针对室内可见光通信（visible light communication, VLC）与射频（radio frequency, RF）通信异构网络中易因瞬态链路变化而导致频繁切换的问题，提出了一种基于时空特性的VLC/RF网络决策机制。首先，在离线采样阶段，通过对室内网格点的空间位置、带宽、接收信号强度、时延与信噪比等特征数据，在控制器中对特征数据进行处理并转换成比例权重，构建数据库。然后，在实时决策阶段，终端将采集到的特征数据反馈给控制器，通过跟控制器数据库中的信息比对来确定用户终端位置，提取出数据库中当前位置的优先链路并连接。最后，利用链路决策模型实时推导出网络通信指数，通过决策函数对数据库中的链路比例权重与网络通信指数聚合计算，求解出最优链路，并使用驻留定时器提高通信质量。仿真结果表明，本文切换算法与传统的模糊逻辑切换算法和等效信噪比（signal-to-noise ratio, SNR）垂直切换决策算法相比，有效降低了平均切换次数、切换中断概率。本文算法可以实现室内异构网络的智能化管理，改善用户体验质量。

Aiming at the problem of frequent switching due to transient link changes in the heterogeneous network of indoor visible light communication (VLC) and radio frequency (RF) communication, a decision-making mechanism based on spatio-temporal characteristics for VLC/RF networks were proposed. Firstly, in the offline sampling stage, the database was constructed by processing and converting feature data such as spatial location, bandwidth, received signal strength, delay and signal-to-noise ratio of indoor grid points into proportional weights in the controller. Then, in the real-time decision-making phase, the terminal feeded the collected feature data back to the controller to determine the location of the user terminal by comparing it with the information in the database of the controller, and the prioritized link of the current location in the database was extracted and connected. Finally, the link decision model was used to derive the network communication index in real time, and the optimal link was solved by aggregating the proportional weights of the links in the database with the network communication index through the decision function, and the communication quality was improved using the resident timer. Simulation results show that the switching algorithm effectively reduces the average number of switching times and switching interruption probability compared with the traditional fuzzy logic switching algorithm and the vertical switching decision algorithm with equivalent signal-to-noise ratio (SNR). The algorithm can realize intelligent management of indoor heterogeneous networks and improve the quality of user experience.