Mobile Location Based Indexing of Disaster Notification System

Abstract

Myanmar is one of the countries that geographically located in the disasteraffected area because of the climate changes and environmental conditions. There is no way to prevent the natural disaster, but its impacts can be reduced or rescued. Adequate prior disaster information can save a significant number of lives and properties. Therefore, the accurate alerts or notifications about disasters are needed for Myanmar people. Now, there are many mobile phone users along with the technological progress around the world including in Myanmar. Thus, mobile devices become the most convenient communication tools which have not time and place limitations. A suitable disaster notification system based on mobile phones is one of the useful things as it is also a requirement of Myanmar people. Because of the fast and easy way, notification messages via mobile phones take benefits in the communication activities. The main service task of this system is the delivery of possibly disaster information to mobile devices which are in the imminent disaster region. The system finds whether mobile is within a predefined area using its current location. The users who are in the imminent disaster area will receive the required notification messages. In this system, the server gets the current mobile position and keeps their location update structurally are a great challenge along with the continuously changing in the position of mobile devices. Therefore, a suitable technique is needed to store and update mobile positions. Moreover, when a message delivery to mobile devices within the specified range, the factor to be considered which mobile should be sent the message first. One of the most suitable solutions is sending the nearest mobile locations as fast as possible. In this system, a two-dimensional balanced structure, a presort-nearest location index tree is proposed that allows maintaining, updating, and circular range querying mobile objects within the required time. It also supports generating nearest locations by index structure from the desired query point. In this structure, all of the location nodes are placed by level order thus nodes at any distance can easily find without traversing the whole tree and the searching time may reduce greatly. It is harmonized to solve nearest neighbor location queries since the locations of the datapoints are based only on their relative distances from each other. In addition, all mobile locations in the range will be available by the distance at one time. Then, the system architecture is built for sending notifications by connecting with firebase cloud messaging (FCM), application server and mobile devices in the imminent disaster area. To overcome being unnecessary updates at the server, Hybrid Update Algorithm is proposed in this system. In this case, a virtual mobile dataset is needed to access several of moving mobile locations for performance evaluation. Thus, a synthetic mobile location generator is proposed that is based on the creation of two-dimensional mobile locations. As a result, this generator is free from location privacy and confidentiality. The necessary performances are tested by using a JUnit testing schema, which can automatically apply to run in testing functions. For performance evaluations, the execution time, updating time and CPU usage are measured by comparing between proposed presort-nearest location index tree, presort range tree and KD tree according to the evaluation of tree construction, range searching and neighbor searching over moving objects. Besides, the distance-based method is applied for comparison of two of index structures.