International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Web Site: www.ijaiem.org Email: editor@ijaiem.org

Volume 4, Issue 3, March 2015

ISSN 2319 - 4847

Fast Image Downloading System For Medical

Domain Using Distributed System
ProfShaileshHule, RajaniPhadtare, VarshaPatil, SnehalRengade, PriyankaBansode
PimpriChinchwad College of Engineering, University of Pune

ABSTRACT
Downloading any file quickly is an integral part of our social as well as professional living. The modern society cannot run
without various downloading facilities. There are many systems which are providing quick downloading of large files for
personal and corporate world. In the current system, if we will focus on medical domain then there is no such system which is
providing any facilities like providing quick downloading for medical domain. Doctor or patients have to carry all the report
images or standard images all time. If in any emergency doctor want to compare the patients image with standard image then
he cannot download it within minimum time as the images are large in size. Our system will provide facility for downloading
image within minimum time with help of Bit torrent. So that if the client who will download the image for the very first time he
will download it from server. But when another user will try to download it, then the first downloaded image will work as
replica for the current downloading. And the process will carried out so on. This system will also use parallel file downloading
protocols. When any user will start his system then his Torrent will also get start automatically, and it will keep record of each
system on which it is installed. So that Torrent will keep track of each image which has downloaded and on which system it has
downloaded.

Keywords:Torrent,Protocols ,Distributed System.

1.INTRODUCTION
Data replication is a key building block for large scale distributed storage system. Parallel download protocols that
establish multiple connections to distributed replica servers and simultaneously download individual parts of the data in
parallel, as depicted in Figure 1, promise a reduction of the data download time by up to a factor of n when using n
parallel connections, compared to a download from a single server only.Moreover, the failure of a replica during the
download reduces the download rate, but does not interrupt the transfer. Finally, parallel download protocols shed load
onmultiple servers and therefore automatically achieve a loadbalancing, resulting in higher server availability.The
challenges in the design of parallel download protocolsarise because network conditions in the Internet are
heterogeneous,dynamic and difficult to predict, and becausefile sizes follow a distribution that varies from applicationto
application. Current parallel download protocols only existas application-layer solutions that target specific
applicationdemands and network environments. CDNs downloadsmall pictures (ads) in parallel from low-latency
servers Slicing protocols, such asBitTorrent [6], split multimedia files of several MB into smaller slices and coordinate
the download of individual slices from multiple peers.

Figure 1 Parallel download from multiple servers

In this paper we develop a mechanism to download the image parallel from multiple client.Suppose there are n number
of clients from that n number of the clients specific number of the clients are having the particular image and if other
client want to download that image it will fire the query with respect to that image,we will maintain an update table
which will contain the information about the images which would be maintained by the server,and if that images is

Volume 4, Issue 3, March 2015

Page 162

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Web Site: www.ijaiem.org Email: editor@ijaiem.org
Volume 4, Issue 3, March 2015

ISSN 2319 - 4847

available the system will find on which process that image is available and depending on the health of the process the
task will be assigned to each processors and parallel downloading operation will be carried.

2.MATHEMATICAL MODEL
Consider a set S consisting of all the elements related to a program. The mathematical model is given as below,
S={s, e, X, Y, F, DD, NDD, Mem_shared, CPUcorecnt | {} }
Where,
s= Initial state
e= End state
X= Input set
Y= Output set
F= F(me) +F(friend)
F(me)= Main function
F(friend)= Supportive function / inbuilt function
DD= Deterministic data
NDD= Non-deterministic data.
CPUcorecnt= No. of cores
Mem_shared= Memory shared by processor.
Success= Desired outcome is generated
Failures= Desired outcome is not generated.
s= User will fire query for image
e= requested image downloaded
e1= image not available
X= Image name
Y= downloaded image or unavailability message
F(me) = {F(search), F(selection), F(partition),F(download),F(merge),F(update)}
search
F(search) = Searching will search requested image on the network.
Input: Image name
Output: Availability message & node count or unavailability message.
selection
F(selection) = Selection will select the node on which image is available.
Input: Total node count
Output: Selected node will provide requested image.
partition
F(partition) = Partition will perform fragmentation of requested image according to node count.
Input: Selected node count.
Output: Fragmented image.
download
F(download) = Download function will parallel download fragmented image from selected nodes.
Input: Fragmented image and selected node count
Output: Download message or failure message.
merge
F(merge) = Merge operation will perform merging of downloaded image.
Input: Downloaded image fragments
Output: Merged image.
update
F(update) = Update will update the availability table.
Input: System on which image is downloaded
Output: System information will be added to availability table.
Success:
If image is available then downloading of image done successfully.
Failures
If image is available but not getting downloaded then it will report error message.

