关键词: 矿井指纹定位, Hadoop 云平台, MapReduce, 指纹匹配

Abstract: Aiming at the current problems of the low positioning accuracy and weak real-time performance of existing mine fingerprint localization algorithms,a Hadoop cloud platform-based mine fingerprint localization algorithm is proposed. The proposed algorithm first preprocesses the multi-base station signal sources in the mine before fingerprint feature extraction to improve the reliability and robustness of the fingerprint features. Secondly,the source fingerprint signal is transformed linearly using wavelet basis functions,and the signals from multiple base stations are decomposed into frequency bands of different scales to obtain representative fingerprint feature vectors. Furthermore,for each location-based signal of the miner or mobile device,it is decomposed into several sub-signal regions,and the correlation coefficient matrix between the test signal source and the feature set is calculated for each region,based on which matching of fingerprint information is implemented. Finally,using the doppler effect and signal attenuation principle,estimate the position of each node to obtain specific location information for each node. A localization experiment is conducted on a mine in China,and the results showed that compared with mainstream fingerprint localization algorithms,the proposed mine fingerprint localization algorithm can quickly and accurately locate people or mobile devices in the mine,with high positioning accuracy and real-time performance. The proposed algorithm can be applied to practical mine safety monitoring and management,providing support for mine safety and efficient production.

Key words: mine fingerprint localization,Hadoop cloud platform,MapReduce,fingerprint matching