模糊聚类下的接入点选择匹配定位算法

Access point selection matching localization algorithm based on fuzzy clustering

针对指纹定位中传统聚类方法难以有效划分物理空间以及信号源不稳定导致定位误差大的问题,为降低数据库存储成本与提高指纹质量,提出了一种基于模糊聚类的精简接入点匹配定位算法。该算法在离线阶段将面积较大的目标区域按信号源特征划分为多个重叠模糊分区,综合考量各分区中信号源的稳定可见性及冗余性等多尺度特征,建立区域最小接入点辨识集合,在提高定位实时性的同时削弱不稳定接入点信号带来的匹配失准的缺陷。在位置计算阶段结合区域接入点稳定性特征分配近邻点权重,以此对传统欧氏距离进行改进,同时依靠待定位用户运动过程中相邻时刻间的速度约束关系筛选定位野值,克服环境和信号源变化对定位结果带来的不利影响,降低定位误差。经实测场景测试,所提算法在对接入点进行有效筛选的前提下降低了定位算法的运算消耗,在显著减少离线数据存储量的同时,将定位场景的平均定位误差控制在了1 m以内,较已有经典定位方法,其定位精度提高了15%以上。

Considering the problem that it is difficult for the traditional clustering method to divide physical space effectively,and that the positioning error is large due to instability of the signal source,aiming at reducing the storage cost of database and improving the quality of the fingerprint,this paper proposes a simplified access point matching location algorithm based on fuzzy clustering.According to the proposed algorithm,in the offline stage,the target space of a large area is divided into multiple overlapping fuzzy partitions by the characteristics of the signal source,the stability,visibility,redundancy and other multi-scale characteristics of the signal source in each partition are comprehensively considered,the smallest access point identification set in the area is established,the positioning speed is improved and the defects of mismatch caused by the unstable access point are decreased.In the position calculation stage,the traditional Euclidean distance is improved by assigning the weight of the neighbor points in combination with the stability characteristics of the regional access points,and the speed constraint relationship between adjacent moments during the movement of the user to be located is used to filter the positioning outliers,to overcome the changes in the environment and signal sources.The unfavorable influence of the locating error is reduced.Tested on actual scenes,the proposed algorithm reduces the computational consumption of the positioning algorithm on the premise of effectively screening the access points,while significantly reducing the offline data storage and controlling the average positioning error of the positioning scene within 1 m.Compared with the existing classic positioning methods,the positioning accuracy of this algorithm is improved by more than 15%.