Research Of Key Technologies On Mobile Computing

With the development of information era, people hope to access data at anywhere and anytime. To meet this requirement, mobile computing is researched as the combine of computing and wireless communication technology. Because mobile computing has characteristic in itself, such as unsymmetrical wireless bandwidth, mobility, disconnection, and so on, the traditional distributed computing technology cannot fit the mobile computing.The author does mostly work on three key technology of mobile computing: cache replacement algorithms, mobile transaction model and QoS of mobile computing.On research of cache replacement algorithm, author proposes two algorithms: Lowest Cost Further Away Replacement and Pre-Defined Regions Semantic Circle Replacement. LCFAR algorithm introduce the cost function as standard of replacement, not only thinking about distance but also frequency of visitation,to improve the weakness of only thinking about semantic segment distance in FAR. PDRSCR algorithm thinks about the relationship between the semantic valid region and cache matching through introduction of pre-defined region and semantic circle. This algorithm is more suited to the query, which has location relativity.On research of mobile transaction model, author mainly thinks about accessed hot data and network environment of committing transaction to propose a new transaction model- Adopt Mobile Transaction Model Based on Weight. This model can be disposed into three sub-models on the base of hot data and network connection. This model also defines the transaction identifier, the mechanism of locking and detecting conflict. At last, three sub-model processing is proposed. This model can gain the advantages of O2PC-MT and Pre-Write Model, and has better performance than these two models.On the research of QoS, mainly work is how to predict the motion track of mobile user. In this paper, author proposes a new method-LPBRBF to predict motion track with RBF neural network, gives the method of history data gain and relativity data maintenance policy. Though the history data and experiments, the RBFNN parameters are confirmed. Moreover, experiment results of location predicted and analysis are given. At last, two resource ReSerVation instances are analyzed and...