提出一种基于可见光通信的BP神经网络室内定位算法, 首先通过MDS-MAP算法和最小二乘法获得全网节点的相对坐标, 再利用信源节点的坐标信息得到网络内所有节点的绝对坐标, 最后通过单隐层BP神经网络优化定位结果。仿真结果表明, 该算法比MDS-MAP算法和MDS-MAP(P)算法的相对定位误差小, 应用于室内定位可以得到更高的定位精度。

In this paper, a BP neural network indoor location algorithm based on visible light communication (VLC) was proposed. Firstly, the relative coordinates of the nodes in the whole network were obtained by both the MDS-MAP algorithm and the least squares algorithm, and then the absolute coordinates of all nodes were obtained from the coordinates of the source nodes. Finally, the positioning result was optimized with the single-hidden layer BP neural network. Simulation results show that compared with MDS-MAP and MDS-MAP(P) algorithms, the proposed algorithm shows smaller relative positioning error, and can obtain higher accuracy in indoor positioning.