3. MATHEMATICAL EQUATION
If you want to download Single file from single node then downloading time=T
If you want to download Single file from multiple node then downloading time=T x

Volume 4, Issue 3, March 2015

Page 163

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Web Site: www.ijaiem.org Email: editor@ijaiem.org
Volume 4, Issue 3, March 2015

ISSN 2319 - 4847

where x=selected nodes from multiples node having images i;

Tx is downloading time of single file.
Ts =Selection time
Tm=merging time
Where,
T=Total downloading time

((Tx)+(Ts)+Tm))<=T
Tx=max(

whereTx is absolute value

Tx=max(
Consider the following example:
If a image i is of 3 mb and network speed is 1mb then time required to download image is 3 unit time .But if we will
consider that same image is existing on the same systems then time required to download same image is 1unit. Because
each system parallely will be downloading 1mb part and merging time and searching time will be negligible so we will
required 1 unit time.

4.DATAFLOW DIAGRAM

Figure 2.DATA FLOW DIAGRAM

5.SYSTEM ARCHITECTURE

Figure 3.SYSTEM ARCHITECTURE

System Modules are

Volume 4, Issue 3, March 2015

Page 164

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Web Site: www.ijaiem.org Email: editor@ijaiem.org
Volume 4, Issue 3, March 2015

ISSN 2319 - 4847

i. Searching:
Searching is the very basic function which is carried out for the quick image downloading. As soon as the user will
press the enter tab, that time searching will be carried out at the back end of the system. The very basic use of searching
feature is that the requested image is present or not in the system server. If the image is present then at how many
places it is present, along with the system name and system count.
ii. Selection and Partition:
As soon as the system will perform searching function that time selection of system from the available system will
carried out. The selection process will be completely based on the system health i. e. first system health will be check
and if the system health is up to the required mark then only those system will be selected. Then image will be
partitioned into number of parts so that at the time of parallel image downloading, the parts of image can get
downloaded quickly and hence can save the download time.
iii. Download and Merge:
When the image will be get partitioned into number of parts, the downloading of image will be done. When image will
get downloaded, the image will be consist of parts and we need to reassemble those part as one image. So the merging
will be carried out within very less time. And then the assembled image will be shown to user.
iv. Updating Availability Table
When the image will get downloaded on one system then the availability table which is used for searching of an image
will be updated. So that more number of system will be available for downloading.

6.CONCLUSION
This paper presents a parallel download from n replica to a single client. It deals with the partitioning of the image and
then merging itsucessfully.Whichprovides the fast downloading in efficient time and would be helpful for the medical
domain.

REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]

http://www.net.t-labs.tu-berlin.de/ roger/papers /tcpparis.pdf

http://en.wikipedia.org/wiki/BitTorrent
http://en.wikipedia.org/wiki/Peer-to-peer
http://wiki.lustre.org/images/8/81/LciPaper.pdf
http://www.vmware.com/_les/pdf/partners/academic /ipdps09.pdf
http://en.wikipedia.org/wiki/Data-ow-diagram
http://en.wikipedia.org/wiki/ IBM-General-Parallel-File-System
https://www.sharelatex.com

Volume 4, Issue 3, March 2015

Page 165