基于深度神经网络的室内定位算法

Indoor localization algorithm based on deep neural network

提出了一种基于深度学习的接收信号强度指示(RSSI)指纹定位算法。该算法将深度神经网络引入到指纹定位的2个阶段中:离线阶段对不同遮挡情况的指纹库进行特征训练,其中指纹数据作为输入,不同遮挡情况的指纹库编号作为标签;在线阶段将实时接收到的数据送入到网络中进行指纹库匹配,然后结合改进的加权K近邻(WKNN)算法进行定位。通过对比实验结果表明:所提算法的定位精度优于其他定位算法,有着很好的定位性能。

A received signal strength indication(RSSI)fingerprint positioning algorithm based on deep learning is proposed.The algorithm introduces the deep neural network(DNN)into two stages of fingerprint positioning:the offline stage performs feature training on the fingerprint database of different occlusion conditions,in which the fingerprint data is used as input,and the fingerprint database number of different occlusion conditions is used as the label;in the online stage,the real-time received data is sent to the network for fingerprint database matching,and then combined with the improved weighted K-nearest neighbor(WKNN)algorithm for positioning.Comparison experimental results show that the positioning precision of the proposed algorithm is prior to other positioning algorithms,and it has good positioning performance.