Summary: | 博士 === 國立成功大學 === 資訊工程學系碩博士班 === 92 === With the advance of wireless communication and VLSI technologies,the emerging mobile computing becomes one of the most attractive research topics in the past decade.Comparing to the previous research in traditional environments,the novel computing paradigm has new factors and limitations in the wireless mobile network,such as mobility of clients, scarce power of mobile devices,limited bandwidth, $ldots$, etc.These distinctive characters bring many new challenges to the database community and the communication society, including location management, data transmission,mobile service management, $ldots$, etc.In this dissertation, we discover certain fundamental and unsolved issues,and proposed effective solutions to these problems based on the current existing mobile computing architecture.
For location management issue, past works focus on improving either the movement operation or the locating operation. We propose the Cache and Carry (C&C) Strategy to improve both of them simultaneously to improve the performance than previous works. In particular, our proposed strategy successfully reduces a lot of global communication operations to local communication operations. Our performance results show that the C&C strategy significantly outperforms the other strategies.
We also address the issue of delivering large-size data for mobile users under a limited bandwidth wireless environment. Fully replicating the data over all cells is a straightforward solution, but is impractical because it not only wastes resource but also degrades the overall performance due to heavy communication load. We resolve this problem in two stages. To fast look up the geographic information, we first devise an encoding scheme to encode the geographic cell, and describe useful properties of the encoding scheme. Then, we propose a dynamic-programming based method to efficiently find out those most possibly resided cells to pre-sent large-size data before the mobile user visits. Experimental results reveal that our proposed algorithm dramatically improves the performance.
We study the optimization issue of the service mobility for the management of mobile services. A mobility controlling mechanism consisted of the request operation and the movement operation is designed for a mobile agent (MA). Then, we develop a probability model and derive the cost functions for the proposed mechanism.Based on the cost functions, the mobile agent is able to determine a best resident location to serve the roaming mobile user. Performance study shows that communication cost of employing our proposed mechanism is dramatically reduced.In addition, we discuss certain implementation issues about maintaining the duplicate data items in a MA-resident site.
Finally, we consider the data scheduling problem for the time-constrained services in the multi-channel data dissemination systems. The time-constrained services are widely used in many applications,but not solved in past related work.We first derive the properties that a broadcast schedule must have in order to satisfy a set of given time constraints. Next, we present a theoretical lower bound on the number of channels for offering the time-constrained services. Then, we propose a scheduling algorithm that owns all the necessary properties and always employs the minimum number of channels. And,the produced schedule always meets the requirement of time constraints of the clients.
|