by authorized users only [5]. Integrity prevents the modification

and to detect any modification made to the message. The confidentiality and integrity should not hinder the availability of data. The data must be available to the authorized users only. Types of cryptography algorithms secret key cryptography, public key cryptography and hash function [19]. In secret key cryptography only one key is used for encryption and decryption while in Public key cryptography one key encryption and another key for decryption of message.

It is worth to recall that security has become an increasingly important factor with the growth of digital world. The Symmetric in which the same key value is used in both the encryption and decryption calculations are popularly used. The AES algorithm is capable of using cryptographic keys of 128, 192, and 256 bits to encrypt and decrypt the message in blocks of 128 bits.

• AES is used as it is simple to implement by using cheap processor and uses minimum amount of memory. It has better resistance against existing attacks and increases the security with less power and high throughput. AES uses four types of transformations namely substitution, permutation, mixing and key adding [7].

• AES such as DES uses substitution. However, the mechanism is different where the substitution is done for each byte.

• Another process found in a round is shifting. Shifting transformation in the AES is done at the byte level: the order of the bits in the byte is not changed [6].

• In encryption, the transformation is known as Shift Rows and the shifting is to the left.

• The mixing transformation changes the contents of each byte by taking four bytes at a time and combining them to recreate four new bytes.



• The functionality in the Add Round Key is matrix addition. Since addition and subtraction in this field are the same, the Add Round Key transformation is the inverse of itself.

Block cipher there are various modes of block cipher have been tested, the serve in this paper is AES [5].

Hash function several widely used hash function are evaluated such as MD2 [15], MD4 [16], MD6 [17] and SHA-1[18].

Cryptographic techniques are widely used for ensuring the secrecy and authentication of database information. The secure protection of information depends upon the cryptography key, which is only known to the authorized users. Maintaining the secret key is one of the challenging issues over the internet [7]. The security of information in encryption system is depends on the technique used to generate the secrecy key for encryption and decryption instead of the encryption algorithm. The encryption system is thus unable to protect the cipher text once the algorithm is broken. The security level of an encryption algorithm is measured by the size of its key space [4]. Larger the size of the key space more time does the attacker needs to do the exhaustive search of the key space. Thus, the level of security is higher. In encryption, the key is piece of information which specifies the particular transformation of plaintext to cipher text, or vice versa during decryption.

Biometrics is an emerging field of technology using unique and measurable physical and behavioral characteristics that can be processed biometric features for identification of a person. The biometric attributes include facial appearance, fingerprint, gait, geometry handwriting, iris, retina, veins and voice. Retina biometric identification is an automatic method that provides true identification of the person by acquiring an internal body image which is difficult to counterfeit [1]. Retina identification has gained its importance in application in high security environments. Retina biometric is unique biometric pattern that can be used as part of a verification system.

The rest of the paper is organized as follows: Section II provides the brief description of a generation of key from retina biometrics. Section III provides the background principles related to the working of the proposed model. All experimental results and related discussion is provided in Section IV. This paper is concluded by summing up the work in Section V.

2. II GENERATION OF KEY USING RETINA BIOMETRIC:

Biometric features such as the number of end points, the number of bifurcation points and the number of islands [25].

The above features are unique to all the retina biometric images, which is the unique method to generate the cryptographic key for encryption and decryption of message using cryptography algorithms. Accept the retina biometric from the database, convert the retina biometric to gray image, the values of gray image in the range 0 to 255 and then gray image to binary image in the form of 0’s or 1’s. From the binary image extract the blood vessels by setting the threshold value, the resultant tree is known as vascular tree as shown in Figure (2).

Fig. 1. Design Diagram

Fig. 3. Vascular Tree from the retina image

Next activity is to thin the generated vascular tree by using morphological operation such as dilation, erosion and open



etc. the broad blood vessels are converted to thin line connected components, since we can find the end points from the thinned image as shown in Figure 3.

Fig. 3. Thinned vascular tree

Figure 3 indicates the number of end points, number of bifurcation points and the number of islands. Further divide this image into for quadrants, and then select the 3rd quadrant which has unique features as shown Figure 5, with reference to Figure 5.

Fig. 5: 3rd Quadrant of a thinned image

Figure 5 has the unique features compare to the remaining quadrants of the thinned image as shown in Figure 3. Find the number of end points, bifurcation points and the islands. Key calculation of number of end points is by using 8 pixel neighborhood structuring element which is further depicted in Figure 4. Move the structuring element from the origin pixel by pixel to find the connect components of a line, if the structuring element unable to find the neighbor element with 1 indicates a end point and the counter will increment by 1 and the pixel coordinate x and y axis with angle will be considered as shown in Table 2.

The same image is further investigated to identify the number of bifurcation points. As per the conventional assumption to detect a line is to check for two adjacent pixel values to be 1. Applying the same principles here, the number of bifurcation points is obtained for the image. Table 3 indicates the bifurcation points, x and y values for those points using MATLAB code. Table 3 thus indicates that for the sampled

retina image there exist 16 bifurcation points with the process of detecting end points and bifurcation points from the retina image [19].

1 1 1

1 1 1 1 1 1

Fig. 4. Structuring element 8 pixels

Finally, find the number of islands in Figure 5 by using principal components analysis to find the island of an image and also find the sum of area of islands which is unique for each image as shown in Figure 5. The extracted features are unique, since these features generate a unique biometric key and it used to encryption and decryption using AES algorithm.

Key Generation method by using the three retina features such as number of end points with angle, Number of bifurcation points with three degree and the number of island with the sum of the island area.

The following steps are used to generate the unique biometric key.

1. Read in the input retina image.

2. Feature extraction such as retina blood vessels

3. Thinning using morphological operations

4. Divide the thinned image into 4 quadrants

5. Select the 3rd quadrant which has unique features compare to other quadrants.

6. Find the number of end points

k1 = � �[�� ∗ y�� ]

7. Bifurcation points

k2 = � �[�� ∗ �� ]

8. K3 is number of islands are 8 with these features we can generate the secured key for cryptographic applications.

9. Key = k1* k2 *k3

3. ENCRYPTION PHASE

In this paper, biometric features are used to generate biometric key for the cryptographic systems. In the encryption method



the part of the retina biometric features are used to generate the cryptographic key, the generated biometric key is converted to 128 bit key. In symmetric cryptography, a same key is used for both the encryption and decryption process. According to this methodology, a key must be same to both encrypting and decrypting method [11]. For encryption and its reverse process, in this work we use the Advanced Encryption Standard (AES) algorithm [19]. The overview of the proposed retina encryption phase is shown in Figure 6.

Fig. 6. Encryption phase using AES

The cipher text is the result of encryption performed on the plain text using an AES algorithm is called cipher. The cipher text is also known as encryption or encoded information because it contains a form of the original plain text which is unreadable by human or computer without the proper cipher to decryption using the same AES algorithm.

The generated cipher text again transforms back to original message by using decryption with AES algorithm as shown in Figure 7.

Fig. 7. Decryption phase

Results are obtained from different datasets such as DRIVE and stare. These are the following results such as number endpoints as shown in Table 2 from the origin.

TABLE 2: The number of end points

No X y

284 41 272 60 194 63 375 76 235 111 327 114 120 119 378 134 327 142 345 146 145 159 382 212 363 224 321 247 323 248

TABLE 3: The Bifurcation points

No X y 226 33 254 36 231 39 290 40 218 41 299 43 224 67 192 85 186 92 263 94 320 119 262 141 281 163 342 200 328 204

Table 3 illustrates the number of bifurcation points in Figure 5 and Table 2 show the number of end points in Figure 5, the complete code written in MATLAB. Finally the last feature numbers of islands in the 3rd quadrant are 8.

Performance measurement was tested to find the amount of time required to perform the encryption and decryption with different keys and different size of message with the existing



method and our method. For each algorithm a number of tests are conducted where time taken is recorded as a sample time for each input message [20], Our approach execution time is less.

Rijindaels Functions: Table 4. Performance of AES

Key size

Message: 2KB Message: 50KB Message: 4 MB

Throughput (Bps)

Throughput (Bps)

Throughput (Bps)

Palm V

Retina Palm V

Retina Palm V

Retina

128 192 256

10673 8989 7764

10500 8910 7775

10711 8947 7673

10612 8932 7661

10522 8769 7516

10496 8712 7498

Figure 8: Comparison throughput of AES between palmv & Retina

DES Functions:

In this we also check the performance measurement results of DES with our approach, the performance of encryption and decryption with our approach results are better than existing methods [20]. We tested few samples with DES and the key is generated from the DRIVE dataset, Stare Data set for DES.

Hash Functions:

We also experimented with Hash function results of MD2, MD5 and our approach. All these algorithms are implemented in MATLAB library. The tested results are as shown in Table 5. By using hash function we found a bit improvement on the hashing speed for few hash functions [20], improvement in the range of 3 % to 5% with existing results. As we increase the

size of the encryption and decryption time also increases tested with few key sizes. The results are as shown in Table 5.

TABLE 5: Performance of Hash

Hash Message: 2KB Message: 50KB

Message: 4 MB

Throughput (Bps)

Throughput (Bps)

Throughput (Bps)

Palm V

Retina Palm V

Retina Palm V

Retina

MD2 MD5 SHA-1

1738 38102 17429

1712 37962 17428

1747 37647 17770

1698 37556

-

1739 37608 17664

1740 37553

-

Fig.9. Comparison of throughput of Hash function between palm & Retina

4. CONCLUSION

The technique used to generate the cryptographic key is unique by using the retina biometric features. To generate the cryptographic key, more permanent features of the retina biometric are used such as the sub graph of vascular tree which has more permanent features compare to other quadrants in which the number of end points, bifurcation points and islands are used to create a cryptographic key for encryption and decryption of message. This paper has put forth the performance measurement comparison which is made between various existing cryptographic approaches and with variation in key and the size of messages. The comparative results has brought out the improvement of existing technique where instead of considering the complete image, it is now possible to generate the key by considering the part of the thinned image which has unique features.

0

2000

4000

6000

8000

10000

12000

key size Palm V Retina Palm V Retina Palm V Retina

Palm V Retina

05000

1000015000200002500030000350004000045000

Palm V Retina Palm V Retina Palm V Retina

MD2 MD5 SHA-1



REFERENCES

[1] Hill, R. B. 1978. Apparatus and method for identifying individuals through their retinal vasculature patterns, US Patent No. 4109237

[2] Chih-Peng Fanand and Jun-Kui Hwang, "FPGA Implementations Of High Throughput Sequential And Fully Pipelined AES Algorithm" International journal of Electrical Engineering, vol. 15, no. 6, pp. 447-455, 2008.

[3] Mehran Mozaffari-Kermani and Arash Reyhani-Masoleh, "Efficient and High Performance Parallel Hardware Architecture for the AES-GCM" IEEE Transactions on Computers, vol. 61, no. 8, August 2012.

[4] A. A. Zaidan, A. W. Naji, Shihab A. Hameed, Fazidah Othman and B. B. Zaidan, " Approved Undetectable-Antivirus Steganography forMultimedia Information in PE-File ", International Conference on IACSIT Spring Conference (IACSIT-SC09), Advanced Management Science (AMS), Listed in IEEE Xplore and be indexed by both EI (Compendex) and ISI Thomson (ISTP), Session 9, P. P 425-429.

[5] Behrouz A. Forouzan, “Cryptography and Network Security”, Tata McGraw-Hill, 2007.

[6] Ramya M., Muthu Kumar A., KannanS. “Multibiometric Based Authentication Using Feature Level Fusion”, IEEE-International Conference On Advances In Engineering, Science And Management (ICAESM -2012), pp. 203-207, Mar 30, 31, 2012.

[7] R. Jubiaya, M Keirthi, “IRIS Authentication Based on AES Algorithm”, IJIRSET, Volume 3, special issue 3, March 2014.

[8] X. Zhang and K. K. Parhi, "High-speed VLSI architectures for the AES algorithm", IEEE Transactions on Very Large Scale Integration Systems, vol. 12, issue 9, pp. 95 967, Sep. 2004.

[9] J. Vijaya, M. Rajaram, "High Speed Pipelined AES with Mixcolum Transform "European Journal of Scientific Research" 2011. Vol. 61 No. 2, pp. 255-264.

[10] C. Nandini & B. Shylaja “ Effective Cryptographic Key Generation from Fingerprint using Symmetric Hash Functions”, International Journal of Research and Reviews in Computer Science, Vol 2, No 4, ISSN 2079 - 2557, Aug 2011. Mohammed Tajuddin, C. Nandini, " Cryptographic Key Generation using Retina Biometric Parameter", International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 1, July 2013, ISSN: 2277-3754.

[11] Mohammed Tajuddin, C. Nandini, “More Secured Cryptographic Key Generation through Retinal Biometric using EBI Algorithm

[12] Kai- Shun Lin and Chia-Ling Tsai, “ Retinal Vascular Tree Reconstruction with Anatomical Realism”, IEEE transaction on Biomedical Engineering, Vol 59, No 12, December 2012.

[13] Stallings, W. : Cryptography and Network Security, Prentice Hall, (2010).

[14] B. S Kaliski, Jr RFC 1319: The MD2 Message Digest Algorithm, IETF RFC 1319 Apr 1992.

[15] Ronald Rivest, RFC 1320: The MD4 Message Digest Algorithm, IETF RFC 1320 Apr 1992.

[16] Ronald Rivest, RFC 1321: The MD5 Message Digest Algorithm, IETF RFC 1321 Apr 1992.

[17] NIST FIPS PUB 180-1 Secure Hash Standard, Apr 1995. [18] Mohammed Tajuddin, C. Nandini “Secured Crypto Biometric

system using Retina”, International Advanced Research Journal in Science, Engineering and Technology Vol. 2, Issue 1, January 2015.

[19] Duncan S, Wong, Hector Ho Fuentes and Angnes H “The performance measurement of cryptographic primitives on Palm devices”, 2012 in CCS. NEU. EDU.



A Robustic Technique to Encrypt Medical Images using Bernstien Polynomial over Prime Field

Smitha Sasi1, Santhosh B.2

1,2Assistant Professor Department of Telecommunication Engineering

DayanandaSagar College of Engineering [email protected], [email protected]

Abstract: Medical image processing techniques requires continuous improve quality of services in health care industry. In the real world huge amount of information has to be processed and and transmitted digitally. Confidentiality and authentication to the transmitted medical images is important in health industry. This work proposed an efficient polynomial based public key cryptographic technique for secured transmission of medical image.

Keywords: Medical Image Processing, Bernstien Polynomial, Encryption, Decryption.

1. INTRODUCTION

Today’s world the responsibility of medical institutions is to keeping patients’ medical records in secured manner. The Physician or the institution should not to unveil any therapeutic data uncovered by a patient or found by a doctor regarding the treatment of a patient to any unapproved individual. Exchanging restorative information from a medicinal database focus to another without applying any cryptographic systems means low level of security for patients. Therapeutic data transmission has expanded with the utilization of telemedicine. Telemedicine is imperative on the grounds that it empowers conferences by remote authorities, misfortune free and quick accessibility of individual patient data, and enhanced correspondence between accomplices in a medicinal services framework. So the therapeutic information must be classified while transmission.

Secrecy implies that just the qualified clients have admittance for the data and this can be accomplished utilizing encryption. Cryptography is an effective way of protecting sensitive information as it is stored on media or transmitted through network communication path. Cryptographic algorithms can be implementing on symmetric key or asymmetric key systems. In the symmetric key crypto system same key used for both decryption and encryption. A drawback of symmetric key cryptography is that the two gatherings sending messages to one another must consent to utilize the same private key before they begin transmitting secure data. In people in

general key cryptosystem open key is utilized to encode the information on sender side and private key on the collector side to decode the information. The essential point of interest of open key cryptography is expanded security and accommodation, private keys never should be transmitted or uncovered to anybody. Open key calculation like RSA the fundamental test is the representation of plain content as a whole number. This paper proposes the polynomial based cryptographic method where the plain text represents as points based on polynomial. The mathematical computation is effective in Bernstein polynomial method. So this paper proposes Bernstein polynomial cryptographic technique.

2. BERNSTIEN POLYNOMIAL

In the scientific field of numerical investigation, a Bernstein polynomial, named after Sergei Natanovich Bernstein, is a polynomial in the Bernstein shape, that is a straight mix of Bernstein premise polynomials. A numerically stable approach to assess polynomials in Bernstein structure is de Casteljau's calculation.

The n + 1 Bernstein basis polynomials of degree n are defined

as n(f, t)=� f � ��

�� ncit

i(1-t)n-i

Where nci is a binomial coefficient. The Bernstein basis polynomials of degree n form a basis for the vector space�n of polynomials of degree at most n. The coefficient nci obtained from Pascal’s triangle. The exponent on the (1-t)th term decrease by one as i increases. • The Bernstein polynomials of degree 1 are

B0, 1(t) = 1-t B1, 1(t) = 1-t

can be plotted for 0<t<1 as shown in fig 1

20 | Dayanan

Fig. 1. linear bernstienpolynomai

• Bernstein polynomial of degree 2 are

B0, 2(t) = (1-t)2 B1, 2(t) = 2t(1-t) B2, 2(t) = t2

and can be plotted for 0<t<1 as shown in fig 2

Fig. 2. Quadratic Bernstein polynom

• Bernstein polynomial of degree 3 are

B0, 3(t) = (1-t)3 B1, 3(t) = 3t(1-t)2 B2, 3(t) = 3t2(1-t) B3, 3(t) = t3

and can be plotted for 0<t<1 as shown in

Fig. 3. ternary Bernstein polynomi

Smitha Sasi, Santhosh B.


ail

omial

in fig 3

mial

A. Mathematical model of polynom

Any of the lower-degree Bernsteincan be communicated as a strapolynomials of degree n. Spepolynomial of degree n−1 can be coBernstein polynomials of degree tB

= nciti+1(1-t)(n+1)-(i+1)

=(nci )/(n+1 ci+1 ) Bi+1, n +1(t =i+1/n+1 Bi+1, n +1(t) (1-t)Bi, n(t) = nciti(1-t)n-+1-i =(nci )/(n+1 ci) Bi, n +1(t) =n-i+1/n+1 Bi, n +1(t)

1/ (nci )Bi, n(t)+1/(n+1 ci) Bi+1, t)n—(i+1)

= ti(1-t)n—i-1((1-t)+t) = ti(1-t)n—i-1 =1/ (n-1ci )(Bi, n -1(t))

3. PROPOSED ALGORITHM

The general form of the Bern

t)=� f � ��

�� nci t

i(1-t)n-i

When n=5; this polynomial is quinti The general form to check the givep=n2-n+41. Generate (x, y) points based ion themap those points on to text. Encryption:

Step 1:

Choose (Kpu, kpr) and (�1, �2) are thWhere Ku is the public key and kr is

Step 2:

Choose the secret reference poinquintic polynomial

Step 3:

Choose the plain text point (Px, medical image

Step 4:

Perform (Px, PY) /(�1, Kpu) mod(another point on the curve. (perform

omial of higher degree

ein polynomials (degree < n) straight blend of Bernstein pecifically, any Bernstein composed as a direct mix of tBi, n(t) = ncit

i+1(1-t)n-i

1(t)

1, n (t)= ti(1-t)n—i+ ti+1(1-

ernstein polynomial is n(f,

intic polynomial.

iven number is prime number

the degree of the polynomial.

the pairs of the keys is the private key.

oint on the curve based on

, PY) is the pixel point on

d(n-n+41)=(a, b) this is the rm point division operation)

A Robustic Technique to Encrypt Medical Images using Bernstien Polynomial over Prime Field


Step5:

(a, b)/secret reference point mod(n2-n+41)=(Cx, CY) this is the cipher text. Cipher text will be the point on the curve. So the final resultant curve which contains all the point of cipher text transmit to receiver side.

Decryption:

On receipt of the cipher text points receiver starts perform decryption.

Step 1:

Perform (Cx, Cy) (secret reference point) mod (n2-n+41)=(a1, b1) (secret reference point will be exchanged between two parties by using any secure key exchange algorithm)

Step 2:

Perform (a1, b1)/(�2, Kr) mod (n2-n+41)= (PX, PY) this is the plain text value, if receiver use proper private key. Relation between Kpu and Kpr is Kpr=n[(b-nKu)/(1-n) +b(1-n)/n] mod n2-n+41. The relation between �1, �2 is �2=n[a-n�1)/(1-n)+a(1-n)/n)] mod n2 -n+41.

4. RESULT

Encryption:

(Px, Py)=(4, 43) is the pixel on image (�1, Kpu)=(7, 3) (a, b)=(Px, Py)/(�1, Kpu)mod(n2-n+41)=(4, 43)/(7, 3)mod(n2-n+41) (a, b)=(19, 26) (a, b)/secret reference point=(CX, CY) Secret reference point=(2, 7) (19, 26)/(2, 7) mod(n2-n+41)=(56, 53)

Decryption:

(CX, CY)x(secret reference point)mod(n2-n+41)=(a1, b1) (56, 53) x (2, 7) mod(n2-n+41)=(19, 26) (a1, b1)/(�2, Kpr) mod(n2-n+41)= (PX, PY)

From the relation; Kpr=n[(b-nKpu)/(1-n) +b(1-n)/n] mod n2-n+41. �2=n[a-n�1)/(1-n)+a(1-n)/n)] mod n2 -n+41. (�2, Kr)=(5, 50) (19, 26)/(5, 50) mod(n2-n+41)=(4, 43)

5. CONCLUSION

Bernstein polynomial over prime field based cryptographic approach provides security and reduces computational complexity. This cryptographic method can implement over medical images to protect the patient databases in confidential manner.compared to other public key crypto methods like RSA and ECC, this method reduces the computational complexity, and supports any order of n, higher order n leads to improve security.

6. ACKNOWLEDGEMENT

I would like to express my gratitude to Dr A R Aswatha who motivated us to write this article and also thankful to all faculties of Telecommunication Department for their support.

REFRENCES

[1] H. Caglar and A. N. Akansu, "A Generalized Parametric PR-QMF Design Technique Based on Bernstein Polynomial Approximation," IEEE Transactions on Signal Processing, vol. 41, no. 7, pp. 2314–2321, July 1993.

