Ph.D. Dissertation Defense - Kim Kihwan

Date: 15 Jan, 2009
Time: 10:00 AM
Location: 223 Atanasoff Hall
Topic: Managing Motion Triggered Executables in Distributed Mobile Databases
Major Professor(s): Ying Cai and Wallapak Tavanapong


Abstract:

This thesis presents a new distributed mobile data management called Geotasking. The hardware platform of a geotasking system is simply a set of position-aware mobile nodes. These nodes connect with each other through wireless networks and together they facilitate the system with storage capacity and computing capability. Geotasking is featured by its executables, each being a special program called a geotask. Unlike regular programs, a geotask is associated with a user-defined geographic region and its execution is triggered by the movement of mobile nodes. In an environment where mobile nodes are the only computing devices, efficient geotask management is challenging. Different geotasks may have different execution triggers. They must be stored among mobile nodes and retrieved for timely execution whenever their execution triggers are fired. As a step toward cost-effective geotasking, we classify geotasks into a number of categories and present a novel technique for efficient management of a basic type of geotasks. Our technique caches geotasks among mobile nodes and does so dynamically to keep the geotasks close to where they are implanted, thus minimizing communication costs incurred in geotask management. With this solution in place, we then investigate how to support other types of geotasks. The proposed techniques convert each of these advanced types of geotasks into a set of basic type of geotasks, making it possible to develop and deploy all types of geotasks under a single unified platform. We analyze the performance of the proposed techniques in terms of mobile communication costs with a detailed mathematical model. We have also developed a detailed geotasking simulator. The simulator allows us not only to verify the accuracy of the analytical model, but also evaluate the performance of our geotask management platform from various aspects.