Research Of Fingerprint Algorithm Based On Hadoop In WCDMA Network

The Value-Added Service which as a new growth performance for operators has beengot more and more attention along with the rise of3rdgeneration business. Andthe realizationof manytelecom value-added service are based on the location technology.Owing to nonline-of-sight affected by wireless transmission,the traditional positioning technology such asTOA, TDOA and AOA, RSSI cannot provide precise location services. Therefore, to find alocalization algorithmwhich can overcome the wireless transmission, simple, easy to realizehas become the research emphasis in this paper.Position Fingerprint Location Algorithm is an innovation which the pattern recognitiontechnology used in the field of wireless communication. Its implementation is divided intostages of off-line training and on-line positioning two stages. The off-line training phasemainly collects signal value information which was transferred from base station to mobileterminal, then through data processing, form the off-line fingerprint database which for thephase of online matching positioning. The mainly work on on-line positioning phase is toinput the fingerprint information which was collected by test terminal, using certain matchingalgorithm to match the fingerprint database which was established in off-line phase, then getthe tested terminal’s estimated position.In order to obtain quite precise localization service, the huge off-line fingerprint databasewas needed to build during the off-line phase. In view of the fact that under the existingfingerprint localization algorithm and huge off-line fingerprint database circumstance, thematching time is too long. This paper combines with the MapReduce distributed frameworkofHadoop cloud computing platform, based on the WCDMA cellular network, Hadoop cloudcomputing platform based fingerprint localization algorithm is proposed.First of all, after consulting numerous relevant documents, this paper introduces cloudcomputing technology, especially the current research of Hadoop open source platform andthe application trend of fingerprint positioning technology. Then, through the analysis of the WCDMA network architecture, contrast the pros andcons of various positioning technology of WCDMA cellular network, according to thecharacteristics of the position fingerprint positioning technology and the technologyadvantage of Hadoop/MapReducedistributed framework, designed the WCDMA networkposition system.Finally, It emphasizes on KNN(K-Nearest Neighbor) algorithm and it’s implementationbased on MapReduce. Mainly design the relative map function and the reduce function. At theend build Hadoop cloud computing platform, using the RSCP and the corresponding latitudeand longitude which were got from the road test measured on WCDMA network as afingerprint database, compare the matching of time using the MapReduced KNN algorithmunder the condition of single node and double node, prove the advantage of MapReducedKNN algorithm.