International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169

Volume: 4 Issue: 5 539 - 542

_______________________________________________________________________________________________________
Big Data Approach for Secure Traffic Data Analytics using Hadoop

Koushalya Bijjaragi1, Poonam Tijare2

1 M.Tech Student, Computer Science and Engineering, CMR Institute of Technology,
132, AECS Layout, IT Park Road, Kundalahalli, Bangalore 560037
30koushalyab@gmail.com
2Assistant Professor, Computer Science and Engineering, CMR Institute of Technology,
132, AECS Layout, IT Park Road, Kundalahalli, Bangalore 560037
poonam.v@cmrit.ac.in

Abstract As the volume of traffic is increasing day by day, it gets difficult to store and process such huge sets of data using traditional
software. A cluster of storage devices is needed to store such huge amounts of data and also a parallel computing model for analyzing those huge
inputs of data. Hadoop is one such framework that provides reliable cluster of storage facility, which stores huge data in a distributed manner
using a special file system, called Hadoop Distributed File System and provides efficient parallel processing feature through MapReduce
framework. Using Map Reduce the filtered traffic data can be fetched easily, to provide end user with traffic analysis and giving useful
predictions.
Keywords HDFS: Hadoop Distributed File System, Map Reduce
__________________________________________________*****_________________________________________________

1. INTRODUCTION handle massive amounts of data where data is segmented and is

Traffic congestion is one of the uncontrollable problems faced stored on different nodes. A large task is divided into small
almost every day in urban environments. The reasons for which tasks, and is processed in a MapReduce model. At the same
could be attributed to increasing growth in the number of time, the system stability and fault tolerance is important.
vehicles, lanes having improper plan or lacking a pre-plan,
major junctions and traffic signals, commonly used roads for 2. Big data can improve the efficiency of transportation
easy access to other roads, roads having no dividers, industry largely. Transportation industry, involving many
unavoidable long routes, bottlenecks and worn out roads and so aspects of work, need to handle massive amounts of data each
on. Apart from these challenges faced in the current day, needs more controlled mode of application and has a great
infrastructure, due to other constraints like space, economic or deal of equipment. In the aspect of improving transport
political, the engineers involved in road construction are efficiency, improving the threshold capacity of the road
obliged to construct better, well laid road networks and network, adjusting traffic demands, big data technology has
flyovers in neutralizing the problems of traffic jams in the obvious advantages.
urban scenario. Hence, the number of risks and challenges Heterogeneous nature of accident data is the major problem
involved in stabilizing the traffic flow increase if the plan made in its analysis. This heterogeneity should be reflected on
to deal with it is baseless. Also the renovations made impact analysis of the data else, few relationships that exist between
the existing map and would possibly lead to unwanted loops or the data might remain unseen.
perhaps new spots for traffic jams would come into the picture The term big data classifies the particular forms of data sets
if addressing the possibilities of such issues in the near future is comprising formless data which will be taken from the
not heeded in a tactful manner. technical computing application layers.
One of the most beneficial applications of the systems
employed in traffic control is the improved ability to control 2. EXISITNG SYSTEM
the road network traffic. Tracking methods are used to capture At present, traffic information is commonly shared between
the position and/or location of massive number of mobile different traffic agencies by means of voice or data
objects. With the help of that tracked information, analysis and communications. For such data communication between
prediction of traffic density in a given network is enhanced. management centers, a common language and a frame of
This renders valuable information for controlling traffic flow, reference is required. There are many possible ways to analyze
prediction of congestion and reducing the number of accidents traffic data like manually counting each vehicle or through
in that network. image processing, video analytics and so on. Manual counting
The field of big data for resolving the above questions involves a team to count each vehicle travelling on the road and
provides a new technical approach. Big data applied to road update the statistics and a separate team is required to analyze
traffic analysis has the following advantages: these results of statistics.

1. Traffic management system that uses big data technology 2.1 Limitations of the Existing System
can handle vast amounts of complex and diverse data. Big data Requires very huge man power.
has resolved three major problems: data storage, data analysis Time consuming and cumbersome.
and data management. Hadoop is born with the ability to Create unmanageable traffic jams.
539
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 5 539 - 542
_______________________________________________________________________________________________________
If there is no traffic, the team personnel map-reduce functions. Hadoop uses this framework to analyze
still need to wait at junctions. large datasets distributed over HDFS because of its unbeatable
Lack of traffic details to users: Present fault tolerance, reliability, distributed and parallel computing
traffic systems fail to provide traffic features.
information including busy/idle time 3.3 System Architecture
and number of accidents happening. Architecture design is a diagram which represents the basic
The expense involved in maintaining the structure of the entire project. It includes the various
field components used is also high. components that are a part of the project and how the
Setting up the framework initially is also components are connected. It also shows the actions performed
difficult. by each component. Figure 3.1 shows the architecture design
where the user needs to first sign in to communicate with the
3. PROPOSED SYSTEM server. The user is provided with options to generate a dataset
A dataset which consists of information about traffic is created. and also to manually insert an entry into the dataset.
Map Reduce is used for processing to get the secure analysis of
traffic data. Based on the analysis, prediction of traffic is made
to show at what time it will be high and low in a day for a
particular area. Prediction is also made for which month will
have most number of accidents. Authentication is provided
using signature, hence user security is ensured. Encryption
algorithm is used to protect data, as it is sent across network.

