ARTİCLES 4

Location Based Early Disaster Warning and Evacuation System on Mobile Phones Using OpenStreetMap Khandaker Mustakimur Rahman, Tauhidul Alam, Mahfuzulhoq Chowdhury Department of Computer Science & Engineering Chittagong University of Engineering & Technology Chittagong-4349, Bangladesh. Email: mustak_cuet@yahoo.com, tauhid03_cuet@yahoo.co.uk, mahfuz_csecuet@yahoo.com Abstract— Natural Disasters have threatened mankind since history started.

 Due to geographic position and climate change, Bangladesh is one of most vulnerable countries to natural disasters. The country also lacks effective disaster preparedness system to confront natural disasters. Timely disaster warning and evacuation guideline can save lives of the people. In addition, a tourist or a blind people may face difficulties in finding safe area or shelter place prior to the occurrence of natural disasters. For this reason, we have proposed a location based early disaster warning and evacuation system for both normal and blind people using OpenStreetMap (OSM). The system is implemented on android mobile phone because of the burgeoning growth of smart phones in Bangladesh. So, our system comprises a third-party server named Disaster Management Server (DMS), android device with our application installed on it and user. The local weather office updates the disaster (tsunami, cyclone or flood) data on DMS. Device user registers on Android Cloud to Device Messaging (C2DM) server to get automatic notification of upcoming disaster otherwise user gets manual notification. The user communicates with DMS to have updated data sending the current position obtained by GPS or network provider. The probable disaster affected area is determined by ray casting algorithm. When our application recognizes the user in probable disaster zone then application will disseminate visual and audio disaster warning and evacuation guideline including shortest path of shelter or safe zone on the map of the application. Evacuation progress is also tracked using DMS and national identification of user. The experimental result demonstrates the effectiveness of our system. Keywords- Natural Disaster, OpenStreetMap, Disaster Warning, Evacuation Guideline, Android I. INTRODUCTION Natural Disaster is the consequence of natural hazards such as cyclone, storm, earthquake, tsunami, flood etc. This earth has already observed the destructive mode of nature which has taken millions of lives. The 2011 Japan earthquake and tsunami, the 2010 Haiti earthquake, the 2008 cyclone Nargis, the 2004 Indian Ocean Tsunami, the 1991 Bangladesh cyclone are some recent examples of deadliest natural disasters [1]. Natural hazards generally end up with disasters where the affected areas are vulnerable. The report of United Nation’s International Strategy for Disaster Reduction (UNISDR) on Mortality Risk Index (MRI) ranks the Bangladesh no. 1 among 200 countries most at risk from earthquakes, floods, tropical cyclones, and landslides [2]. Since Bangladesh is most natural disaster-prone country, so prevention is necessary for protecting lives and properties. Sometimes people may be unaware about the upcoming natural hazards. Lack of preparedness of people causes the major damage during disaster. So, adequate prior disaster warning and effective evacuation system can save significant number of lives in the country prone to frequent disasters. A new comer or a tourist and a blind people in a particular area may face the problem in finding safe area from his current stay during disaster. Hence, we have proposed a location based early disaster warning and evacuation system on mobile phones using OpenStreetMap which is mainly utilized to provide audio and visual messages on the map. OpenStreetMap (OSM) is a rapidly growing open source map of the world because of the availability of map information across the world and the advent of inexpensive portable GPS devices [3]. This open source OSM has recently been employed in many projects like WikiProject Libya and WikiProject Haiti etc. In the WikiProject Libya [4], the roads and places of interest were mapped in details. The WikiProject Haiti facilitated the rescue work and helped in providing relief aid after the devastating earthquake at Haiti in 2010 [5]. So, OSM has achieved the popularity to use instead of using restricted Google map. The demand of location based services is also increasing day by day with the burgeoning growth of smart phones. Our location based system is also an android platform based smart phone application to render location based services showing the warning of upcoming disasters (tsunami, cyclone, and flood etc.) if the user is in the possible disaster affected area or near to that area and demonstrating nearest safe zone or shelters on the map of the application. Our proposed system is developed for both the normal and blind people. The usability of OpenStreetMap (OSM) is ensured for all users as it is free. Users of our application will get both text and audio warning message and evacuation direction on the map. The rest of the paper is organized as follows. In Section II, we discuss about some related works and their shortcomings in b rief. 