[2] Online geometric modelling notes: Bernstein; Visualization and graphics research group; department of computer science, University of California

[3] http://mathworld. wolfram.com [4] William Stallings, “Cryptographyand Network Security”,

Principles and Practices, 3rd Edition, Prentice Hall 2003. [5] Interpolation and approximation of polynomials by Philips, G.

M. ISBN: 978-0-387-00215-6http://www. springer.com/978-0-387-00215-6.

[6] The Encyclopedia of design theory: Galois fields by Peter J. Cameron May 30, 2003.

[7] Error Control Coding by Shu Lin, Daniel J Costello; 2nd edition.

[8] D. S. Abdul. Elminaam, H. M. Abdul Kader and M. M. Hadhoud “Performance evaluation of symmetric encryption algorithms” Communications of the IBIMA Volume 8, 2009 ISSN: 1943-7765.

[9] Dr. SudeshJakhar” Comparative analysis between DES and RSA algorithms” IJARCSSE Volume 2, Issue 7, July 2012, pg no. 386-390.

International Journal of Electrical SciencesOnline ISSN: 2455 6068; Volume 1, Issue © Dayananda Sagar College of Engineerin

22 | Dayanan

An OLearnin

Annina Simon1, Mah1,2Prefinal Y

annina.a3,4Professo

selvamvenkDayanan

Abstract: The possibility of this research papattentiveness among upcoming scholars about retechnology, specifically deep learning an arlearning which finds applications in big datartificial intelligence.

Keywords: Machine Learning, Deep Learning, BiIntelligence

1. INTRODUCTION

Machine learning, by its definition, is a fiescience that evolved from studying pattern computational learning theory in artificial intellearning and building of algorithms that can make predictions on data sets. These procedconstruction of a model from example inputs idata-driven predictions or choices rather thanstatic program instructions.

“A computer program is said to learn from exrespect to some task T and some performance performance on T, as measured by P, experience E.” -- Tom Mitchell, Carnegie Mell

So if we want our program to foresee, for eforms at a busy node (task T), we can run it thrlearning process with data about previous (experience E) and, if it has successfully “learndo better at predictingupcoming traffic patternmeasure P).

We need machine learning in the following cas

• Human expertise is absent. E. g. Navigatin

• Humans are unable to explain their expertiRecognition.

• Solution changes with time E. g. Temperat

ces & Engineering (IJESE) ue 1; January 2016 pp. 22-24 ring, Bengaluru-78


Overview of Machine ning and its Applications

ahima Singh Deo2, S. Venkatesan3, D.R. Ramesh Ba

l Year Dept. of Computer Science and Engineering, [email protected], [email protected] ssor, Dept, of Computer Science and Engineering

[email protected], [email protected] nanda Sagar College of Engineering, Bengaluru

paper is to create recent advances in area of machine ata analytics and

Big Data, Artificial

field of computer n recognition and telligence. It is the an learn from and edures operate by s in order to make an following firm

experience E with ce measure P, if its , improves with ellon University.

r example, traffic through a machine s traffic patterns

arned”, it will then erns (performance

cases:-

ting on Mars.

ertise. E. g. Speech

rature Control.

• Solution needs to be adaptedBiometrics.

• Problem size is too vast focapabilities. E. g. Calculating w

Consider the recognition of spokenspeech signal is converted to ASCIIa word may vary from person to page, gender or pronunciation, so approach is to collect a large collefrom diverse people and learn toanother example, consider routinggrid. The trail maximizing the quato destination changes regularly as A learning routing procedure is ablby monitoring the network traffic.

Machine learning involves two type• Supervised machine learning

on a pre-defined set of “trainifacilitate its ability to reach angiven new data.

• Unsupervised machine learnibunch of data and must find therein.

Consider a situation wherein we algorithm to make predictions. Our

Where and are constants. Fwith x as input, there is a correspknown in advance. We compare predictor with the output y and try tin values by altering and. Aftbeen used for training, we are left wNow, if we provide an input whopredictor function will be able to gestimate.

Babu4

ed to particular cases. E. g.

for our limited reasoning g webpage ranks.

en speech, where an acoustic CII text. The pronunciation of person due to differences in so in machine learning, the llection of sample utterances

to plot these to words. As ng packets over a computer uality of service from source as the system traffic changes. able to adapt to the best path

pes of tasks:- ing: The program is “trained” ining examples”, which then an accurate conclusion when

ning: The program is given a nd patterns and relationships

need a machine learning ur predictoris of the form

. For every training example sponding output y, which is

re values obtained from the y to minimize any differences fter multiple examples have

t with the optimized equation. hose value is unknown, the o give us an almost accurate

An Overvie

23 | Dayanan

2. DEEP LEARNING

A new area of machine learning research, wintroduced with the objective of moving mcloser to one of its original goals: Artificial Inte

Deep learning draws its roots from NeocognitrNeuron Network (ANN) introduced by Kunihin 1980. An ANN is an interconnected networunits emulating the network of neurons in thebehind ANN was to develop a learning methothe human brain. However, this method lost machine learning community owing to the factan impractical amount of time as well as a humof data to train the network parameters fapplication. Deep learning is a method to t(and hence the word “deep”) ANN using little reason why ANN is back in the game. Usingcompare Machine Learning with Deep Learnithat if a machine learning algorithm learns pareyes and nose for face detection tasks, aalgorithm will learn extra features like the deyes and the length of the nose. Hence Deepmajor step away from Shallow Learning Algori

The term deep learning gained traction in mid“vanishing gradient problem” responsible reduction in speed was solved in a publicatiHinton and Ruslan Salakhutdinov. They showelayered feed forward neural network couldretrained at a time, treating each layer unsupervised restricted Boltzmann’s machinsupervised back-propagation for fine tuning.

A Deep Neural Network (DNN) is defined toNeural Network (ANN) with at least one hidde

view of Machine Learning and its Applications


, which has been machine learning

Intelligence.

itron; an Artificial nihiko Fukushima ork of processing

he brain. The idea thod by modeling favor within the

act that it required umungous amount s for any decent train multi-layer

tle data. This is the ing an example to rning, we can say parts of a face like a deep learning distance between eep Learning is a orithms.

id 2000s after the for causing a

ation by Geoffrey wed how a multi-ld be effectively in turn as an hine, then using

to be an Artificial dden layer of units

between the input and output layeradded levels of abstraction, thuscapability. The most popular kindsare known as Convolutional NeurConvNets. These area type of feenetwork, extensively used in comindividual neurons are tiled in suchoverlapping regions in the visual fihave also been successfully apprecognition (ASR). Deep Belief NeDeep Belief Networks are some oarchitectures in use.

There are two disadvantages with Dand computation time. Overfittingvery specific details on the trainlayers. As a result, the DNN perforis given as input, but poorly whenThis problem is solved by a regularization where some units arthe hidden layers during trainingcomputations required here are wewe could speed up the computaenormous processing power.

The figure below illustrates howimages can be achieved using a devery layer learns a single feature acan learn the different edges; in slightly more complex features liksuch as ears, noses and eyes. In theven more complex features like thface shapes. The final represenapplications of categorization.

Applications of deep learning are as

• Optical Character Recognitionextracting text from it.

yers. The extra layers give it us enhancing its modelling ds of Deep Learning models, ural Nets (CNN), or simply feed-forward artificial neural computer vision, where the ch a way that they respond to field. In recent times, CNNs

pplied to automatic speech Networks and Convolutional other popular deep learning

DNNs. They are overfitting ng is when the DNN learns ining data using its hidden

forms well if the training data en the input data is different. method called "dropout" are randomly removed from

ing. The matrix and vector well suited for GPUs. Hence, utations by harnessing their

w categorizing of different deep learning model where at a time. At the first layer it the second, it could learn

like different parts of a face the third layer it could learn the distance between eyes or

sentations can be used in

as follows:-

on E.g. Scanning an image an

Annina Simon, Mahima Singh Deo, S. Venkatesan, D.R. Ramesh Babu


• Speech Recognition E. g. Generating textual representation of speech from a sound clip.

• Artificial Intelligence E. g. Robotic Surgery

• Automotive Applications E. g. Self-Driving Cars

• Military and Surveillance E. g. Drones

3. DEEP LEARNING IN BIG DATA

Deep Learning and Big Data are two high-focus areas of data science. Deep learning algorithms extract complex data patterns, through a hierarchical learning process by analyzing and learning massive amounts of unsupervised data (Big Data). This makes it an extremely valuable tool for Big Data Analysers.

Big Data has 4 important characteristics, namely, Volume, Variety, Velocity and Veracity. They are Learning algorithms are mainly concerned with issues related to Volume and Variety. Deep Learning algorithms deal with massive amounts of data, i. e. Volume whereas shallow learning algorithms fail to understand complex data patterns which are inevitably present in large data sets. Moreover, Deep Learning deals with analyzing raw data presented in different formats from different sources, i. e. Variety in Big Data. This minimizes the need for input from human experts to retrieve features from all new data typesfound in Big Data.

Semantic Indexing, Data Tagging and Fast Information Retrieval are the main objectives of Deep Learning in Big Data. Consider data that is unstructured and unorganized. Haphazard storage of massive amounts of data cannot be used as a source of knowledge because looking through such data for specific topics of interest and retrieving all relevant and related information would be a tedious task. Using Semantic Indexing and Data Tagging, we identify patterns in the relationships between terms and concepts based on the principle that words used in the same context have similar meanings. The related words can then be stored close to each other in the memory. This helps us present data in a more comprehensive manner and helps in improving efficiency. A direct result of such a form of storage would be that search engines would work more quickly and efficiently.

4. DEEP LEARNING IN ARTIFICIAL INTELLIGENCE

Artificial Intelligence is the theory and development of computers which are capable of performing tasks which humans can. Deep learning represents the rudimentary level of attempts towards achieving this task. It is utilized in visual perception, speech recognition, game playing, expert systems, decision-making, medicine, aviation and translation between languages.

In the gaming industry, Artificial Intelligence could be useful as we could have a ‘gamebot’ stand as an opponent when a human player is not available. We could also have deep learning algorithms suggest how enemy spawns could be strategically placed in the arena to obtain different levels of difficulty. The military as well as aviation industries can use Artificial intelligence to sort information related to air traffic and then provide their pilots with the best techniques to avoid the traffic. A medical clinic can use Artificial Intelligence systems to organize bed schedules, staff rotations and provide medical information.

5. CONCLUSION

Deep learning techniques have been criticized because there is no way of representing causal relationships (such as between diseases and their symptoms), and the algorithms fail to acquire abstract ideas like “sibling” or “identical to.” Not much theory is available for most of the methods which is disadvantageous to beginners. Deep Learning is only a small step towards building machines which have human-like intelligence. Further advancements must be made in order to achieve our ultimate goal. Organizations like Google, Facebook, Microsoft and Baidu (a Chinese search engine) are buying into this technology and exploring various avenues available. For example, Facebook is using deep learning to automatically tag uploaded pictures. Google’s Deep Mind focusses on exploring new techniques in this area. Recent trends show that the interest in machine learning has only been growing with time and has sparked an interest in countries like India and Singapore. Thus it has emerged as one of the most promising fields of technology in recent times.

6. REFERENCES

[1] D. Bouchaffra, F. ykhef in “mathematical model for machine learning and pattern recognition”.

[2] Itamar Arel, Derek C. Rose and Thomas P Karnowski in” Deep Learning – A New Frontier in Artificial Intelligence Research”.

[3] Alexander J. Stimpson and Mary L. Cummings in “Accessing Intervention Timing in Computer-Based Education using Machine Learning Algorithms”.

[4] Li Deng, Geoffrey Hinton and Brian Kingsbury Microsoft Research, Redmond, WA, USA in “New Types of Deep Learning for Speech Recognition and Related Applications: An overview”.

[5] Maryam M Najafabadi, Flavio Villanustre, Taghi M Khoshgoftaar, Naeem Seliya Wald and Edin Muharemagic in “journal of big data”.

[6] Deep learning by Nando de Freitas [7] An Introduction to Machine Learning Theory And Its

Applications: A Visual Tutorial with Examples by Nick Mccrea

[8] A Deep Learning Tutorial: From Perceptrons To Deep Networks



Hadoop Framework for Flow Analysis and Congestion Control of the Big Data

Rakshitha Kiran P.1, Jeffrey Shafer2 1Assistant Professor, MCA (VTU) Department

Dayananda Sagar College of Engineering, Bangalore [email protected] 2University of California

[email protected]

Abstract: Flow categorization of the Computer Network traffic displays the sequence and pattern of the traffic in the network. This helps the network administrator to monitor the operations going on in the network, to understand the network usage and to examine the behavior of the user using the network. Tapping of the internet traffic can avoid a huge amount of problems. Flow analysis helps in fault tolerance, traffic engineering, resource allocation and network capacity planning. Due to fast growing network, the volume of the traffic is getting very big day by day. So it is very difficult to collect, store and analyze this huge data on a single machine. Hadoop is a leading framework which is designed to execute tremendous datasets that can be hundreds of terabytes and even petabytes of data. Hadoop performs brute force scan for multiple traces of input data and produces the output for traffic flow identification, flow clustering. In this paper a Hadoop based traffic analysis of internet traffic is done. Once the analysis is done flow control mechanisms are carried out to avoid any kind of congestion in flow.

Keywords: Hadoop, bigdata;

1. INTRODUCTION

Data turns into big data when its volume, variety or velocity exceeds the potential of the storage systems to fetch, gather, store analyze and finally process the data. In many organizations have necessary equipments to handle this large amount of data. The data that is produced by various streams can be structured or unstructured so these systems lack the capacity to mine the data. It’s not just about the volume of the data that is produced every day, but its also about the speed of the data in which the data arrives. For example, online social networking site facebook produces every day 2. 5 billion pieces of content and 500+terabytes of data, similarly data from twitter, weather forecast, call detail reports etc produces very huge amount of data every day.

Hadoop is a emerging framework performs distributed processing of the large data, big data, across big clusters of computer. . Hadoop is designed to process tremendous

datasets and performs brute force scan on the multiple traces of the data and produces a efficient output.

Flow analysis of the traffic helps us to elucidate the pattern and the sequence of the traffic in the network. This information is needed for the network controller to monitor the various operations going in the network. Network controller understands the type of network, its usage, the behavior of the user using the network, monitors the network traffic and bottleneck.

In this paper we talk about the flow analysis and congestion control mechanism for the big data by making use of Hadoop framework. Hadoop performs Map and reduce techniques over the large datasets and gives a summarized output. The output is then analyzed for congestion.

2. LITERATURE SURVEY

The literature survey makes a very significant part in the research process. It is a place from where research ideas are drawn and developed into concepts and finally theories. In the paper The State of Enterprise Network [1] the author tells about the analysis done on the passive and active techniques that were used to measure the state of the communication using Internet. TCP is a major protocol used on the internet traffic but nowadays UDP has risen in the recent years. The author in the paper Traffic Classification on the fly [2] proposes a technique for analysing the traffic which are associated with the TCP connection. The detection of application using TCP flows is an important step in network security. the author proposes a method where the observation lies on the output of first five packets which are associated with the TCP connection to identify application. Based on the observation the author proposes a traffic classification mechanism which has 2 phases: First, Learning Phase and second is traffic classification phase. In the paper Network Traffic Characteristics of Data Centres in the Wild[3] the author analyzes data sets from 10 different data centers which

Jeffrey Shafer, Rakshitha Kiran P.


includes 5 commercial cloud based data centres, 2 private enterprise and 3 university campus data centers. The author collects and analyzes SNMP statistics, packet-level traces and topology. Examination is done on the basis the range of applications that deployed in those data centers, their placements, their flow level and also packet level transmissions of those applications. In the paper The Hadoop Distributed Filesystem: Balancing Portability and Performance [4] the author tells about the Hadoop framework. Hadoop is a popular framework which is open source which implements MapReduce and HDFS for the analysis of large data sets. HDFS is distributed file system which is used for storing purpose and MapReduce is processing technique used to process the large data sets. In the paper Toward Scalable Internet Traffic Measurement and Analysis with Hadoop [5] author tells traffic measurement and analysis of scalable Internet is difficult job because of large amount of data set is required for the computation and the storage resources. Traffic measurement and analysis is used to find out network usage and user behaviour. But because of very high growth of traffic and very high speed it’s very difficult to analyse them. Hadoop is popular framework which is also open source is used for this traffic management and analysis. It makes use of HDFS and MapReduce techniques for massive data analysis because it supports scalable processing of data and storage of data.

3. HADOOP FRAMEWORK

In the traditional framework an organization had a system which would do both storing and processing of the big data. In this framework the data would be stored in Relation Database Management System(RDBMS) like MS SQL Server or Orcale Database system or any sophisticated softwares. These softwares were required to interact with the database. This approach works good when data is of less volume. But when it comes to dealing with large data it will be very difficult. So to overcome this drawback we go for Hadoop framework.

Fig. 1. Hadoop framework

Hadoop makes use of MapReduce algorithm where the data will be processed in parallel on various CPU. In this way hadoop framework is capable of runnig over multiple clusters and thus gives statistical analysis for huge amount of data.

4. PROPOSED METHODOLOGY

The proposed methodology consist of network where the large amount of packets flow, a system to capture the flow of packets, hadoop cluster to perform analysis. .

Packet flow:

Hadoop framework is used for flow analysis and congestion control for big data. Initially we will start capturing a large data from large network. The packet flow from this network will be captured in the form of text file. This packet flow is given as input to the Hadoop Cluster. Hadoop Cluster will accept the internet traffic flow and performs analyzes using by making use of 2 main functions called Map and Reduce. Hadoop cluster then generates an output file which is in the form of text file. The output obtained is then monitored for any congestion, if found flow control mechanism is done.

Flow Analysis:

Define Based on the input given to the hadoop cluster an output file is generated. The input file is a text file that consist of the various information like source address, destination address, protocol, length of the packet, information about the packet. This input file will be given to the hadoop cluster for analysis. The hadoop cluster will then perform Map and Reduce on the input file. Here the mapping is done based on the source ip address and destination ip address. The packets from same source ip address to same destination ip address are mapped together and finally reduce is applied. The summarized detailed of the packet are then stored in a separate output text file. The output file generated will have a detailed analysis of the packets that are exchanged between the hosts.

Congestion Control:

Fig. 2. flow diagram of system

Hadoop Framework for Flow Analysis and Congestion Control of the Big Data


Every path from the source to destination host is provided with some priority. The output text file generated will have details about amount of data passed from one particular source to destination. If the bytes of data have crossed the threshold then priority of that particular route will be updated and the packets will be forced to take up other congestion free path. Fig below shows flow diagram of the system.

5. EXPERIMENTAL RESULTS

Experiments were conducted by making use of 2 host and 4 nodes. The packets would travel from one host to another host and vice versa. Packet flow was captured from both the hosts. The packet information is in the form of text file.

Figure below shows the screenshot of the network and packet information that are exchanged between the nodes.

Fig. 3. screenshot of network topology created

Fig. 4. screenshot of packet information

The packet information collected will be given for the Hadoop cluster for the analysis. Hadoop cluster performs analysis based on Map Reduce technique. The input file i. e. the text file is stored in the separate folder called input. We need to specify the path of input file for the hadoop cluster. Fig below shows screenshot of the hadoop performing Map Reduce technique and then generating the input file.

Fig. 5. Screenshot of Hadoop performing MapReduce

Fig. 6. Screenshot of output file generated

The output text file generated by the hadoop cluster is shown above which is named as part-00000. This output file consists of the summary of the flow of the packet information collected above. Here the packets from same source to same destination are grouped up and the also the byte information of the packets are added. The screenshot below shows the analyzing the output file and updating the flow.

Jeffrey Shafer, Rakshitha Kiran P.


Fig. 7. Screenshot of analyzing output file and updating flow

Fig. 8. Screenshot of updated flow

6. CONCLUSION

This paper gives the work on flow analysis and congestion control on the Hadoop platform. Hadoop is a popular framework which can process a huge amount of data just in few seconds. Here we have provided a detailed analysis on the classification of the packets based on the address, protocol and its bytes of the packets. In this system we show a flow control mechanism applied on the packets flowing in the network where the large amount of packets flow, a system to capture the flow of packets, hadoop cluster to perform analysis and also to avoid congestion.

REFERENCES

[1] M. Yu, L. Jose, and R. Miao,” Software defined trafficmeasurement with opensketch,” in Proceedings 10th USENIX Symposium on Networked Systems Design and Implementation NSDI, vol, 13, 2013.

[2] Scscc J. Shafer, S. Rixner, and Alan L. Cox, “The Hadoop Distribution Filesystem: Balancing Portability and Performance”, in Proceedings of the 10th ACM SIGCOMM conference on Internet measurement ACM 2010.

[3] T. Benson, A. Akella, and D. A. Maltz, “Network traffic characteristics of data centers in the wild,” in Proceedings of the 10th ACM. SIGCOMM conference on Internet measurement. ACM, 2010, pp. 267–280.

[4] A. W. Moore and K. Papagiannaki,” Toward the accurate identification of network applications,” in Passive and Active network Measurement. Springer, 2005, pp. 41-54.

[5] L. Bernaille, R. Teixeira, I. Akodkenou, A. Soule, and K. Salamatian, “Traffic classification on the fly”, ACM SIGCOMM Computer Communication Review, vol 36, no. 2, pp. 23-26, 2006.

[6] Yuanjun Cai, Min Luo, “Flow Identification and Characteristics Mining from Internet Traffic using Hadoop” in 978-1-4799-4383-8/14/ at IEEE 2014.

[7] Apache Hadoop Website, http://hadoop. apache. org/



Intelligent Street Lighting System (extended version of automatic illuminators)

Sanskar Gupta1, Sonika Soni2 1Department of Information Science & Engineering,

Dayanada Sagar College of engineering, Bengaluru, India [email protected]

2Department of Information Science & Engineering, Dayanada Sagar College of Engineering, Bengaluru, India

01sonika. [email protected]

Abstract: Project deals with extended version of automatic street lights. This public street lighting system is quite different compared to traditional illumination system as it enables light to travel along with you, actually a bit ahead of you. This system is more ‘INTELLIGENT’ than the already existing systems as citizen’s safety has been taken into the consideration. We also referred it to as 'extended' as we have included a safety measures into it, which will provide a safe and secure environment for pedestrians, bicyclists and motorist during evening, night time and early morning hours when the possibilities of crime is at its peak. The ‘energy efficient’ attribute of the system is enhanced by making use of SOLAR PANELS to further decrease Energy consumption.

