基于零先验知识的室内指纹定位优化算法

Zero Prior Knowledge-based Indoor Fingerprint Positioning Optimization Algorithm

针对众包方法构建指纹库数据质量低和指纹模糊相似性等问题,提出了一种基于SLAM技术的指纹数据质量优化和指纹唯一性增强的算法。利用ICNN算法完成数据关联优化,并在指纹向量中加入可信度指标来优化指纹更新过程;随后针对指纹模糊相似性设计了基于高斯插值的指纹唯一性增强优化算法,保证了指纹数据质量,从本质上提高了指纹定位的性能。实验表明,该算法能够将指纹定位的中位数误差从原始指纹的3m提高到2m,最大定位误差从8m左右下降到4m以内。

This paper proposes an algorithm based on fingerprint data quality optimization and fingerprint uniqueness enhancement,focusing on the low quality of fingerprint database and fingerprint fuzzy similarity problem.It utilizes the ICNN algorithm to optimize data association,and adds a confidence value in the fingerprint vector to optimize the fingerprint updating process,then designs an uniqueness enhancement algorithm based on Gaussian for fingerprint fuzzy similarity,which guarantees the data quality of fingerprint,and essentially improves the performance of fingerprint positioning.Experiments show that the proposed algorithm can reduce the median error of fingerprint positioning from 3 mto 2 m,and the maximum positioning error is reduced from8 mto 4m.