Research On Keyword Search Method For Wireless Data Broadcast

With the popularity of mobile devices, the development of mobile computing and global positioning technologies, keyword search processing becomes more and more important in our daily life. Broadcasting is an efficient and scalable dissemination information technique in wireless computing environment. Under such environments, servers cyclically broadcast popular data items through the common channels; mobile clients tune in to the broadcast channels, and receive their own interested data items. Compared with the way of point to point data transmission, data item call be sent to all users who simultaneous access the data through data broadcast, and the cost of dissemination has nothing to do with the number of mobile clients, so data broadcast can meet a huge number of users to access information simultaneously. The wireless broadcast model is more scalable, and is suitable for system where there are large amount of users. Another advantage of this model is that the location privacy of mobile users is not leaked to the server. As the mobile devices are more popular, study about keyword search for wireless data broadcast have important theoretical and realistic significance.This paper proposes an effective code-compressed keyword search index in the environment of periodic data broadcast. Keyword search has undergone a great deal of engineering research and development, but considering the specialness, the traditional methods are unable to be properly applied in the wireless data broadcast. Inverted list is an index technique widely used in the keyword search. Inverted list index and hash-based stream index can’t be able to deal with the problem of a too large index structure. This paper proposes a new type of keyword index, and it manages to make the index structure smaller by a way of coding compression. The index converts the representation of inverted list to the form of two-tuple lists. The paper proposes that we can change the order of the document sequence to change the representation of the two-tuple lists during the construction of the index, which has effects on the size of the index structure. And we proves that the problem of finding the optimal document sequence to achieve the smallest index structure is an NP problem. Besides, the paper propose a scheduling algorithm for plain text files data broadcast, combined with the index construction. Lastly the paper compares the index proposed with the inverted list and hash base stream index and take real data set to process our module experiment. The experiment results demonstrate that our index structure is latency-and energy-efficient, and have a promoted performance compared with inverted list index and hash-base stream index.