论文部分内容阅读
目前已有的位置指纹室内定位算法大多都是建立在原始指纹库的基础之上,指纹库的建立精度会直接影响到最终的定位精度。为此,通过对指纹数据的研究,提出一种基于正态检验的室内定位算法。训练阶段,首先对每个指纹点接收到的信号RSSI样本进行正态假设检验,若接受假设则选用正态分布函数对其总体进行概率密度估计,否则选用核函数对其总体进行概率密度估计,最后取大概率信号的均值建立高精度的指纹数据库。在线定位阶段,使用K加权邻近算法(WKNN)估算位置,实验结果表明提出的算法定位精度较均值模型法以及正态模型法都提高了15%以上。
At present, most indoor fingerprint location algorithms are based on the original fingerprint database. The accuracy of the fingerprint database will directly affect the final location accuracy. Therefore, based on the research of fingerprint data, an indoor location algorithm based on normality test is proposed. During the training phase, the RSSI samples received at each finger print point are firstly tested by normal hypothesis. If the hypothesis is accepted, the normal distribution function is used to estimate the overall probability density. Otherwise, the kernel function is used to estimate the probability density, Finally take the average of the probability signal to establish a high-precision fingerprint database. In the online positioning stage, the position was estimated by using K-weighted neighboring algorithm (WKNN). The experimental results show that the proposed algorithm possesses an accuracy of 15% or more compared with the average model method and the normal model method.