Technology Research Of Embedded Mobile Database

As a kind of new rising technology, Mobile Embedded Database is an important branch of traditional distributed database. Because of a quite especially running environment, much technology used in distributed database system cannot go with Embedded Mobile Database. Therefore, further research on Embedded Mobile Database technology has become a hotspot in this field.Embedded Mobile Database, which runs on motive embedded devices, is an essential part of whole mobile database system. Instead of being analyzed and researched isolatedly, a lot of technology involved should be extended and deepened within the whole system. This paper starts from the model of Mobile Database System, analyzes the function of Embedded Mobile Database in the entire system, and explores thoroughly and studies deeply the key techniques in this field according to the motive characteristics, such as the motion of client, the frequent break and connection between client and network, the variety of network condition, the non-symmetry of network communication and so on.This paper stresses on the Management algorithm of Client Cache in Embedded Mobile Database, and mainly extends on three aspects: the Replacement Algorithm, Prefetching Algorithm, and non-Consistency of Data. The Cache Management of Mobile device is based on the mobile database running on multi-disk broadcast environment, which is characterized as not requiring upload channel. In this case, the client can continuously receive expected data stream without sending requests to the server. On the other hand, according to client access requests, the server can attempter the data item broadcasted in sequence so as to make different data items broadcast in different frequency.The paper is organized as follows. Firstly, we studies the Replacement Algorithm of Cache Management by analyzing two sutra Algorithm: LRU and 2Q. We also prove the considerable suitable Cache Replacement Algorithm of multi-disk broadcast-PIX Algorithm by using mathematics method. Through these analyses and investigations, we provide an ideal algorithm - 2QIX, which can run stably in multi-disk broadcast environment. This algorithm has integrated the advantages of 2Q and PIX algorithm. It considers not only the issue of the associate access of data item, but also the broadcast frequency gene. It can estimate the futureaccess probability of data item via the history access records of data item, and, by experiments, it turns out to have better effects.Secondly, we discuss the Prefetching Algorithm. At the beginning, we analyze a reasonable Algorithm - PT algorithm. Using Prefetching Algorithm, the client can continuously receive expected data items in broadcast environment. Therefore, Prefetching Algorithm works quite well in such environment. However, PT algorithm has its restrictions in practice. In this case, we provide a feasible Prefetching algorithm - EPT algorithm. This algorithm makes PT algorithm possible by estimating the future access probabilities via the history access records of data items. But it demands the mobile device of better computing ability, which can complete the comparison of pt value at every broadcast unit interval.Thirdly, we discuss non-Consistency of Data for Cache Management. Considering the necessity and inevitably of Data non-Consistency in application environment, we provide two keys to the problem: Invalidation Setting and Propagation Updating Value. Since these two methods have their own features, it will be a pretty good solution if we can combine them together according to their different status.Finally, we build the executable model of Mobile Database system, testifies all algorithms discussed throughout the paper and validated its performance by trials. The conclusion of this paper will have certain contribution for the application of Embedded Mobile Database.