3.1 Hadoop Distributed File System HDFS

A cluster in HDFS consists of a single name node which is a
master server machine and is responsible for managing the
entire file system. It also provides an access to the file system
as requested by the clients. Each cluster also consists of
multiple data nodes, and each data node in turn consists of
blocks of split data. Data distribution to these data nodes is
maintained by the master node. The main operations of data
nodes include file read and write.
Hadoop adopts the concept of rack awareness, which mainly
helps in storing data into a rack and finding its location in the
cluster. This means that a Hadoop Adminstrator can clearly Figs 3.1 System Architecture
define what chunks of data should be saved on a particular
rack. This way, it is possible to mitigate the data loss if a rack The other operations of server include storing the
fails entirely. Network performance is also improved because dataset into HDFS and fetching the matching data from the
data replication is done on various racks of different machines. directory as requested by the user. It does so by sending the
request to HDFS directory where the Map and Reduce phase
3.2 MapReduce takes place according to the input request data.
MapReduce is a software framework which was introduced by The HDFS directory, when finishes the reduce phase, sends the
Google to carry out parallel processing on large data sets. This result back to the server where the data is encrypted and sent
large data set is distributed over a large number of machines across network. Once the user's signature is verified,
present in a cluster. For quick access, each machine computes decryption takes place and the requested information is sent to
and stores the data locally, this in turn contributes to the user. The user on receiving the result of requested query,
distributed parallel processing. Such a computation involves can analyze the data easily.
two parts Map and Reduce. In the Map phase, data nodes
take raw input data and produce intermediate data based on the 3.4 Signature Generation Algorithm
type of computation and then that data is stored locally. In Input: Public key (A,B,h) , system parameters , message M
Reduce phase, intermediate outputs from map phase is fetched Output: Generate a valid group signature on M
by the nodes and then it is combined to derive final output that Select random numbers a, roM, roR, mus,mux, mueprime, mut,
is stored in HDFS. Name node with its prior knowledge on the muE
data distribution, tries to assign the task to a particular node Computes the following values
based on the locality of data. Developers can write custom map E0 = g * roE
and reduce functions suitable to the application and the E1 = h+(h1*roE)
MapReduce function then takes care of distributing and E2 = h+(h2*roE)
parallelizing tasks across a rack on commodity hardware in the ACOM (A*(a2^rom) mod n).mod n
cluster underneath. s = (Eprime+ke)* roM
Inter-machine communication is managed by the BCOM (B *(w^ roR mod l) mod l
framework, thus programmers only have to focus on actual t = Eprime * roR
540
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 5 539 - 542
_______________________________________________________________________________________________________
V0 = g*muE {
V1 = (g * mux+(h1 * muE) Display as average case
V2 = (g*mux))+(h2* muE) }
Vmpk= (((a1^mux mod n)*(a2^mus mod else
n)).mod(n))*(ACOM^~mueprime mod n)).mod(n)) {
Vrev= ((w^mut mod l))*(BCOM^~mueprime mod l)))mod lE = Display as bad case
E0+E1+E2 }
V = V0+V1+V2 Step 5: Stop
reste = ACOM+BCOM+V+Vmpk+Vrev
Set c = f(E + reste + message) 4. RESULTS AND IMPLEMENTATION
Construct the following numbers
taux = c * (x+mux)
taus = c * (s+mus)
taut = c *(t+mut)
tauePrime = c *( Eprime+mueprime)
tauE = (c*(roE+muE)) mod(o)

Return
=(E0,E1,E2,ACOM,BCOM,c,taux,taus,tauePrime,taut,tauE)
End

3.5 Signature Verification Algorithm

Input: System parameters and signature =
(E0,E1,E2,ACOM,BCOM,c,taux,taus,tauePrime,taut,tauE)
Output: True or False

