一种改进的RSSI指纹库定位算法

Advanced indoor position algorithm based on RSSI fingerprint

为了提高室内定位的精度,进行了信号强度RSSI之间的相关性的分析,提出了ID-WRKL算法。该算法将RSSI排序转换成AP指纹序列对并建立离线指纹库,其稳定性可以减小定位误差;再通过在线AP的选择,过滤噪点AP对定位估计的影响,减少计算量;最后根据Levenshtein距离得到最近邻的度量。在基于Map Reduce框架下的两个集合间的K-AP(P,Q)最近邻查询法基础上进行位置估计,提高了定位的精度。大量的对比传统KNN定位法的实验表明该算法的定位更精确,速度更快。

In order to improve the accuracy of indoor position, this paper proposed ID-WRKL algorithm by analyzing the cor- relation between RSSIs, which were sorted to AP sequence to build the offline fingerprint library. In order to effectively reduce the positioning error, the algorithm performed online AP selection to filter out the noise AP＇ s influence, as well as to reduce the amount of calculation. Next the algorithm utilized Levenshtein distance to obtain the measurement between the nearest neighbors, and adopted K-AP（P,Q） nearest neighbor query method between two collections based on MapReduce framework to estimate the position. It further improved the accuracy of indoor positioning by weighting the K nearest position pairs. A large scale positioning experiments prove that the ID-WRKL algorithm is more accurate and efficient than other indoor positio-ning methods.