RBF与加权质心相结合的室内无线定位算法

A new indoor wireless positioning algorithm combining RBF and weighted centroid

针对传播路径损耗模型的参数,极易受室内障碍物等环境因素影响,导致定位精度低的问题.利用RBF(径向基函数)神经网络算法替代损耗模型,拟合RSSI(接收信号强度)值与距离的关系.采集室内RSSI值和其对应的距离值的实测数据,利用实测数据训练RBF神经网络,建立RSSI-距离拟合模型;利用拟合模型将经过处理的RSSI值转换为距离值,并将距离值按从小到大排序;取前3个离定位节点较近的固定节点的信息,进行加权质心定位计算.研究结果表明:RBF算法的定位精度比路径传播损耗模型算法提高了34.5%,且略高于BP算法的定位精度.在相同的室内环境下,RBF算法能更好地克服环境因素对距离计算的干扰,提高室内定位的精度和稳定性.

Considing that the propagation path loss model＇s parameters are easy to be affected by indoor obstacles and other environmental factors, which would result in a low positioning accuracy, a new ranging method using RBF neural network algorithm, instead of the propagation path loss model to fit the RSSI-distance model is developed. First collect RSSI values and their corresponding distance data measured indoor, trained RBF neural network with the measured data and establish the RSSl-distance fitting model; then use the fitting model convert the pretreated RSSI values into distance data, and sort the distance data; take the former three fixed node information, and calculate coordinates with the weighted centroid localization algorithm. Experiment results show that： the positioning precision of RBF algorithm is improved by 34.5% compared with the propagation path loss model, and slightly higher than the BP algorithm. At the same indoor environment, RBF algorithm can better overcome the interference of environmental factors on the distance calculation, and improve the indoor positioning accuracy and stability.