Compute the following values

taue = (c*( expKe+tauePrime);
tauEG = g *tauE
a0a1 = (a0^c mod n)*(a1 ^ taux mod n))mod n
a2A= (a2^taus modn)*(ACOM^ ~taue mod n)mod n Fig 4.1 Server Page
Vmpk = (a0a1 * a2A)mod n
Bw= ((b^c mod l)*(w^taut mod l))mod l The figure 4.1 shows the server screen. For users request to be
Vrev = (bw*(BCOM^~tauePrime mod l)) mod l processed this page should be running always. The admin has
E = E0+ E1+E2 control over this page and can add areas, view list of areas and
V = V0+V1+V2 delete areas any time he/she wishes to.
reste = ACOM+ BCOM+ V + Vmpk+ Vrev
if c = f(E + reste + message)
Return True
else
Return False
End

3.6 Algorithm used for Busy/Idle Traffic Prediction

Input: Traffic details from MapReduce
Output: Time when the traffic is low, average, high
Step 1: Filter traffic between 6 to 10
Step 2: Find highest number of vehicles
For all traffic details
{
Find: number of vehicle / highest number of vehicles
}
Step 3: Sort average
Step 4: Display time when traffic is less
if (val<0.5)
{
Display as best case
Fig 4.2 User's Main Menu
}
else if (val<0.8)
541
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 5 539 - 542
_______________________________________________________________________________________________________
The figure 4.2 shows the page which appears as the Overall, that data which the user wishes to retrieve can be
user logs in successfully, where in the logged in user's name efficiently fetched using the Big Data approaches such as
will be displayed on the top. The user has full access to all the Hadoop and MapReduce technology, by overcoming all the
buttons as shown in the snapshot. limitations that were encountered by the existing system.
At present the application operates offline, in the
future changes that make it work online should be incorporated.
More useful predictions can be made if the data to be operated
on is the real-time data which can be fetched easily from open
source platforms.

References
[1] Traffic Prediction and Analysis using a Big Data and
Visualisation Approach - Declan McHugh, Department of
Computer Science, Institute of Technology Blanchardstown,
March 10, 2015
[2] Traffic Accident Analysis Using Machine Learning Paradigms
- Miao Chong, Ajith Abraham and Marcin Paprzycki, Computer
Science Department, Oklahoma State University, USA,
December 20, 2004
[3] A Perspective Analysis of Traffic Accident using Data Mining
Techniques - S.Krishnaveni, Dr.M.Hemalatha, International
Journal of Computer Applications, Volume 23 No.7, June 2011
[4] Application of Big Data in Intelligent Traffic System - Gang
Zeng, IOSR Journal of Computer Engineering, Volume 17, Issue
1, Jan Feb. 2015
Fig 4.3 Generation of Dataset [5] A Comprehensive Analysis of The Vehicle Traffic Prediction -
Ravindra M, Lokesha V, Prasanna Kumara, International Journal
As shown in figure 4.3, the user can generate the of Engineering Research & Technology, Vol. 1, Issue 8, October
dataset by specifying the numbers of records to be generated by 2012
the application automatically. [6] Abellan J, Lopez G, Ona J. Analyis of traffic accident severity
using decision rules via decision trees, vol. 40. Expert System
with Applications: Elsevier; 2013.
[7] Tom White, Hadoop: The Definitive Guide, 3d. Ed. CA, 2012.
[8] Improving Decision Making in the World of Big Data
http://www.forbes.com/sites/christopherfrank/2012/03/25/improv
ingdecision- making-in-the-world-of-big-data/
[9] Pakize S, Gandomi A (2014), Comparative Study of
Classification Algorithms Based On MapReduce Model,
International Journal of Innovative Research in Advanced
Engineering (IJIRAE), 1(7), 251-254.
[10] Bhagattjee B (2014), Emergence and Taxonomy of Big Data as
a Service.
[11] Ashish Vijay, Priyanka Trikha, Kapil Madhur, A New Variant
of RSA Digital Signature, International Journal of Advanced
Research in Computer Science and Software Engineering,
Volume 2, Issue 10,October 2012, ISSN: 2277 128X.
[12] Gang Zeng, Application of Big Data in Intelligent Traffic
System, IOSR Journal of Computer Engineering (IOSR-JCE), e-
ISSN:2278-0661, p-ISSN: 2278-8727, Volume 17, Issue 1, Ver.
VI(Jan Feb. 2015), PP 01-04.
Fig 4.4 Traffic Flow Analysis and Prediction
[13] Ashwini E.M, Sunil M.E, Tadoop:Traffic Analysis and Traffic
Solution Using Hadoop,International Journal of Innovative
This snapshot gives the prediction of traffic flow for a Research in Computer and Communication Engineering, Vol.3,
particular date. It indicates which is the best, average and bad Special Issue 7, October 2015.
time to travel based on hourly traffic density.

CONCLUSION AND FUTURE SCOPE

The project presented a secure approach to collect the
user requests for traffic data details. Each user is authenticated
and only upon signature verification the requested details are
provided to the users. The proposed system is designed for
secure analysis and giving predictions about the traffic flow
details in a day and number of accidents happening per month.
542
IJRITCC | May 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________________________