Keywords: plug and play, intelligent system, security mechanism, solar panel, energy efficient, camera.

1. INTRODUCTION

The main objective of the project is to provide Automatic street lighting system with enhanced power saving capability, accompanied with security system for citizens. The system proposed adapts the technique of ‘plug and play’ i. e. We are making use of an electronic kit that fits into existing street lights and turns them into intelligent street lights. The existing street lights need not be replaced by new ones rather just modified by plugging in an electronic kit which will facilitate the required functioning. Power sources are getting diminished and so the power saving is a major consideration. We want to save power automatically rather than doing it manually.

The main functions of the system is-

• The lights dims when no one is in vicinity and glows to full mode as soon as passerby comes within range, and thus, saves a lot of energy.

The street lights illuminates automatically during dark hours if there is some motion detected in the road. But during day hours, even if the motion is detected, street lights remains OFF. A timer circuit is used which provides time delay

proportional to the time constant of the 555 timer so the lamp will glow only for certain duration of time.

Our proposed system also includes security mechanism which becomes active through a switch and danger alert is send to the nearest police station or tolls (as per the availability) and Moreover, it turns ON all the street lights of that particular area. It is also accompanied by a camera for further safety purposes.

2. PAST AND FUTURE

A. History

The concept of intelligent street lighting was initially introduced in 1990s. On April 7, 2006, the first large scale implementation of street lighting system took place in Oslo (Norway) in Europe. The expected results were- reduced consumption of energy by 50 percent, improved roadway safety and minimized maintenance cost.

This Oslo project influenced other cities in Europe, thus paving way for other sustainability initiatives like E-Street initiative.

B. Future Aspect

1) Damage detection – signal can be generated by individual lamp post when maintenance is required as the lamp posts are capable of communicating with each other.

2) Traffic speed sensors - traffic speeds can be managed through dimming of street lights. During evening or night hours, fast average traffic speed could trigger dimmed street lights. This dimming effect would slowdown the speed of motorist in response, eventhough dimming level would be imperceptible to motorists.

30 | Dayanan

3. INTERNAL ARCHITECTURE

A. LDR (light dependent resistor)

LDRs or Light Dependent Resistors are circuits. It is also referred to as photoconductoor photo-conductive cell. Generally LDRresistance of up to the 10, 00, 000 ohms, butdramatically if illuminated. They are capable electrical characteristics when comes in conrays, thus they are also considered as one ofonto sensors. LDRs find application in eldesigns, such as flame or smoke detector, burglights and others, because of following featrugged features and simple structure.

Fig. 1. LDR

B. Block Diagram

Fig. 2. Block diagram of Intelligent stre

• Switching circuit using LDR- this blocktransistors, resistors, LDR

• The LDR is capable of sensing visible ligh

• When no light falls on the LDR, the swsends a HIGH signal. When light falls sends LOW signal.

• Pressure sensor- second block basicaworking of sensor according to the outpblock. The sensor used here is a pressure s

Switch-ing circuit using LDR

Pressure sensor

555 timer



used in sensor ctor, photoresistor DRs have high but the value falls

of altering their contact with light

of the Electronic electronic circuit

urglar alarm, street eatures- low cost,

treet light

ck is made up of

ight.

switching circuit on the LDR it

ically deals with utput of previous sensor.

• Timer circuit - the third block,to pressure sensed by sensor. the 555 timer which is operated

• Lamp- the fourth block i. e. theto the output terminal of the 55high output is obtained as nopressure sensed on pressure setimer which is operated in monlamp will glow for some duratime constant of 555 timer.

C. Pressure Sensor

Fig. 3. pressure

As demonstrated in the fig3, A anplates which are connected via twois grounded and plate A is connectforce is applied on plate A, it comhence triggering 555 timer. Whewithdrawn, then two plates are drestores back to its own position.

D. Security Mechanism

Fig. 4. Block diagram of Se

The main key feature of our securiembedded at certain locations onsame path as that of the street lights

Lamp

Metallic Plate A

Metallic Plate B

ck, gets triggered in response Timer circuit comprises of

ted in mono-stable mode.

the street lamps are connected 555 timer. During night time no light falls on LDR. Any sensor now will trigger 555 ono-stable mode. As a result, ration of time depending on

re sensor

and B are the two metallic wo insulated springs. Plate B cted to 555 timer. Whenever

omes in contact with plate B hen force from plate B is disconnected i. e. plate B

Security mechanism

urity mechanism is the alarm on the street, following the hts.

Intelligent Street Lighting System


• The button in the system, when pressed, activates alarm located at help centers which is the indication that ‘help’ is required at that specific location.

• As alarm activates, simultaneously all the street lights of that particular area (within specified range) are switched ON in order to alert the people nearby of the help required.

• Parallel to the ongoing process, camera provided at the location where alarm is pressed, starts recording the current situation for later reference if required.

• Solar panel acts as a backbone for our security mechanism. In case of power failure, solar energy collected through the panels acts as a backup, to ensure that people’s safety is not compromised in any case.

4. A BETTER ALTERNATIVE TO STREET LIGHTS

A. Need

Illumination of city's street during dark hours is one of the important part of city's infrastructure. One major challenges with public lighting is that they keep on working all night even no one is around. This is the big waste of energy.

This is the problem we are trying to solve. So we need an Intelligent street light technology - a system that dims when no activity is detected but brightens when movement is detected. It automatically senses motion in the road and visible light. Use of this system helps to save energy by switching ON the street lights as and when required and it also reduces manual work at most up to 100 percent. This Technology helps in reducing Light Pollution as well.

Further, the safety mechanism involved will lead to decrease in crime rate. Safety buttons embedded on specific locations on street will allow people to send a signal for help directly to the nearest police station. Thus ensuring quick help and thus security of the people.

B. Advantages

1) There is lighting only where needed thus decreasing energy consumption which results in faster payback or cost recovery.

2) Safety mechanism included, ensures safety and security of people.

3) Reduction of light pollution.

4) Dimming capability allowing to dim each light individually as activity level decreases on street.

5) Reduction in CO2 emissions: with reduction in energy consumption, CO2 emission is also reduced.

6) Elimination of manpower to higher extents.

7) Maintenance of safety: since the lights are dimmed and not turned off completely, safety is maintained.

8) Plays a part in sustainable development.

C Analysis

Fig. 5. Energy consumption index of various sources of light

The above graph represents the energy consumption index.

• Analysis of graph indicates that in comparison to conventional lamps, HSP (High pressure sodium) lamps consumes 25- 35% less energy.

• LED (Light emitting diode) lamps stands out both the above mentioned sources as 50-60% more energy is saved in comparison when HSP is used and the count increases to 70-80% in comparison to the conventional lamps.

D. Drawbacks and its solution

• Drawbacks of this project are –

o Initial cost

o Maintenance

• Initial cost would not be a problem if the long term benefits like faster payback is considered.

Cost of the project can be minimized through good resource planning and advancement in technology.

0

20

40

60

80

100

120

Conventional

lamp

HSP LED

%Energy consumed



5. CONCLUSION

This project is a cost effective as the components used such as LDR, Pressure Sensor etc. are of low cost.

Approximately 80% of energy will be saved by implementing this project, thus making it ecofriendly.

This project is also effective in tackling one of the major problem faced in today’s world scenario – saving of energy. Other than street lights, the project can find application in parks, industries, campuses, parking lots etc.

Thus, ‘Intelligent Street Lighting System’ lends a hand in lighting the way to Safe Street, healthier environment and better bottom line.

REFERENCES

[1] Sharath Patil G. S et al. Int. Journal of Engineering Research and Applications ISSN: 2248-9622, Vol. 5, Issue 6, (Part - 1) June 2015, pp. 97-100

[2] Y. Chen and Z. Liu, “Distributed intelligent city street lamp monitoring and control system based on wireless communication chip nRF401,” in Proc. Int. Conf. Netw. Security, Wireless Commun. TrustedComput. , Apr. 25–26, 2009, vol. 2, pp. 278–281.

[3] W. Yongqing, H. Chuncheng, Z. Suoliang, H. Yali, and W. Hong, “Design of solar LED street lamp automatic control circuit,” in Proc. Int. Conf. Energy Environment Technol. , Oct. 16–18, 2009, vol. 1, pp. 90–93

[4] M. A. D. Costa, G. H. Costa, A. S. dos Santos, L. Schuch, and J. R. Pinheiro, “A high efficiency autonomous street lighting system based on solar energy and LEDs,” in Proc. Power Electron. Conf. , Brazil, Oct. 1, 2009

[5] International Journal of Engineering and Advanced Technology (IJEAT) ISSN: 2249 – 8958, Volume-3, Issue-4, April 2014.



A Secured Protocol for Efficient Discovery of Service in Vehicular Networks

Shalini K. B.1, Anil Sagar T.2, S. Venkatesan3, Darshan H. Yogendra4

1Assistant Professor, Department of Information Science, DSCE shalinikb. [email protected]

2Assistant Professor, Department of Computer Science, DSCE [email protected]

3Professor, Department of Computer Science DSCE [email protected]

4University of Limerick, Ireland [email protected]

Abstract: Wireless access in vehicular environments (WAVE) is a foundation for a broad range of intelligent applications in transport, related to safety, comfort and, traffic management. Opportunities to advertise and disseminate services arise with the technological advances of vehicular networks. Points of interest (POI) to drivers and passengers, e. g. , shops and gas stations, broadcast the advertisements with local services and facilities available to the nearby vehicles by exploiting the road-side unit (RSU) infrastructure and comfort applications. However, RSU access coverage limitation or packet losses can prevent users (drivers or passengers) who are genuinely interested in services from finding them, even with a short distance from the POI. To improve the service discovery, we propose PEDS, a protocol based on opportunistic vehicle contact and a store and forward technique to discover and advertise services in vehicular networks. PEDS (Protocol for efficient discovery of service) is a lightweight and alternative protocol intended for location-aware applications. It supports distributed services fully independent from Internet access. In this paper we describe the key design concepts of PEDS and demonstrate its feasibility by a concept proof testbed. We perform two experiments evaluating the success rate of: i) capturing and storing service messages and ii) searching for a service of interest and receiving reply messages containing it. PEDS feasibility was demonstrated for different vehicle densities and for very short time connections. Results indicate that a vehicle can capture 100% of messages under the coverage of seven POI sending advertisement messages every second, with a contact time of 1. 2s. Moreover, a user can find a service with 95% of success rate by querying one neighbor vehicle.

1. INTRODUCTION

Vehicular networks are promising technologies to enhance safety, convenience, and entertainment for users under different urban and road traffic conditions. Vehicular networks have been defined by the wireless access in vehicular environments (WAVE) standard, which establishes communications channels to non-safety applications for the

road side unit (RSU) and vehicles which are main source for a transport [1], [2].

This creates significant opportunities for the deployment of a wide diversity of applications and services to vehicular environments, especially location-aware services. In this context, protocols for service discovery become an important requirement for vehicular networks. Advances in this field are concentrated in mobile ad hoc network (MANET) concept. However, most MANETs do not take into account the particularities of vehicular networks, such as, highly dynamic network topology changes and short connectivity time, due to fast movement of vehicles, inability to rely on an assumed geographic position of vehicles, and the wide exploitation in the vehicular environment of services dissemination. Therefore WAVE and the ability to tolerate communications failures must be considered when designing network protocols. Different studies on service discovery protocols for vehicular networks have been conducted. Dikaiakos et al. [3] provided time-sensitive information about the traffic conditions and the available roadside services by means of car-to-car communication. Their protocol, however, is not scalable for increasing network density and number of requests. Boukerche et al. [4] proposed a location-based service discovery protocol that is able to discover location-aware and time sensitive services based on the specified location of the requested service in the drivers request. However, failures in service queries significantly affect the protocol performance. In a recent study, Abrougui et al. [5] introduced a fault-tolerant scheme for infrastructure-and-location-based service discovery by integrating service information into the Network Layer and using diverse channels. However, the solution requires two wireless interfaces per vehicle. To improve services discovery in vehicular networks, we propose an opportunistic service discovery protocol (PEDS). Our protocol is different from

Shalini K. B., Anil Sagar T., S. Venkatesan, Darchan H. Yogendra


other protocols in several aspects. First, we designed the protocol taking into account the WAVE standard. Second, PEDS is a layer-2 protocol that runs on dedicated short-range communication (DSRC) [6], whereas previous approaches are either application solutions at layer-7 [3] or layer-3 protocols [4] [5]. Finally, other studies rely on Internet access or on the RSU interconnection to exchange data, whereas PEDS is based on the opportunistic contact and store-and-forward technique, as in [7]. Thus, PEDS can handle RSU or Internet access disruptions. We define opportunistic communications as being occasional communications among vehicles. No assumption is made regarding the existence of a complete protocol stack for infrastructured communications between vehicles. POI services and the destination vehicle might never be connected to the same network at the same moment. Store- and-forward denotes the fact that the service advertisements are buffered in the vehicle to wait for contact from other vehicles searching for services. When a required service matches to the stored one, service information is forwarded. The PEDS is a lightweight protocol at layer-2 that focuses on providing location-aware services for moving vehicles. Instead of users using their smart phone with 3G/4G connections to find the desired services via the Web, PEDS will be integrated into the vehicle system and network, interacting with the users through an interface (e. g. , car’s navigation system) to search for the desired services. The returned information (e. g. , location and price-lists of gas stations, location and menus of restaurants) can help users who are interested in locating nearby roadside services. Furthermore, this information can be merged with GPS position, extending on-board navigation systems and Internet applications functionalities. We implemented and evaluated PEDS to prove, individually, the accomplishment of vehicle-to-RSU (V2R) and vehicle-to-vehicle (V2V) communications. To this end, first, for V2R, we increased the number of services offered to vehicles from 1 to 7. Second, for V2V, we increased the number of neighbour vehicles of a particular vehicle from 1 to 7. According to the results, PEDS was feasible for long and very short time connections and also for different densities of vehicles and RSUs. The remaining of this paper are organized as follows. In Section II we describe a base scenario and provide PEDS definitions. In Section III we detail the PEDS architecture. Section IV presents the testbed, implementation and validation. In Section V we discuss the performance evaluation. Finally, the conclusions are in Section VI.

2. SCENARIO AND DEFINITIONS

A. Envisioned Scenario

Assume an urban-road scenario in which the vehicle network infrastructure is somehow deployed, such that the RSUs and vehicles are equipped with built-in wireless interfaces, minimum processing and memory capabilities, and GPS devices. Security and privacy issues in vehicular networks are

beyond the scope of this paper and will be addressed in future work. For more details about this subject, we refer to the reader to [8]–[12].

We distinguish safety and non-safety services. Safety services will be provided in high-priority channel to enable communication in critical situations, e. g. , finding a hospital in an emergency. Non-safety services will be provided in low-priority channel to enhance users’ experience during their trip. PEDS is intended for non-safety services as an alternative for users. Therefore, users sometimes can experience non-satisfactory or failed requests.

The PEDS can handle the types of services shown in Fig. 1. Roadside services periodically broadcast messages by using beacons with their local information, e. g. , RSU1 broadcasts price-lists of gas stations; RSU2 broadcasts hotel room prices; RSU3 broadcasts parking lot rate. Vehicles around the RSU can hear these beacons and store the information received. Suppose the user of vehicle v2 is interested in receiving information about gas stations. Thus, v2’s PEDS checks its stored data and queries to the other neighbor vehicle, v1, to gather more information. Vehicle v1 replies to v2 the stored information received from RSU1, when it passed by the gas station. Similarly, vehicle v1 received and stored the hotel and parking lot information when it passed by RSU2 and RSU3, respectively. Vehicle v3 is looking for a bookstore, but its does not find one. This is because v4 has not passed by any bookstore. However, vehicle v3 can continue to query any vehicles it meets along the way (e. g. , when meeting v2) until finding the desired service.

Fig. 1. Future scenario in a vehicle network. Vehicles {v1. . . v4} are hearing and storing beacons broadcasted by {RSU1. . . RSU3}

and opportunistically store-and-forward services information.



B. System definitions

The entities in the envisioned scenario are as follows:

• Vehicle: A vehicle v (car, bus, motorcycle or truck) is equipped with built-in devices required for a vehicular network, at least, wireless interface, GPS and an onboard computer. Thus, all vehicles v that implement PEDS are able to receive the service information broadcast by RSU, and store it in a cache c. Moreover, v can query the cache c of neighbour vehicles within the range of its radio frequency (RF).

• RSU: Road-side units are placed along the streets, roads, parks and other such locations. Every T time, they broadcast to v their beacons carrying the local service information offered by the points of interest (POI) in the geographic region.

• Service: Services are characterized as a POI for users. A service s is defined by a 4-tuple _N, P, E, D_, where N is the general service name (e. g. , drugstore, bookstore or gas station), P is the geographic position with the pair latitude-longitude, E is a time-to-live in seconds in which the service expires, and D is an additional description text about the offered service (e. g. , address, business name, and phone number).

Therefore, the problem is defined as follows:

Let RSUk broadcasts services sk = (Nk, Pk, Ek, Dk) and V = {v1. . . vj} the set of vehicles that hear and store sk in their cache cj. When a vehicle vi needs to find sk and sk /∈

ci, vi sends a query qi = _N, D_ to the neighbour vehicles in

V within vi’s range of RF coverage, then {∀vj |sk ∈ cj}, vj replies rj = {s1. . . sn} to vi. For each different sk replied, vi updates its ci.

Hypothesis: sk is broadcasted by RSUk and heard by V. If a vehicle vi does not pass by the RF coverage of RSUk, vi can find sk when vi meets at least one vehicle of V.

3. OPPORTUNISTIC SERVICE DISCOVERY PROTOCOL

In this section we describe the PEDS messages, phases and procedures. First, we define the three types of messages used by PEDS and discuss the trade-off when adopting extensible mark-up language (XML) as a service data format. Second, we describe the PEDS phases. Finally, we explain the protocol procedures in the RSU and vehicle nodes.

A. Types of messages

PEDS is a beaconing-based communication protocol. Beaconing consists of the periodic single-hop broadcast transmission of messages, so-called beacons [13], [14], handled in the physical layer and media access control layer of WAVE. Three types of messages are defined in PEDS:

• M is a beacon with the service s, which is broadcasted by RSU every T time;

• Q is a query message that contains the name of service N and description D. To find a desired service, a vehicle vi sends a query qi to its neighbour vehicles V ;

• R consists of the messages rj, which neighbours V send to vi replying the services {s1. . . sn} found in the cache cj. R is a unicast probe response message.

Every single message has the fields illustrated in Fig. 2. TYPE is an unsigned integer to distinguish the messages types. LENGTH is an unsigned integer to set the size of XMLDATA. In the end, XMLDATA is an XML document.

Fig. 2. Message encapsulated in the beacon and probe frames of WAVE.

It is important to note that these messages will be used by a beacon or a probe with other ordinary beacon/probe fields, according to the frame of the WAVE standard.

The XMLDATA should have the hierarchy of <service> and <description> tags. Example of a service advertisement s in a message M is shown in Code 1. The <service> hierarchy has the mandatory information set _N, P, E_, represented by the attribute tags <name>, <latitude>, <longitude> and <timelife>. The <description> hierarchy is optional. It has the attributes to detail the service, such as <url>, <phone> and <address>. These attributes can be selected to provide the best service detail. The R response message follows the same XML structure of the message M.

<?xml version=" 1. 0 " encoding="UTF−8"?> <service> <name>Un i v e r s i t y < /name> <l a t i t u d e>−22. 001954< / l a t i t u d e> <longitude>−47. 931717< / longitude> < t ime l i f e >1382449758< / t ime l i f e > </ service> <description> <label> Universidade de São Paulo < / l a b e l> <u r l> www. usp. br </ u r l>



<phone> +55−16−3373−9700 < / phone> <address> Av. Trabalhador sao−carlense, 400 </ address> </ d e s c r i p t i o n>

Code 1. Example of a XML document carried by an M message.

For Q messages, the XML document also has both <service> and <description> with simpler attributes, only <name> under <service> is mandatory. The optional attributes in the <description> is intended to limit the search and reduce the number of replied messages R. For instance, an attribute <max_distance> could specify the maximum distance from V to the target service s, which is tolerable by vi.

Vehicle systems and databases will probably contain data in incompatible formats. Many types of services will be offered and each one has different information and formats. Therefore, PEDS addresses the data sharing semantic by using XML documents, which allows software-hardware independent methods of storing data.

B. PEDS phases

PEDS runs in independent phases, as illustrated in Fig. 3. i. Beaconing: Each interval T, RSUs periodically broadcast

the beacons M that conveys the services {s1. . . sn}.

ii. Caching: Vehicle vj passes by the RSU, receives the beacon M, and stores the service s. If s was already stored in the cache cj, then M is dropped. Otherwise, cj is updated, also expired s are removed.

iii. Querying: A query occurs when a user wants to find a service s. Thus, vi broadcasts a query Q to the neighbor vehicles V. All vj that received Q checks in their caches cj to determine if s-related services exists. If at least one service is found, then vj sends a probe response R as a unicast probe to vi with the services {s1. . . sn}. Then, vi updates its ci with the received services {s1. . . sn}.

Due multi-hop beacons can degrade the vehicular network performance [15], for all the phases, the messages M, Q and R are sent in a single hop.

Furthermore, PEDS adopts the probe response R as a unicast message, which is sent to a particular vehicle vi. Multiple vehicles in V might reply with different responses for the queries Q at the same time. In this sense, the unicast message allows distinguishing the replies R and enables the users of vi to choose suitable services.

C. Operation of the protocol

PEDS operates in five states in a vehicle v, as shown in

Fig. 4, it starts listening to the messages M, Q and R or waiting for users searching in the Listen state. According to the type of the received message, PEDS goes to different states. If the type is M or R, PEDS goes to the Update state. In the update state, it processes the message updating the cache c and then returns to the listen state. If the type is Q, the PEDS goes to the Find state to process the query Q by searching in the cache c for the services s. If one or more services are found, PEDS goes to the reply state; otherwise, it returns to the listen state. In the Reply state, the services s are conveyed in the unicast messages R and then it returns to the listen state. When the user wants to find a service, PEDS goes to the Query state and sends the query Q.