Then we present the description of our proposed location based system in Section III. In Section IV, system implementation details are described. In Section V, we have discussed the experimental result. Finally, we conclude this paper stating the future plan in Section VI. II. RELATED WORK Early Disaster Warning and Evacuation System is very common disaster management approach in disaster-prone areas of the world. In recent years, efforts in disaster management has gained the impetus from unprecedented development of Mobile Technology. Currently, mobile phones provide vital support for disaster management in many ways: monitoring, communication, warning dissemination, evacuation, rescue and relief aid. Moreover, the advent of smart phones supporting GPS functions assists in disaster management. To date, different researchers from all over the world have conducted decent number of researches about early disaster management system. Among them, Short Message Service (SMS) is used to collect the upcoming flood warning and send back to all citizens from the server [6], [7]. Here, lots of SMS transfer can cause the network congestion which might impede the voice call communication through the same network. This can make the evacuation process difficult either. In [8], Cell Broadcasting Service is utilized to directly send messages to the users in a specific area. This is helpful not to cause panic and network congestion as well. However, it still fails to help in evacuation process providing interactive information about shelters and safe areas. GSM alarm device for early disaster warning is proposed to place it in the local police station or fire bigrade station [9]. The device gets disaster warning from the weather office and make three different types of warning. The police station or fire brigrade station then controls the evacuation process and informs the center about evacuation progress through that device. Again this alarm based early disaster warning system is not a faster way of evacuation though it can avoid the problem of network congestion. Researchers also propose Area Mail disaster information service provided by NTT DoCoMo for tsunami alert and evacuation system with a view to support fishery workers [10], [11]. The area mail service makes it possible to deliver information simultaneously in a limited area among the vulnerable persons to the damage due to disaster. In this service, there is a center monitoring authority that will observe the evacuation progress of the fishing boats and suggest them according to the data came from the fishing boats using a mobile application. This service is a quicker and efficient one. Nevertheless, the service will not work as fast and efficient for the overall population. Developed countries like Australia and South Korea are planning to use satellite communication for disaster management when the mobile network fails. This service will be more reliable, secure, efficient and robust [6], [12], [13]. This service initialization and maintenance are expensive and can not be affordable for developing countries like Bangladesh. Very few researches have been carried out to provide location based services for disaster management on mobile phones. Moreover, the previous works did not distinguish normal people and blind people. Considering the facts, we have presented a location based early warning and evacuation system. Our location based system will forecast impending disasters, disseminate understandable visual and audio warnings to both normal and blind people, demonstrate evacuation guidance in response to the warnings. III. LOCATION BASED EARLY DISASTER WARNING AND EVACUATION SYSTEM The disaster management consists of four fundamental steps such as mitigation, preparedness, response, and recovery. Among these steps, the emphasis of our work is the preparedness which is the development of a system for the action plan of upcoming disasters. For this, we present a location based early disaster preparedness system comprising audio and visual warning and evacuation process on the map of the system for a developing country like Bangladesh. 

 A. Preliminaries The geographical location of Bangladesh, a South Asian country, situated as it is between the Himalayas and the ocean, on the delta of wide rivers, means that the country is very vulnerable to flooding [14]. The people of coastal areas in Bangladesh have to face several storms each year and cultivable lands disappear in river due to river erosion. Bangladesh is also one of the most susceptible countries to the planet’s climate changes [14]. Climate change is producing a growing number of cyclones, and paradoxically, there is a greater risk of drought, as well. The country is also surrounded by the Bay of Bengal which is indeed north-eastern part of Indian Ocean [14] and can ravage the country, particularly the coastal belt sternly. Hence, there is also the risk of Tsunami in this country. However, our disaster preparedness system protects the people of this country from these disasters and uses OpenStreetMap (OSM) because the development of OSM is very rapid. It is dedicated to encouraging the growth, development and distribution of free geospatial data and to providing geospatial data for anyone to use and share [3]. In Bangladesh, the movement of OpenStreetMap started in 2010. The OSM community of Bangladesh is developing the open source of map of Bangladesh since then. This is one of the projects employing their developed map of this country. This proposed work is also implemented on android mobile phones. Android is an operating system for mobile devices such as smart phones and tablet computers developed by Open Handset Alliance led by Google. Android is more open and comprehensive than other mobile operating systems. So, it is the best-selling smart phone platform worldwide. It also allows the developers, wireless operators and handset manufacturer to make new applications and products at lower cost [15]. The outcome is more personal and more interactive experience to the users. So, the android mobile platform has been used in our proposed disaster preparedness system.

