Maintenance Strategy Of Data Consistency In Mobile Database

Mobile database as a typical application of mobile computing is deeply studied in academic teams. Mobile database is related to many technologies, such as:information processing, distribution computing, mobile communication and real time processing etc. lots of researchers have focused on that. Data consistency is one of the most important research areas. Data consistency problem exists in many mobile database research directions, such as:data replication、mobile transaction process and concurrency control strategy、 data broadcasting and mobile information publication, and we study it together in this paper.In order to expediently discuss the data consistency this paper referred to, a uniform concurrency control simulation platform (based on C++SIM) and performance index are designed. Based on all kinds of recommended parameter value, we can compare all kinds of models and algorithms easily.Data replication is the key problem of mobile database. It requires guaranteeing the consistency after replicas being updated repeatedly in mobile environment. what’s more, it should provide the replication distribution model and the transaction processing specification of the mobile database. However, with the development of the hardware of mobile device and the technology of mobile communication, the traditional architecture of mobile computing (designed for the mobile database) should be changed and is not suitable for mobile computing any longer. According to the feature of the new mobile computing architecture, the classic two-level replication model is extended and the EMANET transactional multi-level replication model with data consistency is proposed. What’s more, two models are analyzed and compared.Concurrency control as the key mechanism of transaction processing is to make sure that the mobile database is in consistent status during the transaction processing. The concurrency control strategy can be divided into two catalogs:Pessimistic concurrency control protocol (abb. PCCP) and optimistic concurrency control (abb. OCCP). In mobile computing environments, the most frequent information operation of the end user is to query the database, according to the patterns of writing and reading data in mobile application, OCCP is more proper for the using condition than PCCP. OCCP can achieve better effect when there are more read operation in transactions. On the other hand, there exists the unexpected frequent disconnection in mobile environment. If using the lock based PCCP, it will waste plenty of time to recognize and detect the conflict of transaction, also will increase the communication cost between the mobile clients and the database server. Therefore, in the frequent disconnection environment, OCCP is more proper than PCCP. By improved time stamp and the method of RW time point, the RW and WW conflict operations in the BOCC protocol (which belongs to OCCP) are optimized. What’s more the improved BOCC-RW validation protocol is proposed with its correctness being proved. The simulation experiments prove that BOCC-RW protocol has the higher transaction throughput, lower transaction mortality rate and the guarantee of data consistency.Data broadcasting divides data into two groups. The first group is called hot data which are frequent required by most end users and the other one is called cold data. The hot data is broad casted to the cell periodically to make sure the mobile device can access the data objects easily, by making use of the bandwidth asymmetry between server and mobile client. The obvious advantage of data broadcasting is that its bandwidth consumption has nothing to do with the scale of the cell; and it can resolve the frequent disconnection. What’s more, it is good to maintain the timeliness and efficiency of data. We improved the existed algorithm called UFO:In the case of UT transactions’occurring frequently, SBS algorithm is deployed in server side and CRS algorithm in client, which allow the MH need not to wait the reset of the BT any more. The improved broadcast scheduling algorithm is more suitable for the read-only mobile clients. We can prove that the new algorithms can short the data access time and decrease the times of transactions reset such that the schedule efficiency is better than the UFO in the same data consistency condition.