Fig. 3. Beaconing, RSU broadcast the service. Caching, vehicles near the RSU hear and cache the information. Query, users that

wants to find a service queries the cache of the other vehicles close to him

Fig. 4. PEDS state machine operating in vehicle node.



4. IMPLEMENTATION AND TESTBED

We implemented and evaluated PEDS in a proof-of concept vehicular test bed set up with IEEE 802. 11a.

A. Implementation

We added the messages M and Q in the native beacon of IEEE 802. 11a at Information Element Vendor (IE Vendor) field, as shown in Fig. 5. The OSPD XML document size is more than 256 bytes, although IEEE 802. 11a limits the IE vendor size to 256 bytes, we used two IE vendor fields to carry the XML. First field is the <service> and second field is the <description>, both set with the same TYPE and the correspondent length was set in the LENGTH field. The beacon with the XML is transmitted over a Socket Raw. Messages R were added in the probe response unicast messages with the destination to vi which sent the query Q. The PEDS implementation at the RSU sends beacons M every 1 s of interval controlled by the system call sleep (). At the vehicle v, we implemented two POSIX Threads, one for sending Q or R and the other for receiving M or R. The receiving thread uses the PCAP library to capture the beacons and probes. Each received M or R, it extracts the XML document by using the Library libxml2 and stores the services in the cache c. The cache c was implemented as a ring buffer in memory. For search the services, PEDS runs on the buffer and returns the services the name N matches with the query R service name N.

Fig. 5. IEEE 802. 11a native beacon. We injected the XML document into the two consecutive IE vendor fields placed at the

end of frame

The PEDS implementation at the RSU sends beacons M every 1 s of interval controlled by the system call sleep(). At the vehicle v, we implemented two POSIX Threads, one for sending Q or R and the other for receiving M or R. The receiving thread uses the PCAP library to capture the beacons and probes. Each received M or R, it extracts the XML

document by using the Library libxml2 and stores the services in the cache c. The cache c was implemented as a ring buffer in memory. For search the services, PEDS runs on the buffer and returns the services the name N matches with the query R service name N.

B. Testbed

We used five access points (AP) with Pc-Engine Alix3D2 and Ubiquity XR5 wireless card set to a frequency of 5. 8 GHz, transmitting at a 6 Mbit/s data rate and 15 dBm of transmission power. The APs were placed in a row, each separated by � 50 cm, with a 5 GHz interference free area. All the APs had the RSU and vehicle PEDS implementations. We instantiated PEDS twice in the same AP in some tests as an effort to test with more than five nodes. The AP3 placed in the middle represents vi or vj, and the other four could be RSU or vj, depending on the test, as shown in Table I.

5. EXPERIMENTS AND RESULTS

We prove the feasibility of the protocol by means of two tests to assess the three phases of PEDS. First, we tested the beaconing and caching phases for evaluating the sending of beacons M by a RSU and the receiving and storing fo M by vj. Second, the query phase for evaluating the service discovery query procedures. This study was not conducted using experiments on the road or in an urban vehicular environment because of the prohibitive costs. Instead, we based the tests on metrics of inter-contact time for the V2V and V2R transmissions. A inter-contact time occurs when a vehicle passes by an RSU coverage area or passes another vehicle in the opposite lane direction in the road. Rubinstein et al. [16] demonstrated the inter-contact time varies between 10 s to 45 s for two vehicles in opposite lanes with speeds varying between 60 km/h and 20 km/h, respectively.

TABLE 1: ARRANGEMENT OF OSPD INSTANCES FOR EACH DENSITY



In the tests, the inter-contact time was managed by an independent POSIX Thread in AP3 that wakes-up PEDS for a fraction of inter-contact time and, after that, puts it to sleep for 3 s. Moreover, the RSU (dr) and neighbour vehicles (dv) densities were increased up to 7 units. We define dr as the number of RSUs that are under the RF range of one vj and dv as a number of neighbour vehicles vj that are under the RF range of one vi.

The PEDS instances arrangement in the APs for densities dr and dv is shown in Table I. We started testing from 1 and increased by 2 the numbers of RSU and vehicles. The maximum density was 7 units because of the limitation of the number of APs we own. The average and confidence intervals of 95% were computed from a sample of 100 measurements for each inter-contact time.

A. Evaluation phases I and II: Beaconing and caching

This test refers to the vehicle vj when it receives beacons from RSUs in V2R communications, as shown in Fig 6(a). According to the densities, as shown in Table I, an RSU broadcasts one or two beacons M with the services s inside every T = 1 second to prevent flooding in the network [17].

We started the test instantiating PEDS in all RSUs, one by one, and then, in the vehicle vj.

Fig. 6. Evaluation scenario examples: (a) Two RSU and one car receiving beacons. (b) One vehicle querying three neighbours

vehicles.

We measured the inter-contact times from 0. 2 s to 2 s with 100 samples observed in vj. One sample reports a total of M received by vj within an inter-contact time. Thus, beaconing success rate, Sbr, is the rate of total of offered services at that time per the total of services received by vj.

The Sbr mean is shown Fig. 7. For dr up to 5, Sbr is less than 1. 0 for inter-contact times less than 1 s. This indicates the choice of interval T = 1 s affected the Sbr performace, as we have a linear increasing behaviour until 1 s.

Fig. 7. Sbr for the beaconing and caching phases. A vehicle vi receives beacons from RSUs under the density dr varying from 1

to 7.

According to this result, an inter-contact time of 1. 2 s appears to sufficient for a vehicle vj under a density dr = 7 to receive 100% of the advertised services at a given time. When dr > 5 and the inter contact time is less than 1. 2 s, the beaconing performance decreases, probably, because of eventuals packet collision.

B. Evaluation phase III: Querying

This test refers to the vehicle vi when it queries the neighbor vehicles V in V2V communications, as shown in Fig. 6(b). The neighbour vehicles are represented by v[1, 2, 4, 5] and PEDS was instantiated according the densities of the Table I, in the way of the previous test.

To prevent the time of vehicles V to search for a service in cache cj impact in the network performance, all vj have only the service searched by vi stored in cache cj, reducing the search time and minimizing the influence in the evaluation.

Thus, each neighbour vehicle in V sends a unicast response probe R to vi. We measured the inter-contact time from 0. 2s to 2s. We obtained 100 samples for each inter-contact time. One sample reports a total of R received by vehicles vi within a given inter-contact time. The query success rate, Sqr is the rate of the total neighbours vj at that time per the total replies R received by vi.

In this phase, there is a double chance of errors: first, when vi sends the query Q and second, when the vehicles in V reply the responses probe R. If the density dv increases, there is greater chance of packet collision during the replies. We understand that the performance can suffer from the synchronised replies, when the vehicles V receive the query Q



and find and send R at the same time. Thus, is acceptable that the performance decreases as the dv increases.

Fig. 8 shows the average of Sqr for vi. The results indicate that Sqr depends only on the dv value. To understand the reason that the inter-contact time did not affect the query phase, we measured the time of vi starting to send Q and receive the first reply R, which was 1658 �s on average. Therefore, 0. 2 s was a sufficient time for vi to send Q and receive the respective R. When dv = 1, one neighbour vehicle in V is able to know the service searched by vi. Thus, there is a 95% chance that vi will query and receive the reply R within any inter-contact time greater than 0. 2 s. If dr = 7, seven neighbour vehicles in V are able know the service searched by vi, so that vi has a 48% chance of receiving all the replies R, i. e. , vi finds the service, but it does not know all that are available.

Fig. 8. Sqr for the query phase. A vehicle vi queries the vehicles in vj under the density dv varying from 1 to 7.

6. CONCLUSION

Here we have studied the problem of providing users a roadside services in the future WAVE environment without Internet access. It is a users location aware information. Such information improve the users experience during the trip through roadside POI advertisements.

To address this problem, we proposed PEDS – a beaconing-based protocol that is able to find and disseminate services using the store-and-forward messages on available connections. Two tests were conducted in a test bed to demonstrate the feasibility of PEDS. First was the beaconing and caching phases and second was the querying phase.

According to the results, we can conclude: i) to interval T = 1 s, an inter-contact time of 1. 2 s is sufficient for one vehicle receives 100% (Sbr = 1. 0) of the beacons M and ii) the

success rate, Sqr, is not the affected by inter-contact time and the density dv is an important for PEDS performance.

In the future work we will study the security issues of PEDS, perform simulations, and evaluations in urban scenario using V-Beacon [18]. Also, we would like exploring the PEDS as a potential protocol for vehicular sensing and social networks [19].

REFERENCES

[1] Jiang, D.; Delgrossi, L., "IEEE 802. 11p: Towards an International Standard for Wireless Access in Vehicular Environments, "Vehicular Technology Conference. VTC Spring 2008. IEEE, pp. 2036, 2040, 11- 14 May 2008.

[2] Uzcategui, R. ; Acosta-Marum, G. , "Wave: A tutorial, " Communications Magazine, IEEE, vol. 47, no. 5, pp. 126, 133, May 2009

[3] Dikaiakos, M. D. ; Florides, A. ; Nadeem, T. ; Iftode, L. , "Location- Aware Services over Vehicular Ad-Hoc Networks using Car-to-Car Communication, " Selected Areas in Communications, IEEE Journal on, vol. 25, no. 8, pp. 1590, 1602, Oct. 2007.

[4] A. Boukerche, K. Abrougui, and R. W. Pazzi, "Context-aware and location-based service discovery protocol for vehicular networks, " in Proceedings of the 6th ACM symposium on Performance evaluation of wireless ad hoc, sensor, and ubiquitous networks. ACM, 2009.

[5] Abrougui, K. ; Boukerche, A. ; Pazzi, R. W. N. , "An Efficient Fault Tolerant Location Based Service Discovery Protocol for Vehicular Networks, " Global Telecommunications Conference (GLOBECOM 2010), IEEE, pp. 1, 6, 6-10 Dec. 2010

[6] Qing Xu, Tony Mak, Jeff Ko, and Raja Sengupta. Vehicle-to-vehicle safety messaging in DSRC. In Proceedings of the 1st ACM international workshop on Vehicular ad hoc networks (VANET ’04). ACM, New York, NY, USA, 2004.

[7] Mooi-Choo Chuah; Yang, Peng; Davison, B. D. ; Liang Cheng, "Storeand- Forward Performance in a DTN, " Vehicular Technology Conference. VTC 2006-Spring. IEEE 63rd, vol. 1, pp. 187, 191, 7-10 May 2006

[8] W. Hu; K. Xue; P. Hong; and C. Wu, "ATCS: A Novel Anonymous and Traceable Communication Scheme for Vehicular Ad Hoc Networks, " IJ Network Security, vol. 13, pp. 71-78, 2011.

[9] David Antolino Rivas; Manel Guerrero-Zapata, "Chains of Trust in vehicular networks: A secure Points of Interest dissemination strategy, " Ad Hoc Networks, Vol 10, Issue 6, pp. 1115, 1133, Aug 2012.

[10] Rongxing Lu; Xiaodong Lin; Haojin Zhu; Pin-Han Ho; Xuemin Shen, "ECPP: Efficient Conditional Privacy Preservation Protocol for Secure Vehicular Communications, " INFOCOM 2008. The 27th Conference on Computer Communications. IEEE, 13-18 April 2008

[11] Marvy Mansour; Ahmed Fahmy; Mohammed Hashem, "Maintaining Location Privacy and Anonymity for Vehicle’s Drivers in VANET, " International Journal of Emerging Technology and Advanced Engineering, Vol 2, Issue 11, pp. 8, 40, Nov 2012.



[12] Malandrino, F. ; Casetti, C. ; Chiasserini, C. F. ; Fiore, M. ; Yokoyama, R. S. ; Borgiattino, C. , "A-VIP: Anonymous verification and inference of positions in vehicular networks, " INFOCOM, 2013 Proceedings IEEE, pp. 105, 109, 14-19 Apr 2013.

[13] Martelli, F.; Elena Renda, M.; Resta, G.; Santi, P., "A measurementbased study of beaconing performance in IEEE 802. 11p vehicular networks, "INFOCOM, 2012 Proceedings IEEE, pp. 1503, 1511, 25- 30 Mar 2012.

[14] Daniel, A.; Popescu, D. C.; Olariu, S., "A study of beaconing mechanism for vehicle-to-infrastructure communications," Communications (ICC), IEEE International Conference on, pp. 7146, 7150, Jun 2012.

[15] Slavik, M. ; Mahgoub, I. ; Alwakeel, M. M. , "Analysis of beaconing message rate in VANET multi-hop broadcast protocols, " High Capacity Optical Networks and Enabling Technologies (HONET), 9th International Conference on, pp. 037, 041, 12-14 Dec. 2012

[16] Rubinstein, M. G. ; Abdesslem, F. ; Amorim, M. D. ; Cavalcanti, S. R. ; Alves, R. S. ; Costa, L. H. M. K. ; Duarte, O. C. M. B. ; Campista, M. E. M. "Measuring the capacity of in-car to in-car vehicular networks. " Communications Magazine, IEEE 47. 11 (2009): 128-136.

[17] Sommer, C. ; Tonguz, O. K. ; Dressler, F. , "Adaptive beaconing for delay-sensitive and congestion-aware traffic information systems, " Vehicular Networking Conference (VNC), IEEE, pp. 1, 8, 13-15 Dec. 2010

[18] Yokoyama, R. S. ; Kimura, B. Y. L. ; Moreira, E. D. S. “V-Beacon: Uma plataforma experimental para Redes Veiculares Sem Fio”. III Simposio Brasileiro de Engenharia de Sistemas Computacionais (SBESC13), Niteroi-RJ Brasil. Pag. 1-6. Nov, 2013.

[19] Uichin Lee, Mario Gerla, A survey of urban vehicular sensing platforms, Computer Networks, Vol. 54, Issue 4, pp. 527, 544, Mar 2010.



Review on Emerging Techniques to Detect Oral Cancer

Santhosh B.

Department of Telecommunication Engg Dayananda Sagar College of Engineering, Bangalore

[email protected]

Abstract: Oral tumor is a kind of head and neck malignancy, is any carcinogenic tissue development situated in the oral hole. Oral growth is one of the hazardous wellbeing issue on the planet. On the off chance that the diseases are recognized in the late stage prompts life undermining. So it important to distinguish such sorts of disease at the preparatory stage. Early recognition assists specialists with providing on time solution which is advantage to the patients. There are numerous restorative picture handling procedures requires consistent enhance nature of administrations in human services industry. This paper gives accessible picture handling procedures to recognize and arrange the oral malignancy.

Keywords: Oral cancer, Medical Image Processing, Segmentation, Feature Extraction

1. INTRODUCTION

Tumor is a strange development of cells. There are more than 100 sorts of tumor, including bosom malignancy, skin disease, lung growth, oral tumor and so on. Oral malignancy shows up as a development in the mouth [10]. Oral malignancy, which incorporates diseases of the lips, tongue, cheeks, floor of the mouth, hard and delicate sense of taste, sinuses, and pharynx (throat), can be life undermining if not analyzed and treated early. Side effects change contingent upon the sort of malignancy. Tumor treatment may incorporate chemotherapy, radiation, and/or surgery. Early assessment of oral precancerous injuries can dramatically affect oral malignancy death rates [13].

In spite of late indicative and restorative advances, the 5-year survival rate for oral malignancy has stayed under half in the course of the most recent 50 years inferable from the accompanying reasons [11]:

• The greater part of oral tumor cases (60%) present with cutting edge stages (III and IV) at conclusion;

• Oral growth has the most astounding danger for the advancement of second essential tumors ('field cancerization marvel') of any maligna.

The finding of Oral pre-tumor location is the test in the wellbeing business. Especially in the location, assessment of right on time stages. In spite of the fact that oral malignancies are distinguished effectively, recognizable proof gets to be troublesome in introductory stages.

The medicinal picture examination assumes a vital part in clinical finding of oral malignancy and treatment of specialist. Assortment of cutting edge imaging procedures, for example, Magnetic Resonance Imaging (MRI), ultrasound, Computerized Tomography (CT) and Radiography and so forth give the needful data to the wellbeing business. For some reasons, the utilization of data by manual examination is troublesome. With the propelled innovation in the field of PC, imaging innovation, cutting edge preparing apparatuses and develop method for taking care of picture information, imaging systems have gone into therapeutic field. This expanded the capacity of comprehension, there by the level of conclusion has enormously enhanced by utilizing the preparing and investigation of diverse restorative pictures. The preparatory determination of oral tumor depends on visual assessment and enlistment of the persistent's oral hole as real nature computerized pictures. Albeit reciprocal systems exist, based e. g. on infrared or fluorescence spectroscopy[14].

The MR pictures might not have high determination due to moving ancient rarities affected by the moving tongue and jaw. So a proficient picture preparing calculation is expected to recognize the suspicious locale in the disease range with high determination. The recognition of malignancy tissues in the early stage is troublesome by dental radiograph [15].

This paper examines an assortment of oral malignancy discovery calculations and picture preparing strategies in the wellbeing business.

2. RELATED WORK

Literature on various techniques are described to detect and classify the oral cancer in digital images. A lot of research has been done on detection of oral cancer.

Santhosh B..


Lalit Gupta et al [1] proposed another system for Feature Selection utilizing Mean – movement and Recursive Feature Elimination strategies to build segregation capacity of the element vectors. The creator portrayed that the assessment of the execution of the calculation done on an in-vivo recorded LIF information set comprising of spectra from typical, dangerous and pre-threatening patients. The scope of affectability and specificity is >95 gotten towards harm utilizing the proposed technique

Sebastian Steger et al [2] have proposed a strategy for novel picture highlight extraction approach that is utilized to foresee oral malignancy reoccurrence. This examination work proposed different numeric picture highlights that portray tumors and lymph hubs. . This work presented the accompanying methodology which is autonomous from human subjectivity: Registration and managed division of CT/MR pictures shapes the base of the mechanized extraction of geometric and composition components of tumors and lymph hubs. To decrease the measure of client cooperation amid subsequent meet-ups this work joins the past examination's division results. The strength and the numeric way of the extricated components make them in a perfect world suited as information for an advanced versatile expectation environment that gauges the probability of oral tumor reoccurrence and helps the clinician to build up a treatment arrangement.

M. Muthu Rama Krishnan et al [3] have proposed a wavelet based composition arrangement for oral histopathological segments. As the traditional technique includes in stain force, entomb and intra spectator varieties prompting higher misclassification mistake, another system is proposed. The proposed system, includes highlight extraction utilizing wavelet change, highlight determination utilizing Kullback – Leibler (KL). In this work creator considered 67 typical and 47 OSF histopathological pictures as per review study outline convention. In the preprocessing level, middle channel and histogram based complexity improvement have been utilized as a part of request to lessen the commotion for normalizing recoloring varieties of epithelium. The division of epithelium from the grayscale picture is refined by changing over the standardized picture into twofold picture utilizing fluffy dissimilarity and post preparing is done utilizing morphological operations. From that point, surface components of the epithelial district are extricated utilizing Haar, Daubechies, Coiflet, Symle biorthogonal wavelet families including Gabor+wavelet. Subsequently, feature selection is done using Kullback–Leibler (KL) divergence and probability density function is also estimated to show the discriminating potentiality of each feature vector. The ideal subset of wavelet based surface components are encouraged to Bayesian classifier and bolster vector machines (SVM) for screening and characterization of OSF. In this study, it is watched that SVM with straight piece capacity gives better

grouping precision (92%) when contrasted with Bayesian (76. 83%). All in all, it can be prescribed that the textural components of typical oral mucosa and OSF show critical varieties that, when measured, could be entertained for supporting in the analytic procedure.

Ranjan Rashmi Paul et al [4] proposed a recognition procedure to recognize oral malignancy utilizing a system called wavelet on neural systems. In the work, the wavelet coefficients of Transmission Electron Microscopy (TEM) pictures of collagen filaments from ordinary oral sub mucosa and Oral Sub mucous Fibrosis (OSF) tissues have been utilized as a part of request to pick the element vector and to prepare the Artificial Neural Network [16].

Anu Radha et. al [5] proposed Detection of Oral Tumor in light of Marker – Controlled Watershed Algorithm. In this paper, a novel system is proposed to recognize tumor cell present in mouth gave by an Orthopantomogram. This paper depicted that, a novel numerical morphological watershed calculation is proposed to protect these edge subtle elements and in addition noticeable ones to recognize tumors in dental radiographs. Applying watershed on pictures prompts over division despite the fact that it is preprocessed. To stay away from the overhead because of over division, the system called Marker Controlled Watershed division is utilized to section tumors. The outcomes got are entirely great and were tried. The Algorithm ventures for Marker – Controlled Watershed division are as per the follow

1. Compute a segmentation function. This is an image whose dark regions are the objects to be segmented.

2. Compute foreground Markers. These are connected blobs of pixels within each of the objects.

3. Compute background Markers. These are pixels that are not part of any object.

4. Modify the segmentation function so that it only has minima at the foreground and background Marker locations.

5. Compute the watershed transform of the modified segmentation function. Marker – driven water

Fig. 1. Enhanced Image Fig 2. Image after Watershed



The info picture is upgraded utilizing straight difference extending which is appeared in Figure 1. Subsequent to preprocessing, watershed division is connected to the picture

Fig 3. (a) Fig. 3. (b)

Fig 3. (c) Fig 3. (d)

Fig 3. (e) Fig. 3. (f)

Fig. 3. (a), (b), (c), (d), (e) Marker – Controller watershed Segmentation results (f) shows tumor area The output image (e)

shows the tumor part with less blurring and noise.

TABLE 1: Comparison of Algorithms

Woonggyu Jung et al [6] proposed a method in oral disease recognition utilizing Optical Coherence Tomography (OCT). For the imaging profundity of 2-3 mm, OCT is suitable for

oral mucosa. They likewise distinguished oral disease in 3-D volume pictures of ordinary and precancerous sores [3].

Neha Sharma. et. al [7] proposed Extracting Significant Patterns for Oral Cancer Detection Using Apriori Algorithm. In this paper, the creator talked about a methodology for disease identification and counteractive action in light of investigation utilizing affiliation guideline mining. The information dissected are relating to clinical manifestations, history of dependence, co-dismal condition and survivability of the growth patients.