 B. Proposed System Our proposed location based disaster preparedness system consists of a Disaster Management Server, GPS supported android mobile phones with our proposed application installed on it and users having national id of Bangladesh. Fig. 1 demonstrates the workflow of our proposed location based early disaster warning and evacuation system. Disaster Management Server (DMS), a third party server, stores disaster prone areas and the details about the users in its database. Regional weather offices can access the database to update the disaster prone areas. The user of our proposed android application can also register Android Cloud to Device Messaging (C2DM) server to get notification on application only when there is any update in the database of DMS. Through network provider or GPS provider, mobile phone gets the current location of its user and sends it to server. Using this current position, our system determines whether the user is in probable disaster affected area or not. To do this, the system employs modified version of Ray Casting Algorithm. The user of our proposed application gets early disaster warning message both in visual and audio message on map of the application. The user also gets the shortest path of safe area or shelter zone on the application. The national id of the user is stored in the internal storage of the mobile phone which is sent to DMS later on. Regional evacuation control authority keeps track of the evacuation progress prior to the occurrence of disaster with the help of DMS which expedites the evacuation process and the stored national id of the user. The steps of our proposed system are as follows:1) Register/Unregister to C2DM Server: Android Cloud to Device Messaging (C2DM) server is a Google server involved in taking messages from the third-party application server and sending them to the device. Cloud to Device Messaging (C2DM) is also a service to send data from server to the applications on Android devices [16]. The service provides a simple, lightweight mechanism that servers can use to tell mobile application to contact the server directly, to fetch updated data. The application on android device does not need be running any service to receive this lightweight message. Fig. 2 shows the registration/unregistration process of C2DM server. User of our application on android device can register or unregister on the C2DM server. After registering on this server, application receives a registration id which is stored into the Disaster Management Server (DMS). Then the device user gets notification from C2DM service if there is any update about the forthcoming disasters in DMS regardless of whether the application is running on android mobile phone or not. This helps the people who are unaware about the upcoming disasters. User can also unregister from C2DM server by which assigned registration id for the user is deleted from DMS 2) Communication with Disaster Management Server (DMS): Our application gets the current position through GPS or network provider from the user mobile phone and application communicates with DMS to send the latitude and longitude of user’s current position. On the other hand, for registered user, when DMS needs to push a message to our android application, it requests to broadcast the message via an HTTP POST to Google’s C2DM servers. C2DM server broadcasts the message to the device of the application. Afterwards, our android application fetches the data from DMS. And for unregistered user, our application receives the message when he wish to from DMS. Fig. 3 illustrates the communication with Disaster Management Server (DMS). 3) Probable Disaster Affected Area Determination: The DMS server runs Ray casting algorithm to determine the probable disaster affected areas. If user’s current position is in these areas, then server returns the feedback message to the device and in turn to the application. This is also shown in Fig. 3. The Ray casting algorithm is the most efficient algorithm among all point-in-polygon (PIP) algorithms. The algorithm counts the number of intersections for a ray passing from the exterior of the polygon to any point. If the count value is odd, it shows that the point is inside of the polygon. And if the count value is even, it shows that the point is outside of the polygon. We have modified the algorithm to get correct result when the point is in the boundary of the polygon, it shows that the point in the boundary is also inside of the polygon. The pseudo code of the algorithm is as follows: count ← 0 foreach side in polygon: if ray_intersects_segment(P,side) then count ← count + 1 if point is on an horizontal polygon boundary then return boundary if point is on the polygon boundary (other than horizontal) then return boundary if is_odd(count) then return inside else return outside 4) Dissemination of Disaster Warning and Evacuation Guideline: When the application finds its user in a disaster zone, it will give him a notification along with an audio message and vibration. Subsequently, the user of our application gets early disaster warning message both in visual and audio message on map of the application. CloudMade, a geographic related service provider, provides necessary