Mobility information and mobile transaction processing

Abstract

In a mobile environment, system and application performance is impacted by mobility factors such as varying bandwidth, frequent disconnection, and changing locations. These factors make mobility information more important in the design of distributed algorithms. However, at present, no general model exists for exploiting the availability of mobility information. Focusing on this issue, this thesis investigates mobility environment modelling; especially the classification, representation, and utilisation of mobility factors. Based on the concepts and methods of modelling mobility factors proposed in this research, an adaptive transaction scheduling (ATS) approach for mobile transaction processing, chosen as a particular application domain, is developed to demonstrate the feasibility of designing better distributed applications by making use of mobility information. The ATS approach is designed to support autonomous operations during network disconnection and dynamically select the site of transaction execution based on mobility informa ion such as bandwidth and disconnection ratio. The performance of the ATS approach is evaluated against fixed transaction processing policies by the means of simulation experiments. A queue simulation model is designed and implemented in discrete-event simulation language. The results of the simulation experiments show that adaptive transaction processing produces better performance and confirms that the design of applications in mobile environments can benefit from available mobility information.