Apriori Algorithm:

The apriori is an excellent calculation for successive thing set mining and affiliation principle learning over the value-based databases[17]. It continues by distinguishing the regular individual things in the database and stretching out them to bigger and bigger thing sets the length of those thing sets show up adequately frequently in the database. The regular thing sets controlled by an apriori can be utilized to decide affiliation rules, which highlight general patterns in the database. Affiliation standards mining utilizing apriori calculation utilizes a "base up" methodology, broadness first pursuit and a hash tree structure to tally the hopeful thing sets proficiently. A two-stage apriori calculation is clarified with the help of flowchart as appeared in Figure 4 and the calculation is specified underneath:

Apriori calculation: Candidate Generation and Test Approach Step 1: Initially, filter database (DB) once to get successive 1-itemset.

Step 2: Generate length (k + 1) applicant thing sets from length k visit thing sets.

Step 3: Test applicants against DB.

Step 4: Terminate, if no regular or competitor set can be Generated. To choose fascinating standards from the arrangement of every single conceivable tenet produced, requirements on different measures of noteworthiness and hobby can be utilized. The best-known limitations are least edges on backing and certainty.

Anuradha, et. al [8] proposed Statistical Feature Extraction to Classify Oral Cancers. The proposed strategy fragments and arranges oral growths at a prior stage. The tumor is recognized by Marker Controlled Watershed division. The components removed utilizing Gray Level Co event Matrix (GLCM) is Energy, Contrast, Entropy, Correlation, and Homogeneity. The separated elements are encouraged into Support Vector Machine (SVM) Classifier to group the tumor as kindhearted or harmful. The exactness got for the proposed strategy is 92. 5%. The proposed square graph appeared in the figure 5

Santhosh B..


Fig. 4. Flow chart of apriori algorithm

Fig. 5. Proposed method block diagram

For the proposed work 27 pictures were picked haphazardly.composition Features are acquired for the divided some portion of the tumors. GLCM elements are removed and its arrangement was gotten. From Table2 watch the component values for the different specimen pictures.

TABLE 2: Feature Extraction

From Table 2, the pictures are named typical and anomalous utilizing SVM Classifier. Additionally the diagram appeared in Figure 6 speaks to the factual element values for favorable and harmful injuries of oral tumor.

Fig. 6. Performance Analysis

Anuradha, et. al [9] proposed Oral Cancer Detection Using Improved Segmentation Algorithm. This work depicted that the discovery of oral growths utilizing Image Processing. Dental X – Rays are utilized as the Input Image for recognition. In the initial step, Linear Contrast Stretching is utilized to expel commotion from the Dental X – Ray Image. Watershed Segmentation and Marker Controlled Watershed Segmentation are utilized to portion tumors from the improved Image. Issue of over division emerges in both the division calculations. In this way, Marker Controlled Watershed Segmentation is moved forward. The Segmentation Algorithms are looked at for pace and precision. The velocity is computed previously, then after the fact Linear Contrast extending. The proposed calculation gives better segmentation

3. DISCUSSION

Author Technique Advantages

Lalit Gupta

A new feature selection and classification scheme for screening of oral cancer using laser induced fluorescence

Increase discrimination ability of the feature vectors Sensitivity : Above 95% and specificity : Above 99%

Sebastian novel image Oral Cancer Prediction



Steger feature extraction approach

Automatically

M. Muthu Rama Krishnan

Wavelet based textureclassification of oral histopathological sections

Improves the Accuracy

Ranjan Rashmi

A novel wavelet neural network based pathological stage detection technique

ProtectionAuthentication

K Anu Radha

Detection of Oral Tumor based on Marker – controlled Watershed Algorithm

Accurecy is more than 90% the quality of the image is enhanced using linear contrast stretching.

Woonggyu Jung

Optical Coherence Tomography

Accurecy is more detected oral cancer in 3-D volume images of normal and precancerous lesions

Neha Sharma

Apriori Algorithm

the highest confidence level, thereby, making them very useful for early detection and prevention of oral cancer

Anuradha

Statistical Feature Extraction to Classify Oral Cancers. Transform

The proposed system segments and classifies oral cancers at an earlier stage.

Anuradha

Oral Cancer Detection Using Improved Segmentation Algorithm

Better speed and accuracy

4. CONCLUSION

In this paper the survey study of various oral cancer detection techniques in the medical image processing is done. The analysis of medical image processing and its applications in healthcare industry are described in this paper. The advanced Medical image processing techniques and algorithms are reviewd.

5. ACKNOWLDGEMNET

I am truly thankful to my Guide Dr. K. Viswanath, Professor, Department of Telecommunication Engineering, Siddaganga Institute of Technology, Tumkur and in addition Department

Telecommunication Engineering, Dayananda Sagar School of building for a wide range of bolster and consolation to do this examination work.

REFERNCES

[1] L. Gupta, S. K. Naik and S. Balakrishnan, “A New Feature Selection and Classification Scheme for Screening of Oral Cancer using Laser Induced Fluorescence”. Medical Biometrics, 4901, pp. 1–8. 2007

[2] Sebastian Steger, Marius Erdt, Gianfranco Chiari and Georgios Sakas, “Feature Extraction from MedicalImages for an oral cancer reoccurrence prediction environment”, World Congress on Medical Physics andBiomedical Engineering, September 7 - 12, 2009, Munich, Germany.

[3] M. Muthu Rama Krishnan, Chandran Chakraborthy, Ajoy Kumar Ray, “Wavelet based textureclassification of oral histopathological sections”, International Journal of Microscopy, Science, Technology, Applications and Education, pp 897-906.

[4] Ranjan Rashmi Paul, Anirban Mukherjee, Pranab K. Dutta, Swapna Banerjee, Mousumi, Pal, JyotirmoyChatterjee and Keya Chaudhuri, “A novel wavelet neural network based pathological stage detectiontechnique for an oral precancerous condition”, Journal of Clinical Pathology, Vol. 58, Issue. 9, pp 932 –938, 2005

[5] K. Anuradha, K. Sankaranarayanan,” Detection of Oral Tumor based on Marker – controlled Watershed Algorithm”, International Journal of Computer Applications (0975 – 8887) Volume 52– No. 2, August 2012

[6] Woonggyu Jung, Jun Zhang, Jungrae Chung, Petra Wilder – Smith, Matt Brenner, J. Stuart Nelson and Zhongping Chen, (2005) “Advances in Oral Cancer Detection using Optical Coherence Tomography”, IEEE Journal of Selected Topics in Quantum Electronics, Vol. 11, No. 4. pp 811 – 817.

[7] Neha Sharma1, Hari Om, " Extracting Significant Patterns for Oral Cancer Detection Using Apriori Algorithm " Intelligent Information Management, 2014, 6, 30-37

[8] K. Anuradha, K. Sankaranarayanan,” Statistical Feature Extraction to Classify Oral Cancers. Transform”. Journal of Global Research in Computer Science, Volume 4, No. 2, February 2013

[9] K. Anuradha, K. Sankaranarayanan “Oral Cancer Detection Using Improved Segmentation Algorithm”, International Journal of Advanced Research in Computer Science and Software Engineering, Volume 5, Issue 1, January 2015 http://www. webmd.com/oral-health/guide/oral-cancer

[10] Dongsuk Shin, Nadarajah Vigneswaran, Ann Gillenwater, and Rebecca Richards-Kortum, “Advances in fluorescence imaging techniques to detect oral cancer and its precursors”, Future Oncol. ; in PMC 2011 May ACS, American Cancer Society, http://www. cancer. org

[11] James J. Sciubba, “Improving detection of precancerous and cancerous oral lesions”, American Dental Association, Vol. 130, October 1999, pp 1445 – 1457.

[12] J. K. Dhlngra, D. F. Perrault Jr. , K. McMillan, E. E. Rebeiz, S. M. Shapshay, R. Manoharan, I. Itzkan, M. S. Feld, S. Kabani, “Spectroscopic analysis of precancerous and cancerous lesions

Santhosh B..


of the oral cavity”, IEEE Lasers and Electro-Optics Society Annual Meeting, Vol. 1, pp. 332-333, 1996.

[13] K. Anuradha, K. Sankaranarayanan, “Identification of Suspecious Regions to Detect Oral Cancer at Earlier Stage-A Literature Survey”, International Journal of Advances in Engineering & Technology, ISSN: 2231-1963, March 2012.

[14] R. Hari Kumar, N. S. Vasanthi, M. Balasubramani, “Performance Analysis of Artificial Neural Networks and

Statistical Methods in Classification of Oral and Breast Cancer Stages” International Journal of Soft Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume 2, Issue 3, July 2012.

[15] Rakesh Agrawal, and Ramakrishnan Srikant,”Fast algorithms for mining association rules in large databases”. Proceedings of the 20th International Conference on Very Large Data Bases, VLDB, pages 487-499, Santiago, Chile, September 1994.



Types of Cognitive Agents Based on Learning Models Devaraj Verma C.1, M.V. Vijayakumar2, Darryl N. Davis3

1Research Scholar, NHCE (VTU) & Assistant Professor, Department of MCA-VTU, DSCE, Bangalore-560078, Karnataka-INDIA

[email protected] 2Professor and Research Director, Department of Computer Science and Engineering,

Dr. Ambedkar Institute of Technology, Near Jnana Bharathi Campus, Malathahalli, Bangalore-560056, Karnataka-INDIA; [email protected]

3Director of Research, Computer Science, University of Hull, HU6 7RX. http://www2. dcs. hull. ac. uk/NEAT/dnd/index. htm

d. n. [email protected]

Abstract: This paper investigates the types of agents based on the concept of learning models. Generally people learn about the concept from theories, books and websites. Whatever may be the learning method, it involves three stages: Perceiving, Processing, and Organizing and presenting the information. The paper discusses various learning models or theories proposed by the researchers Honey and Mumford (1982), Kolb’s, and Howard Gardner [1, 3, and 8].

Keywords: Agent classification, learning style models or theories.

1. INTRODUCTION

People can know about their learning styles from the several competing theories available. There are websites from which any one can take the test to judge what natural learning style they would posses. Available postulates wrap the three main concepts of the learning style how people in the society learn and present the information to the audience: • Perceiving information • Processing information • Organising and presenting information.

A. Perceiving information

In order to get the information from the world around us anyone can apply their senor organs. Few of them are very talented and hence they get the information by applying the single organ, were as others may employ more sense organs for the same task. The VARK system (Fleming, 2001) perceives the quantity of learning behaviour of people rely on the sense organs such as Visual (sight), Auditory (hearing), Reading and Kinaesthetic (other sensations which includes touch and temperature as well as movement).

B. Processing information

Once the information from the environment had been acquired by applying one or more senses such as listening, reading, etc.

then it can be practiced mentally, as you imagine the mechanism of learning things and remember it. Any one will have a normal penchant for how to progress the information mentally.

C. Organizing and presenting information

At the end, there is a mechanism of how one can select it for distribute the information with others. One can have choice about how to

(a) Systematize information: Anyone can organize the information with a holistic overview, or with detailed and logical analysis.

(b) Present information: Anyone can present the information either verbally or using images.

2. LEARNING MODELS

Several learning style theories are available in order to determine how the people in the society learn from their surrounding environment. For example, some students in the class room will grasp and learn while listening to the lecture by taking the notes i, e they learn by the body movement and so they are called as kinaesthetic learners. Others will grasp and learn only by hearing to the lecture and they are called as auditory learners. The learning theories are Honey and Mumford, Kolb’s, VAK, and Gardner’s multiple intelligent.

A. Honey and Mumford Theory

Honey and Mumford (1982) devised a prominent self-test, which can be taken by anyone to determine their learning style predominantly an activist, a reflector, a theorist, or a pragmatist. In order to gather the likes and dislikes of the learning behaviour any one can obtain the test from the



existing websites online and acquire the percentage of learning knowledge they can have.

An Activist can learn good from tasks which involves new experiences and challenges, tasks that involve competitive teamwork and problem-solving, excitement, change and variety.

A Reflector is one who learn good from tasks in which one is allowed or encouraged to watch, think, ponder on tasks, one is having time to think before acting and to assimilate before commenting, one can carry out his research in detailed and carefully, one can have time to review his learning, one can carefully consider his results to produces the report, one can exchange their views and/or results of the work without any danger with the other people by making an agreement, within the boundaries, and one can take decision without any pressure and tight deadlines.

Theorist is one who learn good from tasks that involves what is offered whether it is a part of the system, form, concept or hypothesis, one can discover cleverly the relations and interrelationships between facts, proceedings and situations; one can probe and question basic tactic, assumptions or logic, one is stretched intellectually, for example where one is being asked to analyse and assess the results, then take a broad view based on some associated concepts, one is have the clear purpose to reach the goal in structured situations, one is having good knowledge of ideas and concepts of the relevant work being done by him.

Pragmatists can learn good from the situations that contains an evident connection sandwiched between the area under discussion matter and a real life problem; they gain knowledge of best from the shown methods for doing belongings with obvious hands-on sessions; they study best from the likelihood to attempt out and put into practice techniques with training or response from a believable professional; they discover finest from mock-up that can imitate, or examples / anecdotes.

B. Kolb’s Theory

Kolb’s model is depicted in the below fig 1. The model shows four different learning choices that are based on the four stage learning cycle. The model follows the well known learning technique called experiential learning that is of great value to trainers because it allows the trainers or users to understand and adapt to the employed training technique, based on peoples learning choices.

Concrete experience is the one that involves the features such as seeing, hearing, feeling or doing something that leads to the observation and reflection on what have been just experienced, which leads to abstract conceptualization about whether or not

the learning experience was useful, enjoyable etc to the learner and whether or not the learner would do it again.

If so, how would the learner do it differently? This in turn leads to Active experimentation in order to test the learner’s abstract concept, which leads to Concrete experience and so the cycle continues (Kolb, Honey and Mumford, 2010).

C. Multiple Intelligence Theory

It was anticipated by Howard Gardner in 1983 as a model of cleverness that differentiates brainpower into a mixture of precise first and foremost sensory modalities to a certain extent than bearing in mind it as conquered by a single universal talent.

Gardner argues that there is an extensive range of good cognitive abilities, and their exist few correlations among them. For example, the theory prognosticates that a child who finds to challenge to memorize the multiplication tables might not have an obvious talent than the other child who will gather the multiplication table quickly. The kid that needs enough moment to grasp the easy multiplication 1) may paramount become skilled at to multiply through a diverse approach, 2) may do extremely well in a turf exterior of mathematics, or 3) may even be perceiving at and understanding the multiplication procedure at a vitally deeper altitude, or conceivably as an completely dissimilar route. Such a essentially deeper understanding can result in what the child looks like sluggishness and can conceal a mathematical brainpower potentially higher than that of a kid who rapidly remembers the multiplication table regardless of a smaller amount thorough understanding of the progression of multiplication.

Types of Cognitive Agents Based on Learning Models


The speculation has been met with diverse responses. Traditional cleverness tests and psychometrics have generally found high association between dissimilar tasks and facet of brilliance, rather than the little associations which Gardner's hypotheses anticipate. However many educationalists assist the hands on importance of the methods suggested by the postulate [1].

Gardner uttered quite a lot of criteria for a conduct of the person to be acumen. These were that the intelligences [2]: 1. Prospective for brain separation by brain injure,

2. Position in evolutionary account,

3. Existence of heart operations,

4. Vulnerability to encode the figurative idiom,

5. A discrete developmental development,

6. The survival of savants, prodigies and other extraordinary people,

7. Sustain from investigational psychology and psychometric conclusions.

Gardner presumes that eight talents meet these criteria are spatial, linguistic, logical-mathematical, bodily-kinesthetic, musical, interpersonal, intrapersonal, and naturalistic [3].

D. VAK Theory

This theory classifies the learners into the three categories visual, auditory and kinaesthetic. Some people will learn the surrounding environment through perception and are called as visual learners. There are few of them who will learn through hearing and are called as auditory learners. And, others learn through the body movements and are called as kinaesthetic learners.

3. TYPES OF COGNITIVE AGENTS

TABLE 1: Agents classification based on Honey & Mumford Theory

S. No Agent name Agent’s Description

1. Activist Agent prefer to “have an experience”

2. Reflector Agent prefer to “review an experience”

3. Theorist Agent prefer to “draw conclusions from an experience”

4. Pragmatist Agent

prefer to “plan the next experience” based previous experience

According to the above theories or models, the intelligence is reflected by the collective behaviors of large numbers of very

simple interacting agents, semi-autonomous individual agents, or complex agents. Whether we consider these agents to be neural cells of the brain, individual members of a species within the environment, or a single person in a society within the real world, their interactions with each other will definitely produce the intelligence to reach the goal.

TABLE 2: Agents classification based on Kolb’s Theory

S. NO AGENT NAME AGENT’S

DESCRIPTION

1. Diverging Agent (concrete, reflective)

• Agent visualizes tangible atmosphere from various sights and adapts by examination instead by action.

2. Assimilating Agent (abstract, reflective)

• Agent pulls a number of different observations and thoughts into an integrated whole.

• Agent have enough time to think things (i, e think on observations made) through before performing the adaptive action into the environment.

3. Converging Agent (abstract, active)

• Agent thinks about the action.

• Agent has decision-making, problem-solving, and the practical application of ideas.

• Thinks and perform the action carefully into the environment based the current knowledge of the environment.

4. Accommodating Agent (concrete, active)

• Agent is superior at adapting to the modified situations; finds the answer to the evils in a spontaneous, trial-and-error



comportment, such as breakthrough knowledge.

• Agent performs action into the environment without thinking i, e lack of planning & decision ability.

• Thus consequences of the actions will not be known.

TABLE 3: Agents classification based on Howard Gardener’s Theory

S. No Agent name Agent’s description

1. Linguistic Agent

• Agent is susceptible to the definition and sequence of words.

• Use tricks that involve audible range, listening, impulsive or official verbal communication, dialect twisters, temper, verbal or soundless interpretation, credentials, innovative text, spelling, journal, verse.

2. Logical Agent

• Agent is talented to finger manacles of way of thinking and identify patterns and guidelines.

• Example as in a scientist:

• Use methods that contain abstract notations/formulas, exactness, graphic things for arrangements, numeric series, assessments, expound codes, crisis solving.

3. Musical Agent

• Agent is thin-skinned to pitch, tune, beat, and pitch.

• Example as in a composer:

• Use actions that engage auditory strip, harmony recital, singing on input, whistling, droning,

environmental sounds, drumming ambiance, rhythmic orders, music composition, and tonal schemes.

4. Kinesthetic Agent

• Agent is competent to use the body dexterously and knob objects skillfully.

• Example as in an athlete or dancer:

• Use behavior that rivet character playing, bodily movements, comedy, inventing, sphere transient, sports events, and corporeal work out, body language, cavort.

5. Visual Agent

• Agent assesses the globe precisely and tries to reconstruct or convert aspects of that globe.

• Example as in a carver or airplane pilot:

• Use tricks that entail talents, movies, sculpture, drawings, sketching, intelligence mapping, patterns/designs, tint plans, vigorous mind, metaphors, and chunk construction.

6. Interpersonal Agent

• Agent can comprehend the people and correlation.

• Example as in a salesman or professor and sense by vigorous thoughts of each other:

• Use tricks that grip faction projects, partition of toil, sensing others' motives, getting/charitable opinion, teamwork skills.

7. Intrapersonal Agent

• Agent possesses admission to one's emotional life as resources to comprehend one self and others, manifest by persons with perfect views of themselves.

Types of Cognitive Agents Based on Learning Models


• Agent use tasks that engross passionate dispensation, noiseless mirror image methods, opinion strategies, focus skills, privileged way of thinking, and meta-cognitive techniques.

TABLE 4: Agents classification based on VAK Theory

S. NO AGENT NAME

AGENT’S DESCRIPTION

1. Visual Agent

• These agents have two sub-straits – linguistic (written) and spatial (pictures).

• These agents learn from viewing the written information that involves reading from the book that contains the printed or written text.

• Agent memorizes the written information by them rather than the read text in spite of not read the written text more than once.

• Agents who are visual-spatial can learn the things better by the help of the charts, demonstrations, videos, and other visual things, but do not perform well with the written text.

• These agents can easily envision faces and places by using their thoughts and infrequently get lost in new ambiance.

•

2. Auditory Agent

• These agents talk to each other.

• These agents use the lip movement and read the information a bit louder to understand.

• Agents of this type might find the trouble in reading and writing responsibilities.

• They can perform well in situations like speaking to the other subordinates in their domain or recording the conversation of themselves and hearing it later after recording.

3. Kinesthetic Agent

• The agents will perform in a good manner while sensing and walking.

• This type of agent also have two sub-channels: kinesthetic (movement) and tactile (touch).

• They learn less when there exists a little or nil outside stimulus or body movement.

• For e.g., the agents deployed in the situation want to have the movement of the organs of the body in order to understand the things going in the surrounding environment. Such as in case of the children’s listening to the teacher in the class room simply move their hands to take the notes of the session.

• Generally these agents can use the highlighters and mark the notes by drawing films, arts, or sketching.

•

4. CONCLUSION

This paper considered several learning style theories Honey and Mumford, VAK, multiple intelligence and Kolb’s. All these theories suggested how the people in the society can learn from the surrounding environment situation. Based on these theories, this paper categorized the agents how they can learn from the environment in which they are deployed

REFERENCES

[1] “Waldorf education embodies in a truly organic sense all of Howard Gardner's seven intelligences not simply an amalgam of the seven intelligences. Many schools are currently



attempting to construct curricula based on Gardner's model simply through an additive process (what can we add to what we have already got?). Steiner's approach, however, was to begin with a deep inner vision of the child and the child's needs and build a curriculum around that vision.” Thomas Armstrong, cited in Eric Oddleifson, Boston Public Schools As Arts-Integrated Learning Organizations: Developing a High Standard of Culture for All

[2] “Lynn Gilman, Human Intelligence” [3] “Robert Slavin Educational Psychology”, 2009, p. 117 ISBN

0205592007 [4] Gardner. infed. org (2008-06-15). Retrieved on 2011-10-22. [5] P.E. Vernon. (1950). “The structure of human abilities”.

