The Study Of Multilevel Model Of Mobile Database Based On Clustering In Ad Hoc Network

Mobile Ad Hoc networks are a kind of specially self-organized, point-to-point,multi-hop wireless mobile networks(MANETs). Since1990s, with the popularizationof wireless network technology along with its application, Ad Hoc networks are ableto operate under independent network environment without network facilities whichresults from it’s own characteristics, such as centerlessness, self-organization,multi-hop routing, and dynamic topology, etc.. It is clear to recgonize thosecharacteristics of Ad Hoc networks: changing mobile nodes location, asymmetry ofnetwork communication,frequent disconnection, limits of memory property and rangeof processing capacity.Networks are usually supported by traditional mobile databasemodels which has lots of base stations, mobile exchange centers, and other basiccommunication equipments. Furthermore, they are supposed to be built around placeslike cities areas. There is no need to take environment conditions into account as toapplications.However, in Ad Hoc networks, speaking of mobile database, it is essential toconsider the energy of mobile main computer, limits of memory capacity, and limitsof bandwidth, as well as frequent disconnection with main database server, etc..Therefore, it will bring up a great deal of unsolved issues if we apply traditionalmobile database models to Ad Hoc network directly and recklessly. Given that, weshould improve those traditional database models for the purpose of adapting with AdHoc networks.Faced with these problems, this paper has come up with the theory of buildingnew mobile database models under Ad Hoc network clustering. It focuses onimproving performance of database under Ad Hoc network circumstances, reducingall kinds of expenses generated by database applications.The main works are as following:1.Based on the analysis of strengths and weaknesses of present Ad Hoc networkclustering algorithm,a self-adapting weighted clustering algorithm has been putforward on account of database application and massive data transmission. It realizes improvements on performance of data transmission under Ad Hoc network bycalculating clustering weights using parameters like relevance of nodes, relevantmobility, surplus energy value, communication bandwidth, and so on.Improvements are made especially towards mobile database models, in additionwith multilevel local database servers on the ground of Ad Hoc network clustering.We creates the Clustering Multilevel model of Mobile Database (CMMDB) whichconsists of three categories of nodes, i.e. mobile computer, multilevel local databaseservers, and main database server. Besides, this paper explains structures,characteristics, operating principles and application technology of CMMDB at length.3.This paper also brings up solutions to the problems which emerged along withadding multilevel local database servers.First of all, it gives solutions on reading and writing synchronizationrespectively. Especially on special cases of writing synchronization. We designedboth strong synchronization algorithm and weak synchronization algorithm ofstatistics in order to adapting different conditions.Then, it shows solutions on read-write conflict by proposing a new conceptionof reading data transactions, using broadcast within clustering to solve it.Secondly, it demonstrates solutions for repetition of transactions undermultilevel structures, analyzing all sorts of situations that could possibly occur. Andput up with solutions for each kind of those situations.