Académique Documents
Professionnel Documents
Culture Documents
net/publication/290626961
CITATIONS READS
7 371
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Shamim Akhter on 01 March 2016.
1 Introduction
The recent progress in virtualization technologies and the rapid growth of Cloud
computing services have opened a new opportunity for complex scientific workflow
such as Disaster Management System (DMS) [1, 3]. Cloud services such as Amazon
EC2, Google Cloud services etc. can provide reliability, scalability and interactive
platform to increase the performance [5].Therefore in this paper we have described a
Cloud based implementation of Disaster management system [2].Cloud can provide a
large amount of computing power over short periods of time during a disaster - so
several government agencies as well as NGOs and other associations like RedCross,
UNO etc. can respond more efficiently to anything in the world during disaster [4].
Moreover, in case of disaster people never knows when it might have a spike for a
need in compute power or disk storage. In such case, Cloud platform allows resources
to be used on an elastic basis. In addition, Cloud platform drives down costs by shar-
ing resources and being more communal, it allowsquicker communicating response to
emergencies and disasters to be more agile. Furthermore, for massive information
sharing among government and other agencies in such situation,Cloud computing
environment can be the most helpful [3]. Finally, Cloud computing allows for rapid
scaling when needed, it allows for significant flexibility and reduces cost tremendous-
ly. Therefore, most experts agree that when it comes to information technology, and
especially a complex, uncertain and dynamic system likesdisaster management, Cloud
computing is the best way to go. In this paper, the design of a complete Cloud based
DMS is described along with its different functionalities.
© GPC 2013
2 Related work
Several ICT based DMS have been implemented already [1,10]. An ICT based
DMS should be well designed to deal with all four stages of the life cycle of DMS [2]
such as (1) planning, (2) emergency response, (3) recovery and (4) post-planning.
However, most of existing DMS model mainly deal with first two stages.For example,
Emergency response system [1] and warning system for mass people [10] are two
significant example of existing DMS model which mainlydesign a warning system for
people, vehicle and transport. On the contrary, the proposed system architecture of a
DMS in this paper deals with all four stagesalong with following contributions
1. Improve the intelligent system of traditional DMS and make easier the deci-
sion making process
2. Increases the efficiency of the system through task distribution among differ-
ent services
3. Simulate the performance and compare that with traditional system to evaluate
the efficiency
In this paper, the implementation structure of a DMS has been proposed based on-
Cloud environment for following reasons
1. Huge data can be computed easily and quickly.
2. During natural disaster ICT infrastructure also get damaged, such as some
servers of IEEE damages due to recent violent flood sandy [5]. However as the
data will be stored in Cloud, those will have replicated back up.
3. The computation and decision making process for DMS is too complex and
need apparently large amount of time. However, distributed environment in
Cloud has expedite decision making process.
4. For data storage, Cloud is superior in providing security, easier sharing and-
migration, flexible access and rights management. if Cloud is used for imple-
menting DMS data storage and management can be taken care of by Amazon
S3 and Google BigTable [5].
5. Improve discovering different class and characterizes resources.
The proposed DMS consists of six major components such as (1) Web Portal, (2)
Role Manager, (3) Workflow Engine, (4) Workflow Scheduler, (5) Workflow Moni-
tor and (6) Notification depicted as in figure 1.Among these components, Workflow
Engine (WE) and Workflow Scheduler (WS) are the most important components.
Fig. 1. Components of Proposed DMS Workflow
Work Management System
Web Portal
Web portal provides a Graphical User Interface (GUI) that helps users to edit
workflow. Mainly all kind of agent can access different pages of this web portal aac-
cording to their role. All task management and workflow management can be done
through this portal.
tal. Although tasks are prepared from collected data and assigned with
number of required resources as well as priority by WE.
WE However the portal also su sup-
ports manual task configuration for dynamic and uncertain problem domain.
Workflow Edit
In this work, Workflow is organized
organ as DAG [2, 6], which has been converted to
xml schema for implementation.
implement . Figure 3 represents the schema of a workflow in the
proposed DMS. Each workflow consists
consist of a number of parameters such as Agent,
Flow connection, list of resources, environment parameters, performance, assigned
resource status, overall workflow
workflow status, task status and task. In figure 4, different
parameters of a workflow in this proposed DMS have been described.