University of Michigan [6] J.B. Carroll. (1993). “Human cognitive abilities: A survey of

factor-analytic studies”. Cambridge University Press [7] D. Wechsler. (1997). “Wechsler Adult Intelligence Scale III”. [8] Gardner (May 1984), "Heteroglossia: A Global Perspective"

Interdisciplinary Journal of Theory of Postpedagogical Studies. [9] Medina, Suzanne L. (1993). “The Effect of Music Upon

Second Language Vocabulary Acquisition. ERIC Clearinghouse on Languages and Linguistics, Center for Applied Linguistics”.

[10] Gardner, H. (1995). “How Are Kids Smart: Multiple Intelligences in the Classroom --Administrators' Version ISBN# 1-887943-03-X -- National Professional Resources Dr. Howard Gardner, along with teachers and students from Fuller Elementary School in Gloucester, MA, discuss the theory behind Multiple Intelligences and demonstrate how they have

integrated it into their classrooms and community”. (41 minutes)

[11] Gardner “Interpersonal Communication amongst Multiple Subjects: A Study in Redundancy,” Experimental Psychology (2002)

[12] Gardner, Howard. (1999) “Intelligence Reframed: Multiple Intelligences for the 21st Century.” New York: Basic Books.

[13] Tupper, K. W. (2002). “Entheogens and Existential Intelligence: The Use of Plant Teachers as Cognitive Tools”. Canadian Journal of Education 27 (4): 499–516. doi:10. 2307/1602247. http://www. csse-scee. ca/CJE/Articles/FullText/CJE27-4/CJE27-4-tupper. pdf.

[14] “This information is based on an informal talk given on the 350th anniversary of Harvard University on September 5, 1986. Harvard Education Review”, Harvard Education Publishing Group, 1987, 57, 187-93.

[15] Helding, L. (2009). “Howard Gardner's Theory of Multiple Intelligences”. Journal of Singing, 66(2), 193-199. Retrieved from EBSCOhost.

[16] “http://www. newcityschool. org/ WhatisMI_19. aspx” [17] Willingham (2004), “Check the Facts: Reframing the Mind”. [18] Gardner, Howard (1998). "A Reply to Perry D. Klein's

'Multiplying the problems of intelligence by eight'". Canadian Journal of Education 23 (1): 96–102. doi:10. 2307/1585968. JSTOR 1585790.

[19] Klein, Perry D. (1998). "A Response to Howard Gardner: Falsifiability, Empirical Evidence, and Pedagogical Usefulness in Educational Psychologies". Canadian Journal of Education 23 (1): 103–112. doi:10. 2307/1585969.



Nonintrusive Method for Liquid Level and Volume Measurement

V.G. Sangam1, Vinyl Ho Oquino2 1Professor & Head Electronics and Instrumentation Engineering Dept,

Dyananda Sagar Engineering College, Bangalore, India; [email protected] 2Assistant Professor, ASTU Ethio, Institute of Integrated Electrical Engineers

philippines [email protected]

Abstract: Liquid level is an important process parameter; it plays a key role in many industrial processes. Accurate gauging of liquid level is necessary when computing liquid volume in the process tanks. The accuracy of the measurement depends upon the measurement system used and the conditions in which these level measurement techniques are made. When the liquid in the tank is corrosive, it generally gets evaporated, and hence the contact type techniques are not recommended.

In this paper an attempt has been made to develop a system which measures liquid level and volume using nonintrusive technique. We have proposed He-Ne laser based technique with ARM-7 controller. A prototype experimental set up has been developed and the measurement has been carried out. Results obtained are compared with conventional dipstick method; It has been found that that the proposed method shown accuracy of ±2. 5% with sensitivity of 7mv/ml for water and 10mV/ml for slurry. Repeatability of around 0. 3% is observed. Also volume is calculated and compared with obtained experimental results and shows accuracy of ±3% with sensitivity of 90mV/m3 for water and 100mV/m3 for slurry, with the dynamic range supported by the proposed method is up to 250mm. The proposed method is well suited for closed tank level measurement.

Keywords: Liquid level, computing, nonintrusive method, dipstick method, Arm 7, and dynamic range.

1. INTRODUCTION

Liquid level and volume measurement techniques are generally based on resistive, inductive, capacitive, piezoelectric, ultrasonic and fiber-optic transducers [1]. In recent times, many researchers have proposed and reported several liquid level sensing techniques based on fiber-optic sensors [2]-[6]. These techniques measure liquid levels in terms of discrete levels. At the same time, they are characterized by a very small dynamic range of measurement in spite of their good sensitivities. The ultrasound sensors are also used to measure liquid level [7]. In ultrasound measurement techniques, the system parts should have good acoustic reflection properties. Otherwise, the sound waves are scattered. This may be a problem for different container

shapes. Furthermore, if there are gas bubbles in the liquid the waves are scattered through the bubbles.

Number of physical principles is in practice to measure liquid-level sensors [8-10]. Generally, the choice of the most suitable level sensor for a specific application is based on the requirements such as; Dynamic range of measurement, Resolution, Accuracy, Characteristics of the liquid and Environment [11].

The presented technique provides better resolution of liquid level. At the same time, this method can be used for measurement of liquid level in excess of 150 mm in depth, whereas [2–5] do have 5–100 mm. Further, this technique can be used for precise measurement of liquid levels and volumes in biochemical, pharmaceutical laboratories and field applications.

2. MATERIAL AND METHODOLOGY

2. 1 Laser source and detector.

In the proposed method the linearly polarized Helium-Neon laser is used. It operates at frequency 474Hz, having wavelength of 632. 99nm with speed of 2. 99x108 m/sec and wave time of 2. 1115E−15 sec with power output 5mv [Jain-laser, India make].

There are various sensors used for sensing laser beam, few sensors works on Photo resistive and photoconductive principles are commonly used for their various advantages [11]. Here photo resistive sensor (LDR) is used has; operating temperature range of -55˚C to +125˚ C, ¼ Watt power rating, temperature coefficient of resistance ±15 ppm/˚C, with tolerance in resistance of ± 0. 25%, dark resistance of 100K�, maximum operating voltage of 150V and with diameter of 2. 3cm.

The light-detecting resistor (LDR) works on the principle of change in the resistance for the change in the intensity of the



incident light [12]. This property of the LDR is mainly used in proposed method where in the laser light beam is incident on the LDR and the output of the LDR is connected to the electronic circuitry for measurement and further processing the physical parameter.

2. 3 Principle of measurement

The principle of measurement used in the proposed technique is quite different from the already reported fibre-optic liquid level sensors techniques [2-6].

When a laser beam is incident on the surface of a liquid, part of the incident light is reflected from the liquid surface, part of it undergoes refraction at the air-liquid interface and the remaining light is scattered in the bulk of the liquid and also on the liquid surface. The beam of light that is reflected from the liquid surface is of interest in liquid level measurement.

Fig. 1. Principle of measurement

The reflection coefficient of liquid level can be calculated from Fresnel reflection law.

�lq = [�n� − n��/�n� + n��]^�� (1)

Where #�= refractive index of medium in which incident beam is travelling. #�= refractive index of medium at the surface of which reflection occurs.

For air-water interface $%& at 18˚ Celsius is 0. 02 [3]. When a 5Ww laser is used the optical intensity of the reflected beam is around 100µw.

The laser beam is incident on the liquid surface at an angle ‘i’ as shown in Figure-1. A distance ‘D’ on a measurement plane separates the incident beam position and reflected beam position. Two LDRs are moved at a constant velocity ‘V’ on the measurement plane, at the position of incident beam the LDR1 triggers a timer ON and at the position of the reflected beam the LDR2 triggers the timer OFF, if ‘t’ is the time indicated by the timer (i. e. time taken by the detector to move from incident beam position to reflected beam position) then distance ‘D’ covered by the sensor head between the incident point and the reflected point of the laser beam on the liquid surface is deduced as;

' = (��)��2�

The liquid level (L) of the liquid in the process tank is determined using,

+ = , − ℎ��3�

/ℎ010�ℎ = '2 � tan 1

, = ℎ045ℎ)�67�)ℎ0�81690::�);#<

The time taken by the moving sensor between the two beam points is measured and processed to calculate the liquid level using ARM-controller and displayed in 16 x 2 LCD display.

3. METHOD

3. 1. Process Tank Design

The measurement range and the angle of incidence of the incident beam primarily influence the design of tank. For a measurement range of 0-30cm and an angle of incidence of 30 degrees, the measurement range on the measurement plane is given by,

' = 2��ℎ�� tan 29. 5��4�

Where h = distance of measurement plane from maximum level of 30cm. With h = 40cm, D = 45. 26cm. The width of the tank is selected to be greater than 45. 26cm; it is taken as 60cm. The height of the tank is chosen to be greater than 40cm; it is taken as 60cm. For convenient machining the tank is designed with as 60cm x 60cm x 60cm cube. The prototype tank is fabricated using a 1. 3mm thick iron sheets. The mouth of the tank is closed using a 60cmx 60cm iron lid.

3. 2 Design of Measurement Head

The measurement head comprises of a lead screw, guide rod, with its housing bearing and D. C. motor to rotate the lead screw and slide element to hold detector. The mechanical system is fixed into the tank wall. The lead screw has square threading with a pitch of 6mm. It is fitted into a square block housing with bearings. The lead screw is coupled externally to a dc motor having 200 rotations per minute. Gears are used to

Nonintrusive M

55 | Dayanan

reduce the rotation of the lead screw to 5minute. The dc motor used is operated at 12Vdtorque of 3kg. The slide element holding the Linto a nut which passes through the lead screwThe slide element travels with a velocity oSchematic diagram of proposed prototype mvarious parts of measurement head are shown i

Fig. 3. 1. Proto type Model

3.3 Signal Conditioning Circuitry

The schematic diagram of signal conditioning in Figure-3. 2. It consists of voltage amplifier The change in resistance of the LDR due to laser beam (incident and reflected beam) is to voltage pulse. In order to realize this series LDR’s are connected as a feedback resistor in amp voltage amplifier with an input voltage ofdetector passes through the position of laoperational amplifier produces the output volt3V to 3. 5V. A comparator with a fixed refer2V is fed along with the output of the operatThe output of the comparator is initially high when the op-amp output goes to 3. 5V the comgoes low (logical = OV).

Fig. 3. 2. Signal conditioning circu

Method for Liquid Level and Volume Measurement


52 rotations per Vdc, 3A and has a

LDR’s is welded ew and guide rod.

of 0. 47cm/sec. method showing in Figure-3. 1.

circuit is shown and comparator. the incidence of be obtained as a combination of an inverting op-

of 0. 5Vdc. As the laser beam, the

oltage between 0. ference voltage of rational amplifier. (logical 1= 5V),

comparator output

rcuit

Similarly, when the second LDRbeam, the operational amplifier ou0. 045V to 1. 3V. The comparatorthis case is 0. 7V. So whenever outhan 0. 7V the comparator outputpulse is produced whenever the dlaser beam.

The output of the two comparatorsThe output of the gate is high only comparators is high. Thus a vowhenever the detector system crovoltage pulse is used as an externalcontroller as shown in circuit schem

Fig. 3. 3. Circuit Schem

The relay driver circuit is used position of LDR mounted on slideither zero volts or five voltsmicrocontroller ports for driving thIC 74LS244. The output of this is One is used to switch ON/OFF the to change the polarity if the shaftdirection motors controlled througmoving head reaches one extreme is directed in reverse direction.quantity calibrated and displayed in

4. RESULTS AND DISCUSSIO

To validate the proposed work usreal time experiments were carprototype tank of dimensions 60cexperimental results obtained are pl

Experimental result shows the sensitivity of 7mv/ml for water waThe table, 4-1 and 4-2 shows thereadings and error for two liquidslurry.

R passes over the reflected output voltage changes from or reference voltage given in output of the op-amp is more ut goes low. Thus a voltage detector system crosses the

ors is feed to the AND gate. when the output of the two

voltage pulse is produced rosses the laser beam. This al interrupt to drive the ARM ematic diagram Figure 3. 3.

ematic diagram

to drive and control the lide. The control voltage of lts is taken out from the

the relay through the driver connected to pair of relays. DC motor and other is used

aft movement. And also the ugh the software. When the

part of the tank, the motor n. The measured physical in 16 x 2 display

ION

using non-intrusive method, arried out with developed

60cm x 60cm x 60cm. The plotted in Figure 4a-4c.

accuracy of ±3% with was produced and calibrated. the actual reading, obtained uid containers of water and



Experimental plots shown in the Figure, 4a-4c has also confirmed sensitivity degradation with a decrease in the transparency of liquids. At the same time, it is also found to be higher for liquids having higher refractive indices. The results obtained by the proposed method shows highly linear with resolution of 10mV/ml approximately.

TABLE 4-1: Results comparison of level and volume for water

Liq

uid

Typ

e

Dip

stic

k va

lue

in m

m

Lev

el m

easu

red

in m

m

Lev

el E

rror

in m

m

Act

ual v

olum

e of

liqu

id in

m

3

Vol

ume

mea

sure

d in

m3

Err

or in

in m

3

wat

er

10 9. 78 . 22 0. 0036 0. 00352 0. 08 ∗ 10-3

25 24. 72 . 28 0. 009 0. 00889 0. 10∗ 10-3

50 49. 39 . 61 0. 018 0. 01778 0. 220∗ 10-3

75 74. 20 . 8 0. 027 0. 02671 0. 3∗ 10-3

100 98. 68 1. 32 0. 036 0. 03552 0. 475∗ 10-3

150 146. 20 3. 8 0. 054 0. 05263 1. 36∗ 10-3

200 195. 30 4. 7 0. 072 0. 07030 1. 7 ∗ 10-3

TABLE 4-2: Results comparison of level and volume of slurry

Liq

uid

Typ

e

Dip

stic

k va

lue

in m

m

Lev

el m

easu

red

in m

m

Lev

el E

rror

in m

m

Act

ual v

olum

e of

liq

uid

in m

3

Vol

ume

mea

sure

d in

m

3

Err

or in

in m

3

Slu

rry

10 9. 2 0. 8 0.

0036 0. 00331 0. 29∗ 10-3

25 23. 8 1. 2 0. 009 0. 00856 0. 432∗ 10-3

50 48. 4 1. 6 0. 018 0. 01742 0. 576∗ 10-3

75 72. 9 2. 1 0. 027 0. 02624 0. 756∗ 10-3

100 97. 41 2. 59 0. 036 0. 03506 0. 933∗ 10-3

150 145. 2 4. 8 0. 054 0. 05227 1. 73∗ 10-3

200 194. 6 5. 4 0. 072 0. 07005 1. 944∗ 10-3

The measurement technique presented in this paper is a very simple, low cost, and innovative technique. The performance of the system can be improved by further optimization and fine-tuning in certain key aspects related to the laser light focusing arrangement and the sensitivity of the detector.

5. CONCLUSION

The proposed work has the following advantage; it provides continuous measurement of liquid level with wide measurement dynamic range up to 250mm with liquid level sensitivity of approximately 10mV/ml. It is non-intrusive technique helps measure the level and volume of corrosive and other highly acidic liquids.

The hardware and software strategies chosen assure the following:

High static performance; the accuracy and repeatability being contained within the transducer resolution; Acceptable dynamic performance; it strongly depends on the liquid viscosity but could be improved by either increasing the ADC conversion rate or changing the software strategy; High reliability; a further improvement could be obtained by increasing the number of lasers.

0 20 40 60 80 100 120 140 160 180 200 220

30

40

50

60

70

80

90

100

110

120

LDR

Res

ista

nce

in K

Ohm

s

Liquid level in mm

Water Slurry

Fig. 4a. Experimental plots for two liquid containers: (a) LDR resistance variation with water and slurry level;

Nonintrusive Method for Liquid Level and Volume Measurement


0 50 100 150 2000.0

0.2

0.4

0.6

0.8

1.0

1.2

Out

put V

olta

ge

Liquid level in mm

Water Slurry

Fig. 4b. Experimental plots for two liquid containers: (a) proportional DC voltage with water and slurry level;

Fig. 4c. Experimental plots of sensitivity forliquids;.

REFERENCES

[1] P. Nath, P. Datta, and K. C. Sarma, “All fiber-optic sensor for liquid level measurement,” Microw. Opt. Technol. Lett. , vol. 50, no. 7, pp. 1982–1984, Jul. 2008.

[2] W. Zhang, K. Sudgen, S. Grice, and I. Bennion, “Liquid level sensing by use of digital formatted optical spectrum spreading technique,” J. Eur. Opt. Soc. , vol. 09022, pp. 1–4, 2009

[3] P. Nath, H. K. Singh, P. Datta, and K. C. Sarma, “All fiber-optic sensor for measurement of liquid refractive index,” Sensors Actuators A, Phys., vol. 148, pp. 16–18, 2008.

[4] F. Reverter, X. Li, and G. C. M. Meijer, “Liquid-level measurement system based on a remote grounded capacitive sensor,” Sens. Actuators A, Phys. , vol. 138, no. 1, pp. 1–8, Jul. 2007.

[5] K. Romo-Medrano and S. N. Khotiaintsev, “An optical-fibre refractometric liquid-level sensor for liquid nitrogen,” Meas. Sci. Technol., vol. 17, pp. 998–1004, 2006.

[6] F. Pérez-Ocón, M. Rubiño, J. M. Abril, P. Casanova, and J. A. Martínez, “Fiber-optic liquid-level continuous gauge,” Sens. Actuators A, Phys. , vol. 125, pp. 124–132, 2006.

[7] S. E. Woodard and B. D. Taylor, “A wireless fluid-level measurement technique,” Sens. Actuators A, Phys. , vol. 137, no. 2, pp. 268–278, Jul. 2007.

[8] C. Vázquez, J. Garcinuño, J. M. S. Pena, and A. B. Gonzalo, “Multisensor system for level measurements with optical fibers,” in Proc. 28thAnnu. Conf. IEEE Ind. Electron. Soc. , 2002, pp. 5–8.

[9] S. Khaliq, S. W. James, and R. P. Tatam, “Fiber-optic liquid-level sensor using a long-period grating,” Opt. Lett. , vol. 26, pp. 1224–1226, 2001.

[10] H. F. Norton, Handbook of Transducers. Englewood Cliffs, NJ: Prentice Hall, 1989.

[11] G. C. Barney, Intelligent Instrumentation—Microprocessor Application in Measurement and Control, 2nd ed. New Delhi, India: Prentice- Hall, Aug. 1992, 0-87692-783-5.

[12] A. K. Shawney, A Course in Electrical and Electronic Measurements and Instrumentation, 18th ed. New Delhi, India: DhanpatRai, 2010, pp. 1408–1411, 81-7700-016-0.

[13] J. M. Senior, Optical Fiber Communications—Principles and Practice, 2nd ed. New Delhi, India: Prentice-Hall, 2005, p. 873, 81-203-0882-4.

[14] J. P. Bently, Principles of Measurement Systems, 3rd ed. Singapore: Longman Singapore Publishers (Pte) Ltd. , 1995.

[15] G. Betta, A. Pietrosanto, and A. Scaglione, “A Gray-code-based fiber optic liquid level transducer,” IEEE Trans. Instrum. Meas., vol. 47, no. 1, pp. 174–178, 1998.



Analysis of Routing Protocols in Ad-hoc Networks Using NS2 Simulator

Bindu Bhargavi S.M.1, Rekha Jayaram2, Prathima Mabel J.3, Manasa Manjunath4 1,2,3Assistant Professors, Department of Information Science and Engineering,

Dayananda Sagar College of Engineering, Bangalore, 4Clemson University, Clemson 1bindu. [email protected], [email protected], [email protected], [email protected]

Abstract: A mobile ad-hoc network is an infrastructure-less network that connects mobiles nodes. Frequent changes in the location of these nodes require best possible performance management mechanisms. Various routing protocols are available that perform optimal routing of data. These protocols are categorized as proactive and reactive. This paper presents an analysis of application specific performance metrics of these protocols. In this proposed work, DSDV and DSR routing protocols are implemented using NS2 for variable queue length and performance is compared. We also highlight the efficiency of AODV, a reactive routing protocol which is a combination of DSDV and DSR.

Keywords: Ad-hoc networks, mobile nodes, proactive routing, reactive routing, DSDV, DSR, AODV.

1. INTRODUCTION

With the development of communication networks the need for wireless networks has increased over time with effective capabilities for efficient communication and mobility between nodes. A wireless network is composed of a collection of nodes using a wireless data connection for communicating between the nodes. These nodes are primarily responsible for sensing the environmental data and further transmit the data over the network to other nodes.

A wireless network can be categorized as an infrastructure based network and infrastructure less network. The network which requires wired components along with gateways and base stations form the infrastructure based network. The base stations are similar to central sink nodes which collect the data from over several nodes across the network. On the other hand the infrastructure less network, also called as ad hoc network consists of ad hoc nodes. The functionality of each node is to collect the sensed data and transmit the same all over the network such that sensed data reaches the concerned destination nodes. The characteristic features of ad-hoc networks are that they are robust in nature making the network as flexible as possible. The mobile nature of each node aids in the transmission of data in a very competent manner. In order to carry out this there is a need for routing protocols which facilitate in establishing efficient, possible and shortest route

such that overall energy consumption in the network in minimized. The routing protocol has to also take care of resource utilization and resource conservation in the network. Ad hoc networks can be widely used in applications like search and rescue operations, data acquisition in case of terrestrial or unreachable landmarks, human policing and personal area networking [1].

Many routing protocols have been proposed so far involving the characteristic features of ad-hoc networks namely self configuration and maintenance capabilities, flexible scalability and decentralized architecture. The basic constraints required for the design of routing protocols are preserving the energy of the nodes along with estimation of an optimal path. The behavior of a routing protocol is better understood when it is implemented and a comparison is done with other protocols. This gives us an avenue for enhancements for a protocol. The paper is organized as follows: Section 2 deals with the detailed categorization of the routing protocols along with the prominent ones that are currently in use. Section 3 provides performance overview of existing routing protocols along with the associated set of routing metrics. Section 4 provides the parameters used for implementation including experimental results showcasing the most effective routing protocols in terms of throughput and end-to-end delay as the routing metrics, followed by conclusion and the open research issues is Section 5.

