基于接收信号强度(Received Signal Strength Indicator,RSSI)位置指纹定位算法可避免RSSI测距问题,通过滤波处理可构建高质量指纹数据库.针对高斯滤波运算时间长而均值滤波定位误差大等问题,通过分析RSSI分布规律,基于部分RSSI重复出现的特性,提出一种众值滤波方法,通过逐次选取当前数据中众数,直到所选众数的频数累积和与数据总量之比大于95％时取平均提取特征RSSI.实验表明,基于众值滤波构建的指纹数据库在73％的情况下与高斯滤波定位精度相近,但运算效率优于高斯滤波.同时,相对于均值滤波,众值滤波将定位精度提升约0.15 m,在95％的情况下将最大定位误差限制在2 m以内,可有效提高指纹数据库质量.
Establishment of Indoor Location Fingerprint Database Based on Crowd Filtering
The fingerprint location algorithm based on the received signal strength indicator can avoid the problem of RSSI ranging.High-quality fingerprint database can be constructed by filtering.Aiming at the problems of long operation time of Gaussian filtering and large positioning error of mean filtering,a method,of crowd filtering based on the repeated occurrence of some RSSI,has been proposed according to the distribution of RSSI.By selecting the mode in the current data one by one until the ratio of the frequency cumulative sum of the selected mode to the total data is greater than 95％,the feature RSSI is extracted by averaging.According to the experimental results,the fingerprint database based on crowd filtering is close to Gaussian filtering in 73％of cases,but the operation efficiency is better than Gaussian filtering.At the same time,compared with the mean filter,the crowd filter improves the positioning accuracy by about 0.15 m and limits the maximum positioning error within 2 m in 95％of cases,which can effectively improve the quality of the fingerprint database.
NETL
NSTL
万方数据
