The effectiveness of a full-mobility architecture for wireless mobile computing and personal communications
Abstract: The emergence of wireless mobile computing, especially the mobility of computers, introduces challenging new problems that were encountered in neither the design and implementation of conventional mobile networks or in fixed computer networks. A critical and challenging problem, for mobile computing and personal communications, is how to fully cope with the special characteristics of the mobile wireless environment, i.e., to make balanced usage of computation and communication while taking advantage of and support the user's mobility, via: location-awareness, knowledge of a user's mobility behavior, i.e., mobility-awareness, and aggressive mobility management. The last allows systems to dynamically configure themselves and makes applications aware of the characteristics of the dynamically changing connection(s) (such as bandwidths, latencies, etc.) so that they can adapt as necessary. This dissertation addresses the above stated mobility problem for wireless mobile computing and communication systems. A partial solution is presented in our novel full-mobility architecture, consisting of a mobile-API, mobile-floating agents, location-sensitive information management, mobility/location-aware dynamic caching and prefetching techniques, and a predictive mobility management scheme. This full-mobility architecture explores the various mobility aspects relevant to mobile systems: personal mobility, terminal mobility, service mobility, resource mobility, data mobility, and computing environment mobility. The mobile-API provides a framework for integrating location-sensitive information and predictive mobility management, in order to explore the mobile nature of mobile users and location-sensitive nature of mobile systems. The mobile-floating agents apply concepts developed originally for "virtual-memory" to mobile systems, in order to support mobile virtual-distributed systems. They provide both the means and methods to integrate caching and prefetching techniques with the mobile feature of mobile systems. In addition, they unbundle services and resources from the underlying network and allow them to follow the user. The concepts of location-aware dynamic caching and prefetching have also been adapted to the wireless mobile environment and to be aware of mobility. Furthermore, a predictive mobility management scheme has also been proposed in this thesis, which makes it possible to have service pre-connection and resource pre-allocation. Thus, it makes mobile systems more intelligent and allows anticipation of the change in location and the mobility behavior of a specific user. By combining predictive mobility management, mobile-floating agents, and location-aware dynamic caching and prefetching techniques, a mobile system can dynamically cope with the mobile nature of mobile users by providing service and resource mobility and intelligent service pre- connection, resource pre-allocation and data structure pre-configuration. This makes it possible to allow the systems to dynamically configure themselves and to make applications aware of impending changes in connection bandwidths, latencies, etc., so that they can adapt as necessary. It can also provide a "soft data-structure handover" scheme to efficiently avoid the problems associated with the "long communication pause during the handover between cells". In this thesis, both detailed theoretical studies based on modeling and performance analysis and practical design and implementation were conducted. We have analyzed and evaluated the performance of this full-mobility and mobile-floating agent scheme with location-aware dynamic caching and prefetching strategies. Detailed analysis and evaluation of these schemes in terms of latency reduction and increased system overhead versus mobility density with different assignment algorithms are given. The basic concepts were also explored via prototypes which have been designed and implemented. The performance results clearly indicate the feasibility and the performance gain of the proposed solution.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.