Role Manager
This module is consists of an Agent Role Manager (ARM) which is in charge of
defining access permission for agents on different modules and section of DMS.
ARM also differentiates tasks among all agents. The next sub module is Task Distri
Distrib-
utor which distributes the task to appropriate Agentas
Agentas soon as a task is produced and
defined by the Workflow
orkflow Engine. In order to monitor the activities of different agents
and prepare the Audit report, there is another sub module named Monitor.
Workflow Engine
This is the core component of this proposed DMS. Workflow Engine (WE) is
consists of five basic components for managing such as resource, task, agent, data and
audit activity as shown inFigure
in 5. Resource is the key element of this proposed sys-
sy
tem. two different sub modules Cost Monitor and Performance Monitor continuously
monitor two most important
ortant attributes
attributescost
cost and performance of the resources. Re-
R
source Monitor will investigate the status such as idle, out of work, running. Another
important component of WE is Task. Four different sub modules Task Manager, Task
Scheduler, Task Executor and Task Monitor are in charge of managing tasks. Besides
secured data management another major challenge of this proposed system is data has
to be shared beyond geographical boundary, among different countries, different
NGOs, different organizations and take necessary decisions for future. All sub mo mod-
ules in Data component of WE have been integrated to attain the ultimate goal.
In order to distribute the roles among agent this MAS based system, it has a Role
Manager. All activities of agents are monitored by Agent Monitor. The main functifunctio-
nality of Agent Access is to keep trace of the permissions allowed for different agents.
Finally all internal communications among different agents as well as among different
modules are managed by Agent Manager.Finally,
Manager. all
ll activities and actions taken by
the proposed system are continuously audited by the Audit WfMS.
Workflow Scheduler
The MAS based workflow scheduling algorithm proposed by S. Akhter et al. [2]
has been deployed in the workflow scheduler of the proposed DMS.
Figure 6 depicts that the workflow scheduler has six main components. At first all
tasks of current workflow prepared by WE are put in to Ready Queue. The priority of
tasks can be changed dynamically by On-time priority changer. After tasks are or-
dered according to their priority in Ready Queue, Scheduler runs the scheduling algo-
rithms [2]. All successful jobs are stored in Successful Task list along with their re-
port and failed tasks are stored in Failed Task. Failed task are rescheduled by Re-
Scheduler and sent back to Ready Queue.
Workflow Monitor
. All performance and status of currently running as well as previous workflow are
generated by Workflow Monitor. Later all reports can be view from web portal based
on performance parameters described in table 1.
Name Description
Number of Succeed Task per workflow Defines the success rate
Number of failed task per workflow Defines the failure rate
Number of used resources Define the resource business
Number of total assigned resource Define the actual resource capacity
Resource Status Define the resource availability
Notification
This is an important component of the system. There are two different types of ac-
tivities those have been performed by Notification. One is to implement Cloud to
device messaging (C2DM) in order to broadcast notification to mass people. Another
is to notify several components within the system.
3.2 Cloud Implementation
The most important contribution of this proposed DMS is cloud based implement menta-
tion.Figure
.Figure 7 describes the three layer of Cloud environment. PaaS is the preferred
model over fully outsourced data processing and handling [3],, presumably gaining
support for having clear visibility, ownership and control over all the data. At the
same time, system can quickly obtain the benefits of a fully-maintained
maintained software sol
solu-
tion on a subscription basis. With PaaSsystem can get full control over data encry encryp-
tion and security. Therefore in this proposed system on PaaS, data related to all deci-
dec
sions taken for several past as well as current disaste
disasters
rs for various locations for diffe
differ-
ent type off incidents and tasks are store.
store. This historical data are used as Heuristic data
storage for further workflow scheduling. In addition in this layer, all record as a result
of continuous audit performed by different
different agents, success and fail report for different
workflow, status and performance evaluation of different resources, comparative
analysis for different type of tasks in different workflow for different regional places
are stored. The next layer (IaaS)
(IaaS) is the most important layer. Amazon EC2 can be a
suitable candidate as IaaS. The main components of this proposed DMS such as
Workflow Engine (WE),, Workflow Scheduler (WS), Monitor, Cloud web services as
well as temporary data storages are put in this layer.
la The proposed web portal is ees-
tablished in SaaS layer.
