基于蜂窝网格划分的指纹定位算法的改进

Improvement of fingerprint positioning algorithm based on cellular meshing

针对传统的K近邻算法计算量大、定位实时性低的问题,提出一种基于蜂窝网格的改进方法。对收集的RSS采用高斯滤波处理,正确显示无线信号的波动特性;针对定位现场会出现不同楼层的情况,提出根据位置指纹内的MAC地址区别不同楼层指纹的方法;使用K-means算法对指纹库聚类,缩小指纹搜索空间,在二分法的投票机制下确定最终的定位区域。仿真结果表明,在缩小定位区域之后,在保证定位精度的前提下,大幅度缩短了定位时间,保证了定位的实时性。

Aiming at the problem that the traditional K-nearest neighbor algorithm has a large amount of computation and low real -time positioning, an improved method based on cellular grid was proposed. Gaussian filtering was applied to the collected RSS to correctly display the wave characteristics of wireless signals. According to the location of different floors, a method was proposed to distinguish the fingerprints of different floors according to the MAC address within the location fingerprint. K-means was used to cluster the fingerprint database, the fingerprint search space was narrowed down, and the final location area was determined under the voting mechanism of the dichotomy. Results of simulation experiment show that, after the positioning area is reduced, the positioning is greatly shortened on the premise of ensuring the positioning accuracy. Time guarantees real-time positioning.