高斯过程回归与参考点排序相结合的新WiFi室内定位系统

为解决WiFi指纹定位系统的定位精度易受环境变化影响的问题,提出了一种新的WiFi室内定位系统。该系统包括一种新的定位算法和指纹更新算法,定位算法首先通过将参考点与测试点之间的信号强度误差转换为参考点排序,然后利用模糊思想将处于同一范围内的排序视为与测试点具有相同的相似度,以削弱信号自身动态变化对定位的影响,进而筛选出相似度较高的前k个参考点进行加权定位;指纹更新算法利用WiFi信号的传播模型和基于Matern核的高斯过程回归进行指纹更新。其次,利用在真实环境下采集的数据进行算法的测试验证。实验结果表明,提出的定位算法和指纹更新算法相较于传统算法,定位精度和指纹更新精度能够分别提升25%和34%以上。

WiFi Indoor Positioning System Based on Gaussian Process Regression and Reference Point Ranking

In order to solve the problem that the positioning accuracy of the WiFi fingerprint positioning system is easily affected by environmental changes, a new WiFi indoor positioning system is proposed in this paper. A new positioning algorithm and a fingerprint updating algorithm are consisted in this system. In the positioning algorithm, the received signal strength error between reference point and the test point is converted into a reference point ranking, and fuzzy theory is used to consider the rankings in the same range as the same similarity with TP to weaken the impact of the signal dynamic changes on positioning, and then k reference points with top similarity are selected for weighted average positioning. In the fingerprint updating algorithm, the WiFi signal propagation model and Gaussian process regression algorithm based on Matern kernel are used to update the fingerprint. Then, the WiFi data is collected under real environment to verify the effectiveness of the proposed algorithms. Experimental results show that the proposed positioning algorithm and fingerprint updating algorithm can improve positioning accuracy and fingerprint updating accuracy by more than 25% and 34%, respectively, compared with traditional algorithms.