4 Experimental Evaluation
For simulation icanCloud [4] has been used in this proposed DMS system. Different
types of damage for which we have collected data from two different data sources [7]
and [8] are listed in table 2. The same data were used in [2].
Table 2.Query criteria for Ten Cases used in Simulation
Data Source Disaster Type Location Year
Case-1 DMSS Tsunami All Region 1974-1986
Case-2 DMSS Flood All Region 1990-2010
Case-3 DMSS Epidemic All Region 1990-2010
Case-4 DMSS Flood + Epidemic North and South coast 1990-2010
Case-5 DMSS Forest Fire North and South coast 1990-2010
Case-6 DMSS Tornado All Region 1990-2010
Case-7 DMSS Strom All Region 1990-2010
Case-8 NDDB Earthquake Asia Zone NA
Case-9 NDDB Cyclone + Flood Asia Zone NA
Case-10 NDDB Tidal wave Asia Zone NA
In this paper the implementation of this MAS model for DMS in real time system
is described and designed. Moreover, the design and implementation plan of a web
portal based on proposed DMS is described. The system architecture and components
of the proposed DMS are also described in this paper. Following, but not limited to,
are some contribution of this proposed work:
1. Web Portal Implementation for Automated workflow model for DMS
2. Real time implementation of DMS on Cloud
3. Eliminate dynamic on time dependency rather than providing proactive de-
pendency calculation
4. In corporate more parameters to DMS such as time, cost, performance,
Quality of Service etc
5. Redistribution of task among agents during idle time
Future work will focus on further analysis and validation of different stages of disas-
ter management system life cycle, and on broadening the scope of this work to real-
time operational, decision making and strategic management of DMS. Moreover,
another suitable extensions of this proposed work can be implementation of Cloud
based augmented reality to detect damaged area and possible easier transport route
which can be a great contribution for recovery system in case of natural disaster.
Reference
1. D. Mendona and W. A. Wallace. Studying organizationally-situated improvisation in re-
sponse to extreme events. International Journal of Mass Emergencies and Disasters, 22(2),
2004.
2. M. Habiba and S. Akhter, MAS Workflow Model and Scheduling Algorithm for Disaster
Management System, In Proceedings of the 1st International conference on Cloud Compu-
ting technologies, Applications and Management, ICCCTAM 2012.Dubai, UAE, Decem-
ber 2012.
3. Kazusa, Shuhei, Director for Disaster Management, Cabinet office, Government of Japan,
“Disaster Management of Japan” Published in 2011.
4. EminNazrov, Emergency Response management in Japan, Final Research report, ASIAN
Disaster Reduction Center, FY2011A Program, 2011
5. SurajPandey, DilebanKarunamoorthy and RajkumarBuyya, Workflow Engine for Clouds,
Cloud Computing: Principles and Paradigms, R. Buyya, J. Broberg, A.Goscinski (eds),
ISBN-13: 978-0470887998, Wiley Press, New York, USA, February 2011.
6. Masahiro YOSHIZAKI, Disaster Management and Cloud Computing in Japan, Report
from Ministry of International Affair and Communication ,December 2011
7. An Analytical Overview March 2007 published by Asian Disaster Reduction Center.
8. Disaster Management System Srilanka, http://www.desinventar.lk/, Last visited
11.11.2012
9. ArcGIS as a System for Emergency/Disaster Management, ArcGIS as a System for Emer-
gency/Disaster Management,Last visited 11.11.2012
Alazawi, Z.; Altowaijri, S.; Mehmood, R.; Abdljabar, M.B.; , "Intelligent disaster man-
agement system based on Cloud-enabled vehicular networks," ITS Telecommunications
(ITST), 2011 11th International Conference on , vol., no., pp.361-368, 23-25 Aug. 2011