data so that the application uses that data to draw the shortest path around the area of the user’s current location and shelter or safe area with proper guideline. OpenStreetMap renders its map for our application showing the two different places with different icon over the map. The TTF (Text-to-Speech) engine provides the audio direction to the user as evacuation guideline. In Fig. 4, it presents the utilizing tools of the application to disseminate warning message and evacuation guideline.5) Tracking Evacuation Progress: If the user is in any disaster zone, the application will automatically start another service on the mobile phone to track the user so that the authority can rescue him from probable disaster affected area. The user is identified with stored national id in the application. The application uses GPS to detect the phone’s current position and sends this data to the DMS for tracking the evacuation progress. When user comes to the safe area then application removes the tracking process. Fig. 5 shows the tracking evacuation progress process of our system.Utilizing OpenStreetMap, C2DM server and Disaster Management Server (DMS) as a third-party server, we have developed an android-based mobile application that provides early disaster warning and also helps to monitor the evacuation progress of the upcoming disaster. Our application will run on android 2.3.3 supported mobile phones and upgraded versions. The mobile phone should have GPS supported location identification facility. However, our application can identify the location of user through network provider of mobile operator. Our application has two major parts: (1) Early Disaster Warning and (2) Tracking Evacuee User. A. Early Disaster Warning: The application provides early disaster warning in two ways: 1) Automatic Warning Notification: There is a toggle button in our application’s homepage which starts or stops a service. When user starts the service as shown in Fig. 6(a), the application sends a registration request to the C2DM server and fetches a registration ID which is sent to the Disaster Managerment Server. It provides users an immediate notification on the application when there is any change in the server. The application in this case automatically starts a broadcast service that indeed communicates with the server taking user’s current position to check whether current area of user is probable disaster affected area or not. If the application can determine the current user area as vulnerable area to disaster then the application automatically notifies that to the user as Risk Message with vibration and audio sound which is demonstrtaedWhen user stops the service, the application unregisters the C2DM server and it doesn’t get any automatic notification from C2DM server even if there is any update in DMS database. 2) Manual Warning Notification: The application also has a reload button which connects application with the DMS and checks the current user area from the perspective of upcoming disaster. B. Tracking Evacuee User: The service which checks that if the user currently is in any probable disaster affected area, starts another service. This service sends the location of user to DMS when user changes his location. DMS stores the data in tracking of users as depictedOur application also has a button labeled as Show Message on Map. While the user presses that button it shows an OpenStreetMap on the screen as given in Fig. 8, identifying user’s current position and displaying shortest path to nearest shelter area.The menu has two options as shown in Fig. 9(a), one is to show the message as like Fig. 9(b) and the other one is to provide the audio direction to go to the shelter area.

The application also has an option to set the user’s National ID so that the application can track the user of that identity. The weather office updates the area based data of different disasters (tsunami, cyclone and flood) into DMS as shown in Fig. 10. Figure 10. Updating a region of probable disaster affected area along with disaster name and specific warning. V. EXPERIMENTAL RESULT We have tested our system assuming that a cyclone will hit in the coastal region of Chittagong, Bangladesh. We have utilized ten users (one of them is blind) in specified area who have installed the proposed application on their mobile phones. The total scenario is described according to the time in the Table I. TABLE I. SCENARIO OF THE EXPERIMENT Time Scenario 09:00 Experiment starts. Every user has started the service of the application on their mobile devices. 09:05 Weather Office updates a new warning of cyclone named Test for the coaster region of Chittagong. 09:06 Everyone in the updated area gets quick notification about the disaster. They check the message and follow the map to reach the shelter zone. The blind user follows the audio direction. The users are also tracked on the DMS. 09:15 Everyone except one can’t reach the shelter zone. 09:16 The rescue authority notices the unreached user and sends a rescue team to rescue him. 09:20 Everyone has reached safely at the shelter zone. The application stops tracking users. The above table shows how fast and reliably our proposed application has responded to the users. 

