Distributed Fingerprint Matching Positioning Algorithm Research Based On Hadoop

With the development of mobile Internet, the user terminal access network more and more, the location based service (LBS) is a new mobile value-added business is developing rapidly, wireless positioning technology is the key technology of location based services has received wide attention. The location technology based on fingerprint matching has become a hot spot in wireless location technology in recent years because of its dependence on the network infrastructure and small interference by the wireless communication environment. How to apply the traditional fingerprint matching positioning algorithm in the wireless location environment with large amount of data has become an urgent problem to be solved.In this paper, the Hadoop distributed computing framework is used to realize the fingerprint matching and locating algorithm.Firstly, this paper designs fingerprint matching algorithms for UMTS network data provided by Huawei. Based on the nearest neighbor fingerprint matching algorithm, a fingerprint matching and location algorithm based on maximum expectation is proposed to improve the localization accuracy and efficiency. To improve the accuracy and efficiency of UMTS network data analysis, a hash algorithm based on maximum similarity of base station and a hash algorithm based on outlier detection are proposed. In this paper, the distributed implementation of these four localization algorithms is designed in detail, and a distributed fingerprint matching and location algorithm is proposed.The Hadoop cluster is build on Three PCs.Four sets of localization algorithms are applied to the five sets of data, and the results of the twenty sets of positioning accuracy and the time consuming of the algorithm are obtained. The experimental results show that the localization accuracy of the four algorithms is improved, the localization algorithm based on the maximum expectation and the localization algorithm based on the maximum similarity of the base station have improved the efficiency on the basis of the former algorithm. Compared with the localization algorithm based on the maximum similarity of the base station, the location accuracy is improved and the locating efficiency is almost the same as that based on the outlier detection. In this paper, we test the efficiency of the hash algorithm based on outlier detection on a single-node cluster and two-node cluster. Which verifies the efficiency of distributed computing in comparison with stand-alone computing.