2. ROUTING PROTOCOLS

Routing is defined as the act of data transmission from a source to a destination through an optimized path. The activity of routing involves the discovery of an optimal path, establishment of the path followed by transmission of data to each node constituting the network. Based on these activities routing can be categorized as static routing and dynamic routing [2]. When the complete knowledge about the network is known prior to the routing activity then it is termed as static routing. The localized knowledge about the network is maintained in the form of routing table at each node. This table holds the necessary details required for routing. The



dynamic routing on the other hand requires the complete status of the network where each node in the network is solely responsible for the data transmission in the network [3]. The ad-hoc networks are dynamic in nature and they can be further classified into three major categories namely proactive protocols, reactive protocols and hybrid protocols. Proactive routing protocols or table-driven routing protocols require each node of the network to be aware of the network status with respect to which all the nodes undergo periodic updations of their routing tables [4]. Though only a few nodes participate in the routing process, this category of routing requires all nodes in the network to be updated and be ready to participate in the routing process. The effect of this is the reduction in overall performance of the network which can be expressed in terms of energy efficiency and bandwidth consumption. A classic example of a proactive protocol is the Destination-Sequenced-Distance-Vector (DSDV) routing protocol [5, 6, 7].

Fig. 1. Classification of routing protocols [1, 5, 9].

In reactive or On-demand routing, the routing process is initiated by the routing source node only when there is a need for data transmission in the network. This requires an

estimation of all possible routes from source node to the destination node and finally selecting the shortest optimal route for routing. The Ad-hoc On Demand Distance Vector (AODV) [8] and the Dynamic Source Routing (DSR) protocols are examples of reactive protocols. A detailed categorization of the routing protocols is represented in Fig 1.

Another class of routing protocols is the Hybrid protocols which are adaptive in nature and is a combination of the features of both proactive and reactive routing schemes.

3. EXISTING SYSTEM

The development of a suitable routing protocol for the application under development requires a detailed analysis of all the existing routing protocols in Wireless networks. The routing protocols proposed for ad-hoc networks are evaluated on various parameters such as throughput, packet delivery fraction (pdf), end-to-end delay, processing overhead and others.

A Review of current Routing Protocols for Ad-hoc Mobile Wireless Networks proposed by Elizabeth M Royer et al [1] provides a comparison of eight different routing protocols by taking into account the updates required to each node to perform an effective routing. Both bandwidth and battery power are considered as the scarce resource and performance analysis has been carried out with an aim of minimizing the use of these resources. This work provided an overview of the application areas based on the evolution of the routing protocols. The performance comparison of ad hoc routing protocols for the networks with node energy constraints was proposed by Anne Aaron et al [10] which reviewed the performance of two routing protocol classes with respect to energy consumption at each node of the ad hoc network in terms of its battery life and the maximum threshold reached. The evaluation of the protocols is dependent on the attributes namely mobility, traffic pattern, physical layer architecture, and the Medium Access Control (MAC) layer interfaces. The review was supported with simulation results which have proved that the reactive routing protocols perform better than a proactive routing protocol in case of high node densities. This is because of the occurrence of overhead due to more number of message exchanges between the nodes involved in routing. Xin Zhang [11] conducted the performance evaluation of the routing protocols in case of networks which are densely populated with around 50, 000 nodes. The primary routing metrics considered here were the size of the network, traffic load and mobility. The analysis included the study of reactive AODV and DSR using the Georgia Tech N/w Simulator (GTNets) [12, 13]. The effect of mobility on scalability of the network was investigated in terms of packet delivery ratio along with the variations in maximum speed and pause time. Based on the observations conducted the parameters selection was optimized.



The metric based study of the routing protocols was carried out in [14] using the NS2 [17, 18] simulator. The routing protocols considered for the study were AODV, DSR and DSDV and the behavior of the routing protocols were understood by changing the number of nodes, data rates, mobility rate of each node and the pause time. Packet delivery ratio, average end - to - end delay and routing load were the metrics taken into account. The performance analysis for Multicast routing protocols [15] like the On - Demand Multicast Routing Protocol (ODMRP), Multicast Ad-hoc On-Demand Distance Vector (MAODV), Protocol Independent Multicast (PIM), Multicast Open Shortest Path First (MOSPF) and Distance Vector Multicast Routing Protocol (DVMRP) were performed using Qualnet 6. 1 simulator. to understand the etiquettes with respect to the size of the multicast group and corresponding metrics achieved with multicast throughput, average end - to - end delay, average jitter, packet delivery ratio, control overhead and link utilization.

The performance evaluation of the routing protocols depends on the combination of the metrics because of which no clear result about the best routing protocol can be obtained. So there is a need for the analysis of the protocols to obtain a stable performance under variable network conditions.

4. PROPOSED SYSTEM

The proposed system involves the comparison of a proactive protocol i. e. DSDV and reactive protocols namely DSR and AODV. When a specific set of conditions are considered, the analysis [16] has shown that none of the proposed protocols outperforms the other under all the conditions. So it is very much necessary to understand behavior of these protocols under complex, dynamic topologies with nodes having high mobility speed. The features that are taken into account are the number of nodes, maximum moving speed of the mobile node, maximum number of connections, pause time of the node, data rate and range of movement of each node.

All these features are set for varying network conditions such that protocol has to provide a stable performance, higher efficiency, strong adaptability towards a large scale or small scale network topology. The overall system design involves the activities associated with the Data link layer, Network layer and Application layer. The network layer is responsible for laying down the topological constraints along with route calculation and packet forwarding, while the application and data link layer exchanges the data with other respective layers. The connectivity between the nodes is brought about as a peer-to-peer network using wireless links.

A modular approach of the proposed system provides us with detailed architecture of the system. The important modules identified are:

• Node Configuration

• Link development

• DSR

• DSDV

• AODV

A. Node Configuration

The network includes mobile ad-hoc nodes with bi-directional connection links. The queue interface type that is set for each node is either Queue, Drop Tail or a Priority queue. In this case Drop Tail condition is chosen to check whether the buffer capacity of the output queue has exceeded because of which the last packet received is dropped. If e new queue type is required it can be explicitly defined by the user. As each node is a wireless transceiver, it is designed as an Antenna model to transmit and receive the radio waves equally in all directions as in case of Omni directional antennas or in respective directions as seen in the directional antennas. This is provided with supporting APIs which uses the General Operation directors to create the suitable interface with number of nodes specified by the user.

B. Link

The link i. e. set up between the nodes is basically a wireless channel used to send data from one location to the other. In some cases, multiple nodes may have to access the same channel and this is taken care by the Medium Access Protocol (MAC) which brings about a co-operative usage of communicating bands. The network interface layer is a hardware interface using which a mobile node can access the channel. For a wireless application, the interface used is the Phy/Wireless Phy which has to facilitate in propagating the radio signals. The radio signals can either be data signals carrying messages, connecting signals and disconnecting signals. The strength of the signal transmitted is formulated as the radio propagation model which takes into account the frequency, distance coverage between the nodes and several other conditions.

C. DSR

The Dynamic Source Routing protocol is a simple and efficient on-demand routing protocol which is mainly suitable for Wireless Mesh Networks. Instead of using the localized routing table information, this protocol follows source - routing at the intermediate nodes between the source node and destination node are collected there by enabling an easier route discovery. Based on the collected information, an analysis is carried out for route selection. The routed packets have to include the address of all the intermediate paths along with the address of the destination node but this scenario results in a very high overhead in case of long paths. In order to overcome



this problem, DSR protocol forwards the data packets on a hop-by-hop basis. This protocol constitutes two major phases which are Route Discovery and Route Maintenance.

D. DSDV

Destination-Sequenced Distance-Vector (DSDV) Routing is a table-driven routing scheme for Ad-Hoc mobile networks based on the Bellman–Ford algorithm. It is adapted from the conventional Routing Information Protocol (RIP) to Ad-Hoc networks routing. It adds a new attribute, sequence number, to each route table entry of the conventional RIP. Using the newly added sequence number, the mobile nodes can distinguish stale route information from the new and thus prevent the formation of routing loops. In this protocol, each mobile node of an Ad-Hoc network maintains a routing table, which lists all available destinations, the metric and next hop to each destination and a sequence number generated by the destination node. Using such routing table stored in each mobile node, the packets are transmitted between the nodes of an Ad-Hoc network. Each node of the Ad- Hoc network updates the routing table with advertisement periodically or when significant new information is available to maintain the consistency of the routing table with the dynamically changing topology of the Ad-Hoc network.

F. AODV

Ad-hoc on demand distance vector routing protocol (AODV) is a reactive on-demand routing protocol which works towards providing an efficient routing protocol for transfer of data within a MANET. The functionality of this protocol is similar to that of Bellman-Ford algorithm [19] where a routing path establishment is initiated based on the request from the source node. This loop free routing protocol is made suitable for a mobile atmosphere but the route once selected has to be maintained until the end of data transmission of last data packet. The sequence numbers of the nodes are used as the primary entities for carrying out routing process and this enables in keeping the nodes active and loop free. The overhead caused during the routing process can be reduced to a great extent in this protocol as it uses a standard predefined set of control messages and does not follow source routing. We do not see a periodic updation of the protocol parameters as routing is not initiated unless there is a requirement. Each node is associated with a routing table which contains the details like destination IP address, sequence number of nodes that is followed along the path to be traversed, neighbor nodes that are capable of becoming a part of routing process and the termination time for each route. The overall functioning of this protocol is understood in terms of the usage of control messages which are defined as follows:

Route discovery: This control message is initiated whenever a source node has to transfer some data to the destination node

placed elsewhere in the network. So when the request is initially generated, routing table of the source node and its neighbor nodes are checked to search if there is a path that is already existing and in case if a path is present the packets are readily transferred to destination node. If in case there is no path existing between the source and destination, then a route request (RREQ) [20] message is transmitted to the neighbor nodes in return of a reply from those neighbor nodes within a predefined time or a timer. RREQ is composed of source node ID, destination node ID and a broadcast ID. The broadcast ID is used if a RREQ is transmitted more than once in the network for a particular transmission. The destination node on accepting the RREQ replies back sending back a route reply (RREP) therefore conforming the route that can be followed for data transmission. The traversal path of RREP back to the source node is made possible by maintaining an entry of the source node IP address, number of neighbor nodes existing on the path and the sequence number of nodes that are followed. Upon receiving RREP, the source node starts with data packet transmission.

Route maintenance: This feature is very much important as we are taking into consideration the nodes in an ad-hoc network which are dynamic in nature. The basic requirement of data transmission in such networks is that it requires the nodes to be stable, such that data transmission is made possible without any link breakages. The route error messages are used to indicate the erroneous conditions in the network. The source node has to stop with data transmission in case if a RERR message is generated.

5. SIMULATION RESULTS

The simulation setup for the proposed system with associated parameters and data transmission details are listed in the table 5. 1. For a queue length of 50, we can observe a huge difference in the average end-to-end delay.

TABLE 1: Throughput achieved for a queue length of 50.

Parameter DSR DSDV

No of packets generated 6259 13602

No of packets received 6206 13298

No of packets dropped 0 67

Packet delivery ratio 99. 1532% 97. 765%

Average end-to-end delay 79. 2549 ms 125. 8 ms

There is a minimal variation in case of packet delivery ratio but a huge difference is seen in case of the packets dropped. Similar tests are conducted for a queue length of 10 and the results are shown as in table 5. 2.



TABLE 2: Throughput achieved for a queue length of 10.

Parameter DSR DSDV

No of packets generated 6963 18986

No of packets received 6776 18348

No of packets dropped 87 258

Packet delivery ratio 97. 3144% 96. 6396%

Average end-to-end delay 57. 6259 ms 57. 8193 ms

When the queue length is 10, we can see that the packets dropped difference is much higher than when the queue size was 50. There is one percent change in packet delivery ration when the queue size is 10. The average end-to-end delay is not substantial.

Fig. 2. Transmission Window vs Time for DSDV.

This graph shows that in the beginning there is high packet delivery ratio as the transmission begins almost instantaneously. This is because DSDV maintains Routing tables at every node from the starting itself.

Fig. 3. Transmission Window vs Time for DSR

The graph below shows that it takes DSR protocol some time to create paths, once created the packet delivery ratio is quite high with less packet drops.

6. CONCLUSION

After performing the above experimentations and simulations and referring to the results in tables and graphs, it has been observed that when the number of nodes is less – DSR is more effective whereas when the number of nodes increases, DSDV performs better.

The Dynamic Source Routing protocol (DSR) provides excellent performance for routing in multi-hop wireless ad hoc networks. DSR has very low routing overhead and is able to correctly deliver almost all originated data packets, even with continuous, rapid motion of all nodes in the network. Hence, DSR has a better Delivery Ratio. A key reason for this good performance is the fact that DSR operates entirely on demand, with no periodic activity of any kind required at any level within the network and also because it exploits aggressive caching and maintains multiple routes to destinations. This cache can become a problem if we increase the mobility and simulation time as then routes will be changing more frequently and cache will have stale routes mostly, therefore, in that case it will not help DSR in better performance. This entirely on-demand behavior and lack of periodic activity allows the number of routing overhead packets caused by DSR to scale all the way down to zero, when all nodes are approximately stationary with respect to each other and all routes needed for current communication have already been discovered.

The Destination Sequenced Distance Vector (DSDV) can work with High frequency of updates and High routing overheads. Random Waypoint Mobility Model has been used in this study to generate node mobility; where we take an average of 10 randomly generated scenarios so to make a detailed Performance analysis. We find that the performance varies widely across different network sizes and results from one scenario cannot be applied to those from the other scenario. As far as Throughput is concerned, DSR performs better than DSDV even when the network has a large number of nodes. Overall, the simulations performed here shows that, DSR performs better when the number of nodes is small. Average End-to-End Delay is the least for DSDV and does not change if the number of nodes is increased.

Ad-hoc on demand distance vector routing protocol (AODV) is a combination of DSR and DSDV. Its performance is closely comparable with DSR because of its reactive nature. Route discovery is initiated only when there is a demand. The destination node replies to the first RREQ message that it receives thereby decreasing the end-to-end delay when compared to DSR. Further studies show that average packet loss rate and throughput of AODV are the key parameters that prove it to have higher efficiency and adaptability making it suitable for small scale networks. Each of these protocols can be used in their own scenarios and must be used in the same



way to get the best results for that network. Thus, no one protocol completely puts down any other.

7. OPEN RESEARCH ISSUES AND FUTURE ENHANCEMENTS

In future, this study can be improved by comparing more number of protocols under each group of protocols. More On-Demand protocols can be compared to add on to this study. Hybrid Protocols such as TORA (Temporally Ordered Routing Algorithm) can be evaluated and compared with these results. Also, more varying parameters like jitter, number of traffic sources, and other traffic than CBR (like TCP transfers); simulation area, measurements and estimation of power and energy consumption along with processing costs, route length, type of traffic, pause time, routing overheads, etc can be checked. We also look forward to further development of the protocols for quality of service (for real-time and non-real-time traffic), analysis of interworking functions for Mobile IP, intermediate route rebuilding and various interconnection topologies with fixed networks and the internet. Future search can be done on increasing the lifetime of the nodes by energy efficient load balancing techniques. As this study will grow, in the future, we can be able to know for certainty that which protocol would be under which circumstances or even with a creation of such a protocol, by learning from such comparative studies, which outperforms all the other protocols and works more efficiently under all possible conditions. Although the field of Ad-Hoc networks is rapidly growing and new developments are coming day by day, it might even give rise to many challenges that would then need to be tackled and evaluated.

REFERENCES

[1] Elizabeth M. Royer, Chai-Keong Toh, "A Review of current routing protocols for Ad-Hoc Mobile Wireless Networks" IEEE Personal Communications, April 1999.

[2] Krishna Gorantala, "Routing Protocols in Mobile ad-hoc Networks", Umea University, 2004.

[3] Scott M. Ballew. “Managing IP Networks with Cisco Routers”, Orilley 1st Edition, 1997.

[4] Charles E. Perkins. “Ad Hoc Networking”, Addision Wesley, 2001.

[5] Ginni Tonk, S. S. Tyagi, "Performance of Ad-Hoc Network Routing Protocols in Different Network Sizes", International Journal of Innovative Technology and Exploring Engineering

(IJITEE) ISSN: 2278-3075, Volume-1, Issue-2, July 2012. [6] C. E. Perkins and P. Bhagwat, “Highly Dynamic Destination-

Sequenced Distance-Vector Routing (DSDV) for Mobile Computers,” Comp.commun. Rev. , Oct. 1994, pp. 234–44.

[7] L. R. Ford Jr. and D. R. Fulkerson, “Flows in Networks”, Princeton Univ. Press, 1962

[8] C. E. Perkins and E. M. Royer, “Ad-hoc On-Demand Distance Vector Routing, Proc. 2nd IEEE Wksp. Mobile Comp. Sys. and Apps. , Feb. 1999, pp. 90–100.

[9] Xiaoyan Hong, Kaixin Xu, and Mario Gerla, “Scalable Routing Protocols for Mobile Ad hoc Networks”, 2002.

[10] Anne Aaron and Jie Weng, "Performance Comparison of Ad-hoc Routing Protocols for Networks with Node Energy Constraints", Stanford University, 2001.

[11] Xin Zhang, George F. Riley, "Performance of Routing Protocols in Very Large-Scale Mobile Wireless Ad Hoc Networks", Department of ECE, Georgia Institute of Technology, Atlanta, GA, 2005.

[12] G. F. Riley, “The Georgia Tech network simulator,” in Proc. ACM SIGCOMM Workshop on Models, Methods, and Tools for Reproducible Network Research, 2003, pp. 5-12.

[13] G. F. Riley, “Large-scale network simulations with GTNetS,” in Proc. 2003 Winter Simulation Conference, 2003, pp. 676- 684

[14] Vimalapriya M. D, Dr. Santhosh Baboo S. Reader, “Analysis of Metrics in Different Scenarios of Routing Protocols in Ad Hoc Networks", Department of Computer Applications, D. G. Vaishnav College Chennai, India, 2011.

[15] Kanwalpreet Kaur, Krishan Kumar Saluja, “Performance Analysis of Multicast Routing Protocols in Ad-Hoc Networks”, PIT, 2014.

[16] Ragb O. M. Saleh, Md Yazid Mohd Saman and M. Nordin A. Rahman, “A Simulative Comparison of AODV and DSR on-Demand Routing Protocols for Mobile Ad-Hoc Networks", 2014.

[17] Eitan Altman, Tania Jimenez, "NS2 Simulators for Beginners, Lecture Notes", 2003 - 2004

[18] Francisco J. Ros Pedro M. Ruiz, “Implementing a New MANET Unicast Routing Protocol in NS2 ", Dept. of Information and Communications Engineering University of Murcia.

[19] B. Karthikeyan, N. Kanimozhi, S. Hari Ganesh, "Analysis Of Reactive Aodv Routing Protocol For MANET", 2014 World Congress On Computing And Communication Technologies

[20] Daxesh N. Patel, Sejal B. Patel, Hemangi R. Kothadiya, Pinakin D. Jethwa, Rutvij H. Jhaveri, "A Survey of Reactive Routing Protocols in MANET", ICICES2014 - S. A. Engineering College, Chennai, Tamil Nadu, India



Smart Transportation System Using Big Data Analytics

Ajay Kumawat1, Hardik Singhi2, Anil T. Sagar3 1,2,3Dept. of Computer Science and Engineering, Dayananda Sagar Institutions, Bangalore, India

[email protected]; [email protected]; [email protected]

Abstract: The advent of Big Data era has transformed the outlook of numerous fields in science and engineering. The transportation arena also has great expectations of taking the advantage of Big Data enabled by the popularization of Intelligent Transportation Systems (ITS). In this study, the viability of a proactive real-time traffic monitoring strategy evaluating operation and safety simultaneously was explored. The objective is to improve the system performance of urban expressways by reducing congestion and crash risk. It can be used for smarter navigation and elimination of jams. A most significant use will be during emergency and ambulance navigation. when raised an emergency, the system automatically roots for the shortest path, halting other cross paths for few minutes till ambulance gets the way through, can be proved as life savior.

1. INTRODUCTION (THE PROBLEM)

Over the past years “Big Data” as the word suggests, has afected larger communities. the millions of terabytes of data is day by day ever incereasing. This can be utilised for better purposes like- smarter transportaion system.

The aim is to develop an intelligent decision-making framework for handling conflicting objectives in traffic management and to visualize them. This unified platform will have tools to harness big data analytics and decision making capabilities and enable policy makers to balance multiple contradictory objectives.

It can prevent injuries and avoid delays. Can route ambulance when raised an emergency. It can plot out the shortest travel routes, avoid traffic snarls and estimate least time they will arrive at their destinations.

2. DATA GATHERING

Analyzing traffic jams in Japan using multiple sources, including traffic data from navigation maps such as Google maps.

The information from GPS mixed with satellite connected to sensors at cross sections roads can be proved helpful in data

collection. The classic system does not have this power, but can contribute in supplementing data.

3. RISKS

With growing roadways and traffic, experts predict that the exhaust gases from vehicles—already implicated in heart disease and cancer—can injure brains cells and drivers. U. S. traffic annually spend 140 hours, and meanwhile, a 2013 survey by workplace solutions suggests that traffic congestion and crowded public transportation systems are top causes of stress and declining productivity among employees.

traffic congestion also can and does have a significant impact on quality of life and the economy in many major cities around the world. For example, a 2013 report from the Texas A&M Transportation Institute found that the financial cost of congestion in the United States in 2011 was $121 billion, translating to $818 per U. S.commuter.

} Wasted Fuel: 2. 9 billion gallons } CO2 Emissions: 56 billion pounds

The expenses in 2013 totaled $200 billion (0. 8% of GDP) and is expected to rise to nearly $300 billion by 2030



4. POSSIBLE SOLUTIONS

Implementing unified data center. Deploy servers running on Intel processor series and a storage space information.

Implementing Hadoop. Use of HBase and Hadoop system.

Implementing Trust way supervision system. Open platform for analytics can be used for data mining.

In-memory computing technique, can analyze large amounts, was able to process the hundreds of million records in just over one second, as compared traditional relational database technology.

“Magnetic sensors in the road at every intersection send real-time updates about the traffic flow. The computer system, which runs software the city itself developed, analyzes the data and automatically makes second-by-second adjustments, adapting to changing conditions and using a trove of past data to predict where traffic could snarl, all without human involvement.”

The system can automatically adjust the delay between lights and get back on schedule.

5. IMPLEMENTED METHODS

These are previously implemented projects on similar approach:

Google’s cars project IBM’s project variety.