 VI. CONCLUSION Our location based disaster management system is an android mobile phone application employing OpenStreetMap (OSM), Google C2DM server, Disaster Management Server (DMS) as third hird-party server. Our application provides visual and audio disaster warning and evacuation help on the map of the application to user if the device user is in probable disaster affected area considering the user’s current location. This helps both normal and blind people to go to the safe area or shelter place prior to the disaster. Our application also facilitates the work of authority to track his evacuation progress ceaselessly so that they can take immediate steps if needed. Lack of details on OSM of Bangladesh is the main challenge of our work. The more volunteers working on OSM can make map of the country rich. In that case, our application will perform better in the context of Bangladesh. Moreover, we have a future plan to implement another application to assist in rescue and relief operation after the disaster and a better server side application to totally automate the system of detecting disaster prone area. REFERENCES [1] List of natural disasters by death toll [http://en.wikipedia.org/wiki/List_of_natural_disasters_by_death_toll] [2] United Nations International Strategy for Disaster Reduction Secratariat (UNISDR), Mortality Risk Index, 15 June 2009. [3] OpenStreetMap in Wikipedia [http://wiki.openstreetmap.org/wiki/OpenStreetMap] [4] Project Libya [http://wiki.openstreetmap.org/wiki/WikiProject_Libya] [5] Project Haiti [http://wiki.openstreetmap.org/wiki/WikiProject_Haiti] [6] D. H. Jeong, “National Disaster Warning System in Korea,” Sep. 2009. [7] Marius Cioca, Lucian-Ionel Cioca, and Sabin-Corneliu Buraga, “SMS Disaster Alert System Programming,”, pp. 260-264, Feb. 2008. [ Second IEE International Conference on Digital Ecosystems and Technologies] [8] Tobias SchernerandLothar Fritsch, “Notifying Civilians in Time – Disaster Warning Systems Based on a Multilaterally Secure, Economic, and Mobile Infrastructure”. [11th Americas Conference on Information Systems, Omaha, NE, USA, Aug. 2005] [9] Jayasinghe, Gamini, Fahmy, Farazy, Gajaweera, Nuwan, and Dias, Dileeka, “A GSM Alarm Device for Disaster Early Warning,”, pp. 383- 387, May 2007. [ 1st IEEE international Conference on Industrial and Information Systems] [10] Hidenori Torii, Jun Sawamoto, Norihisa Segawa, Eiji Sugino, and Yukinori Nomura, “Tsunami Early Alert and Evacuation Support System for Fishery Workers by Mobile Phones,” pp. 704-709, Apr. 2010. [24th International Conference on Advanced Information Networking and Applications Workshops] [11] Yasuaki Teshirogi, Jun Sawamoto, Norihisha Segawa, and Eiji Sugino, “A Proposal of Tsunami Warning System Using Area Mail Disaster Information Service on Mobile Phones,” pp. 890-895, May 2009. [International Conference on Advanced Information Networking and Application Workshops] [12] Anas Aloudat and Katina Michael, “Toward the Regulation of Ubiquitous Mobile Government: A Case Study on Location-Based Emergency Service in Australia,” Journal of Electronic Commerce Research, Vol. 11, Issue 1, Article 3, pp. 31-74, Jan. 2011. [13] Dugkeun Park, “One of the Nowcasting Applications: Early Warning Systems for Natural Disasters in Korea,” Oct. 2006. [14] Sufian Latif, K.M. Rakibul Islam, Md. Monjurul Islam Khan, and Syed Istiaque Ahmed, “OpenStreetMap for the Disaster Management in Bangladesh,” pp. 429-433, Sept. 2011. [IEEE Conference on Open Systems] [15] Android [http://en.wikipedia.org/wiki/Android_(operating_system)] [16] Android Cloud to Device Messaging Framework.