一种基于蓝牙室内指纹定位的贝叶斯改进算法

贝叶斯估计是重要的位置指纹定位算法,但传统的等值贝叶斯先验概率在动态定位中不适用。针对该问题,本文提出了一种基于贝叶斯指纹定位的改进算法。首先,借助陀螺仪获取的航向信息和高斯核函数模型建立概率投票算法,计算先验概率;然后,结合先验概率和信号强度计算待测点位于参考点上的后验概率;最后,选取概率最高的参考点,以概率为权重计算待测点的最或然值。以智能手机为试验对象,在规则路径试验中,改进算法的平均定位误差为1.15 m,定位误差小于2 m的概率为96.1%,不规则路径试验中,平均定位误差为0.50 m,定位误差在1 m的可信度为94.8%;并且改进算法对定位中位置跳变的现象有明显改善,具有较好的稳健性。

An improved algorithm of Bayesian fingerprint localization based on bluetooth

Bayesian estimation is an important position fingerprint localization algorithm, but the traditional equivalent Bayesian prior probability is not applicable in dynamic localization. In view of this problem, an improved algorithm based on Bayesian fingerprint localization is proposed in this paper. Firstly, by means of the heading information obtained by gyroscope and Gaussian kernel function model, the probabilistic voting algorithm is established to calculate the prior probability. Then, the prior probability combined with the signal strength is used to calculate the posterior probability of the point to be measured at the reference point. Finally, the most probable reference point is selected and the most probable value is calculated with the probability as the weight. In the regular path experiment, the average positioning error of the improved algorithm is 1.15 m, and the probability of positioning error less than 2 m is 96.1%. In the irregular path experiment, the average positioning error is 0.50 m, and the reliability of positioning error is 94.8%. In addition, the improved algorithm can improve the location hopping phenomenon and has good robustness.