Data-driven network management

6. CASE STUDY

Case studies of big data analytics in network management. Here, case studies of the use of big analytics in areas of network management such as network planning and engineering, and network operation are presented.

Case I: Network operation (failure prediction and detection)

By analyzing a huge amount of unstructured data (syslog messages, SNS messages, etc. ), we can develop methods of detecting failures that cannot be detected by existing network failure monitoring systems, as well as methods of analyzing the root causes of such failures. Machine learning is used to extract useful information for network operation (Fig. 2).

Fig. 2. Network failure detection and root cause analysis for failure prediction.

Ajay Kumawat, Hardik Singhi, Anil T. Sagar


We are also developing methods for analyzing Twitter messages (tweets). Twitter disseminates short messages in a real-time manner.

Case II: Network operation (security)

Security is becoming a major concern in network operation areas. The use of botnets is spreading widely, which is threatening Internet security. Blacklists are used to contain the botnets. To prevent users from inadvertently accessing botnets, we use a blacklist of command and control (C&C) servers of the botnets and block the communication from users to those C&C servers. The botnets are sophisticated enough to expand their coverage by creating and spreading new C&C servers throughout the Internet. To cope with these ever-expanding botnets, we need to maintain and update the blacklists of the C&C servers (Fig. 3). We therefore analyze traffic data to improve our coverage of the blacklists.

Fig. 3. Communication pattern analytics used against botnets

We have developed a method to find new unknown C&C servers by exploiting the rule of thumb that a user’s personal computer (PC) that accesses a C&C server of a botnet and is consequently infected by malware, is likely to access other C&C servers as well [2]. We analyze traffic data and calculate the co-occurrence score between the server communicating with an infected server and the already-known C&C server communicating with the infected server. We assume that a server with a high co-occurrence score is a newly discovered C&C server. We expand the blacklist by registering the newly discovered C&C servers in the blacklist.

Case III: Network planning and engineering

The Internet is becoming ubiquitous and is playing a fundamental role as a social infrastructure. Accordingly, numerous Internet applications have emerged, and the complexity of traffic carried over the telecommunication networks is increasing. In particular, video traffic accounts for a large amount of the total traffic. We therefore need a

detailed understanding of video traffic to implement effective network planning and engineering. Video services are categorized as broadcast, VoD (video-on-demand), and OTT (over-the-top). These services have a large amount of content to meet customer needs. Customer behavior is quite different from that of traditional telephone networks. Analyzing customer behavior in their viewing of video services is crucial to understand video traffic. Thus, we take into consideration concurrent weather and temperature data and social event information to better understand video traffic.

7. PROJECT RATIONALE

The problem of traffic management is both a complex and challenging issue, often involving numerous factors and variables. At times, they are intertwined; in certain other situations they are contradictory to each other or both intertwined and contradictory to each other. Even though with the flood of data available from different sources which offer us an opportunity to turn to analytics solutions to extract meaning from the huge volume of data, we need to determine the relationship between those variables. Once the relationship between those variables is obtained, its objective function is determined. Most often, those objective functions obtained are conflicting to each other, and they need to be simultaneously optimized. Hence, we need to have a tool to handle this situation, to consider the trade-off between them.

Project Methodology With big data and analytics (both descriptive and predictive), practitioners are able to derive much useful insights pertaining to the various data sets.combined with association rules, relationships between specific values of categorical variables in large data sets are detected. These powerful multivariate exploratory techniques enable analysts to uncover hidden pattern in large data sets. However, what is lacking here is the capability to visualize how one objective function affects another objective function, and the trade-off between them.

8. TECHNICAL RESULTS

• Achieved high I/O processing function. The Intel Xeon processor E5 series enhances I/O processing. Now a single server now allow synchronous transmission of a 500KB picture with an average speed of 250 times per second, or asynchronous concurrent storage of 2, 000 times.

• Provided greater-performance HBase. Hadoop enabled complex data queries in the vehicle monitoring system. It now takes fewer than a second to accurately search for plate numbers or the driving record of a vehicle from over 2. 4 billion records.



• Improved capacity. Apache Hadoop provided a mass data storage solution with high fault tolerance and throughput, allowing reliable storage.

• Easy access to vehicle analysis data. Investigating traffic cases that require complex inquiries, such as data from multiple checkpoints or multiple vehicles, now takes only 10 seconds.

• Improved supervision of motor vehicles. Can now easily retrieve plate numbers and the driving track of a passing vehicle from the over 2. 4 billion records in the system.

9. CHALLENGE

In transportion, data collection is often inconsistent, which may exploit fully big data. This is case for installing devices and gathering data takes certain time.

10. CONCLUSION

By implementation of Apache Hadoop with in memory computing technique, we are able to Deploy Trust way key

vehicle dynamic supervision system and data which provides high end intelligent decisions on traffic management and a powerful system with sophisticated data that could be helpful anytime in future. Less traffic cognition and successful implementation of features including cost elimination, fuel conservation, managed roads, time and traffic management etc are implemented.

REFERENCES

[1] Bloomberg L. P. Research update: Drivers Avoid traffic Jams with Big Data and Analytics BY JOE MULLICH september 2013

[2] CASE STUDY: Intel Processor E5 Xeon Series [3] POSTNOTE: Big and Open Data in Transport Number 472

July 2014 [4] NTT Technical review: Applications of Big Data Analytics

Technologies for traffic and Network Management Data––Gaining Useful Insights from Big Data of traffic and Network Management

[5] http://opengovernanceindia. org/gigusbf/quick-glance-on-bangalore-city-traffic-data



Design of Reconfigurable Rectangular Patch Antenna using PIN Diode

Banuprakash R.1, Dr. Hariprasad S.A.2, Sai Raghuvamshi 3 1Department of TE, BMS Institute of Technology and Management, Bangalore, India

2 Department of E&CE Dayanand Sagar University, Bangalore, India 3Department of E&CE, University of Florida, Gainesville, FL

1r. bhanuprakash@bmsit. in; [email protected], 3sairaghuvamsi@ufl. edu

Abstract: In this paper, frequency reconfigurable rectangular antenna is presented. The proposed antenna consists of a Microstrip patch antenna. In this design it makes use of two patches with different dimension in order to tune for multiple frequencies. The antenna design make use of micro strip feeding technique, PIN diode is used for switching between the patches to tune for different frequencies. The antenna is capable to reconfigure up to two different frequencies i. e. , 9. 2GHz and 6. 2GHz when no PIN diode is connected. When a single PIN diode is connected it will generate a frequency of only 6. 2GHz and when two PIN diodes are connected it will reconfigure itself to generate two different frequencies i. e. , 2. 2GHz and 3. 5GHz. This antenna is simulated and measured results are used to demonstrate the performance of the antenna.

Keywords: Reconfigurability, Feeding Technique, Microstrip Antenna and Rectangular Patch

1. INTRODUCTION

With rapid development of wireless communication especially in depth research on MIMO techniques Reconfigurable antennas gaining great attention. Different characteristics (such as resonant frequency, radiation patterns, polarization, etc) of these novel antennas can be reconfigurable through the change of the structures. . Recently, frequency reconfiguration has attracted significant attention due to the introduction of future wireless communication concept such as cognitive radio which employs wideband sensing and reconfigurable narrowband communication [1]. Frequency reconfigurable antenna has the reconfiguration of the resonant frequency by the change of the structure, while the radiation patterns and polarization remain unchanged. So, frequency reconfigurable antenna can be applied among a very wide arrangement of frequency band or among multiple frequency bands.

Depletion of available frequency resources has been one of the major problems in wireless communication systems. To solve this, cognitive radio is considered to be a promising solution [5]. In the cognitive radio system, antenna characteristics are required to be as equivalent as possible at any selected frequency band. Various types of effective reconfigurable

antennas were proposed [6]. Microstrip patch antennas have found extensive application in wireless communication system owing to their advantages such as low profile, conformability, low-cost fabrication and ease of integration with feed networks[3]. Electrical reconfiguration techniques are based on the use of switches to connect and disconnect antenna parts as well as to redistribute the antenna currents. Radio frequencymicro-electromechanical systems (RF-MEMS) have been proposed for integration into reconfigurable antennas since 1998[7]. P-i-n diodes or varactors have appeared to be a faster and a more compact alternative to RF-MEMS. The switching speed of a p-i-n diode is in the range of 1–100 nsec[8].

Several methods are used to feed microstrip antennas. These methods can be classified into two categories- contacting and non-contacting. In the contacting method, the RF power is fed directly to the radiating patch using a connecting element such as a Microstrip line[4]. In the non-contacting scheme, electromagnetic field coupling is done to transfer power between the Microstrip line and the radiating patch. The four most popular feed techniques used are the Microstrip line, coaxial probe (both contacting schemes), aperture coupling and proximity coupling (both non-contacting schemes).

In this design we used contacting feed such as Microstrip Line Feed. In this type of feed technique, a conducting strip is connected directly to the edge of the Microstrip patch. The conducting strip is smaller in width as compared to the patch and this kind of feed arrangement has the advantage that the feed can be etched on the same substrate to provide a planar structure. However as the thickness of the dielectric substrate being used, increases the surface waves and also spurious feed radiation, which hampers the bandwidth of the antenna. The feed radiation also leads to undesired cross polarized radiation. This method is advantageous due to its simple planar structure. In this paper, a reconfigurable rectangular patch antenna using PIN diodes is presented to provide multiple- frequency operation for various applications. Switches can be modeled at different levels of complexity, depending on the



required accuracy and available computational resources. At the basic level, the switch can be modeled simply by a metal tab; switching between the ON and OFF states is then just a matter of simulating the model with and without that piece of metal.

2. ANTENNA DESIGN CALCULATION

Step 1: Calculation of Width (W)

The width of the Microstrip patch antenna is given as:

Where, c is velocity of light, f0 is Resonant Frequency & �r is Relative Dielectric Constant, of course other widths may be chosen but for widths smaller than those selected according to the width equation [3], radiator efficiency is lower while for larger widths, the efficiency are greater but for higher modes may result, causing field distortion.

Step 2: Calculating the Length (L)

Effective dielectric constant (�eff)

Once Width is known, the calculation of the length which involves several computations; the first would be the effective dielectric constant [2]. The dielectric constant of the substrate is much greater than the unity; the effective value �eff will be closer to the value of the actual dielectric constant �r of the substrate. The effective dielectric constant is also a function of frequency. As the frequency of operation increases the effective dielectric constant approaches the value of the dielectric constant of the substrate is given by:

Effective length (Leff) The effective length: This can be found by

Length Extension (�L) Because of fringing effects, electrically the micro strip antenna looks larger than its actual physical dimensions. For the

principle E – plane (x-y plane), where the dimensions of the path along its length have been extended on each by a distance �L, which is a function of the effective dielectric constant and the width-to-height ratio (W/h). The length extension is:

Calculation of actual length of patch (L)

Because of inherent narrow bandwidth of the resonant element, the length is a critical parameter and the above equations are used to obtain an accurate value for the patch length L. The actual length is obtained by:

Feed Point Location

After selecting the patch dimensions L and W for a given substrate, the next task is to determine the feed point (x, y) so as to obtain a good impedance match between the generators Impedance and the input impedance of the patch element. It is observed that the change in feed location gives rise to a change in the input impedance and hence provides a simple method for impedance matching.

From the above equation we see that if the feed is located at x = xf and 0 � yf � W, the input resistance at resonance for the dominant TM10 mode can be expressed as:

Where xf is the inset distance from the radiating edge and Rr is the radiation resistance at resonance when the patch is fed at a radiating edge [5]. The inset distance xf is selected such that Rin is equal to the feed line impedance, usually taken to be 50�. Although the feed point can be selected anywhere along the patch width, it is better to choose yf = W/2 if W L so that TM0n (n odd) modes are not excited along with the TM10 mode. Determination of the exact feed point requires an iterative solution. Below equation provides a useful guideline for the purpose. Kara has suggested an expression for xf that

Banuprakash R., Hariprasad S.A., Sai Raghuvamshi


does not need calculation of radiation resistance. It is approximately given by

3. ANTENNA DESIGN

The basic structure of the proposed antenna, shown in Fig. 1, consists of three layers. The lower layer, which constitutes the ground plane, covers the partial rectangular shaped substrate with a side of 42×46mm. The middle is the substrate, which is made of Teflon, has a dielectric constant r=2. 1 and height 1. 5 mm. The upper layer consists of two rectangular patches. The rectangular patches have dimensions 33×16. 8 mm and 33×17. 5 mm that covers a portion of the substrate. The patch is fed by a Microstrip line with 50� input impedance.

The three essential parameters for the design of a rectangular Microstrip Patch Antenna:

1. Frequency of operation (f0): The resonant frequency of the antenna must be selected appropriately. The Antenna was designed for radar communications, satellite uplink, Mobile Broadband Wireless Access and WiMAX. Hence the antenna designed must be able to operate in these frequency ranges. The resonant frequencies selected for our design are 6. 2 and 9. 2 GHz.

2. Dielectric constant of the substrate (�r): The dielectric material selected for our design is Teflon which has a dielectric constant of 2. 2. A substrate with a high dielectric constant has been selected since it reduces the dimensions of the antenna.

3. Height of dielectric substrate (h): For the micro strip patch antenna to be used in wireless application, it is essential that the antenna should be compact. Hence, the height of the dielectric substrate is selected as 1. 5mm.

Fig. 1. The proposed geometry of patch antenna

Simulations were performed using HFSS. Convergence was tested for a number of times. Once convergence was obtained simulations were conducted in order to obtain sweep frequency response. Initially we started with one patch with no PIN diodes, then went with both the patch operating with input to the second patch are through only one PIN diode, then finally checked with two PIN diodes, however it was observed that in order to achieve proper impedance patch position and dimensions need to be adjusted accordingly.

The proposed antenna is designed to operate at two different frequencies with single patch with no diode ON at the first stage, then when both the patches are connected together using one PIN diode with specification of PIN diode are Resistance-1. 2Ohms, Inductance-0. 6nH and Capacitance-0. 3pF produces one frequency that’s second stage, Finally in last stage when two PIN diode are used it produces still two more frequencies. In this antenna microstrip feed is used with the length of 13mm and width of 1. 5mm, the gap between the two patches is 0. 7mm.

The designed antenna is as shown figure 2, Blue color indicates substrate (S), dark blue color indicates patch (P1) and (P2), red indicate feed (F), pink indicates PIN diodes (D1 and D2) and white indicates feed port (Fp). In the design feed is given to patch P1.

Fig. 2. The designed microstrip rectangular patch antenna.

4. DESIGN ANALYSIS

There are three stages of operation each one is discussed below:

Stage 1: In this case when both PIN diodes D1 and D2 are OFF, only patch P1 is radiating and patch P2 is not operating or radiating, is as shown in the figure 3.



Fig. 3. Patch P1 is radiating.

Stage 2: In this case when PIN diode D1 is ON and D2 is OFF, now both patch P1 and P2 is radiating, the current flow from patch P1 to P2 is through diode D1, is as shown in the figure 4.

Fig. 4. Patches P1 and P2 are radiating connected using a single PIN diode D1.

Stage 3: In this case when both PIN diode D1 and D2 are ON, now both the patches P1 and P2 are radiating, the current flow from patch P1 to P2 is through both diodes D1 and D2, is as shown in the figure 5.

Fig. 5. Patches P1 and P2 are radiating connected using two PIN Diodes D1 and D2.

5. RESULTS AND OBSERVATION

We observe what happens in each case discussed in above stages,

Stage 1: In this case we can observe that the antenna can produce two frequencies such as 9. 2GHz and 6. 2 GHz which are used for radar communication and satellite uplink communication is as shown in figure 6(a) with a gain of -26. 02 and -16. 33 dB. For these two frequencies we can observe the voltage standing wave ratio as 1. 10 and 1. 37 which is within the acceptable range in practically (ideally should be 1) is as shown in figure 6(b).

Fig. 6(a). Plot of frequency v/s gain(dB)

Fig. 6(b). Plot of frequency v/s VSWR.

Banuprakash R., Hariprasad S.A., Sai Raghuvamshi


Stage 2: In this case we can observe that the antenna can produce frequency of 6. 2 GHz which is used for satellite uplink communication is as shown in figure 7(a) but in this the gain grows to -24. 22 dB from 16dB compare to previous stage. For this frequency we can observe the voltage standing wave ratio of 1. 07. Which is within the acceptable range in practically (ideally should be 1) is as shown in figure 7(b).

Fig.. 7(a): Plot of frequency v/s gain (dB).

Fig. 7(b). Plot of frequency v/s VSWR.

Stage 3: In this case we can observe that the antenna can produce two frequencies such as 3. 5GHz and 2. 2 GHz which are used for WiMAXand Mobile Broadband Wireless Access (MBWA) is as shown in figure 8(a) with a gain of -15. 66 and -8. 71 dB. For these two frequencies we can observe the voltage standing wave ratio as 1. 39 and 2. 36, which is within

the acceptable range in practically (ideally should be 1) is as shown in figure 8(b).

Fig. 8(a): Plot of frequency v/s gain(dB).

Fig. 8(b): Plot of frequency v/s VSWR.

From all the above results we can observe that the frequency will be maximum when all the PIN diode is OFF and with the more gain, as we connect one PIN diode between the patches the frequency slightly decrease and when we connect second PIN diode ( as both PIN diodes are connected) the frequency still reduces. Finally we can say that once PIN diode is connected we can reconfigure the antenna for different frequency applications from highest frequencies to lowest.

6. CONCLUSIONS

In this paper the aim is to design a Reconfigurable rectangular patch antenna and to study the responses of the same. The



antenna has been designed with two different patch dimensions, taking into consideration like patch dimensions, selection of the substrate, feeding technique and also the Operating frequency to design the antenna. The antenna is designed to operate in four different frequencies using PIN diode to switch between the patch in order to reconfigure the frequencies. By using PIN diode for switching we get three stages of operation and we studied all the three stages of design and also we observed their gain and VSWR with respect to the frequency in all the three stages in detail.

REFERENCES

[1] Mojtaba Fallahpour, Mohammad Tayeb Ghasr, and R. Zoughi, “Miniaturized Reconfigurable Multiband Antenna for Multiradio Wireless Communication” IEEE Transactions on Antennas and Propagation, Vol. 62, no. 12, December 2014, pp 6049 - 6059.

[2] I. H. Idris, M. R. Hamid, M. H. Jamaluddin, M. K. A. Rahim, J. R. Kelly, H, H. A. Majid “Single, Dual and Triple-band Frequency Reconfigurable Antenna”, RADIOENGINEERING, Vol. 23, no. 3, September 2014, pp 805-811.

[3] Pradeep Kumar Neha Thakur, Aman Sanghi”Micro strip Patch Antenna for 2. 4 GHZ Wireless Applications “International Journal of Engineering Trends and Technology (IJETT) – Volume 4 Issue 8- August 2013

[4] N. Romano, G. Prisco, F. Soldovieri “Design of a Reconfigurable Antenna for Ground Penetrating Radar Applications” Progress In Electromagnetics Research, Vol. 94, pp 1-18, 2009

[5] Linda E. Doyle, Essentials of Cognitive Radio, New York: J. Wiley & Sons, 2009.

[6] C. G. Christodoulou, Y. Tawk, S. A. Lame, and S. R. Erwin, Reconfigurable Antennas for Wireless and Space Applications” Proceedings of the IEEE, vol. 100, No. 7, pp. 2250-2261, July, 2013

[7] E. R. Brown, ‘‘RF-MEMS switches for reconfigurable intgrated circuits, ’’ IEEE Trans Microw. Theory Tech. , vol. 46, no. 11, pt. 2, pp. 1868–1880, 1998.

[8] G. H. Huff and J. T. Bernhard, ‘‘Integration of packaged RF-MEMS switches with radiation pattern reconfigurable square spiral microstrip antennas, ’’ IEEE Trans. Antennas Propag. , vol. 54, no. 2, pp. 464–469, Feb. 2006.

[9] Y. Tawk, J. costantine, and C. G. Christodoulou, ‘‘Cognitive radio antenna functionalities: A tutorial, ’’ IEEE Antennas Propag. Mag., vol. 56, no. 1, pp. 231–243, 2014.



AIMS & SCOPE:

International Journal of Electrical Sciences and Engineering (IJESE) aims at stimulating new research ideas and foster practical application from the research findings. IJESE provides a platform to share innovative original ideas in Electrical and Allied Science related subjects. It is an open access, international and multidisciplinary journal. Published in January and July every year, this bi-annual journal is dedicated to serve the society by quality research work. We encourage the submission of manuscripts in one of the form of reports of empirical studies, review papers, meta-analyses and theoretical position papers.

JOURNAL TOPICS: Contributions are welcome in the following topics but not limited to:

Computer & Information Science & Engineering: Image processing & computer vision, Software engineering, big data analytics, Information storage management, knowledge management, web semantics, usability and human computer interaction, pattern recognition and computer architecture,

Electronics & Communication: VLSI & embedded systems, Communication Signals and systems, control systems

Medical Electronics & Instrumentation Technology: bio-medical image/ signal processing, Reproductive bio-engineering, obesity and endocrinology, rehabilitation engineering, bio medical instrumentation,

Computer Applications: e-commerce, e-business & e-governance.

SUBMISSION PROCEDURE:

Papers are accepted all through the year.

Full papers have to be submitted as per the IEEE format given in the journal home page.

Complete details of the corresponding author and co-authors have to be provided.

Authors are requested to submit the manuscripts in Word document file as per the given format to the following E-Mail ID: editor. ijese@dayanandasagar. edu

SELECTION PROCEDURE: The papers received will be first screened by the editorial board for suitability and then subject them to plagiarism check. Later the papers will be evaluated by “double blind refereeing system”, by both Indian reviewer and foreign reviewer. After receiving the reports from the reviewers, the Final decision on publication will be taken by the editorial board. Authors have to sign a copyright declaration form and submit to the editorial office before publication.

International Journal of Electrical Sciences and Engineering (IJESE)

DAYANANDA SAGAR COLLEGE OF ENGINEERING, BENGALURU-560078

http://dayanandasagar.edu/dsce/index.php/e-learning

Admissions 2016

· pdf file2 ise dr. naveen n.c. dr. sunanda dixit ... upgrade and tra nsform the methods and...

Documents