基于改进RSSI与CNN融合的抗多径干扰可见光室内定位方法OA

In view of the problems in traditional visible light indoor positioning,such as the received signal strength indicator(RSSI)being susceptible to multipath interference,relatively low positioning accuracy,and the difficulty of a single algorithm in adapting to complex indoor environments,a visible light indoor positioning method for anti-multipath interference based on the fusion of improved RSSI and a convolutional neural network(CNN)is constructed.A visible light indoor positioning system integrating lighting and positioning is built.Based on the Shadowing model,the RSSI propagation model is improved.A multipath correction factor and an environmental attenuation coefficient are introduced to weaken the impacts of multipath reflection and environmental disturbances.The RSSI data is pre-processed by the statistical mean method to enhance signal stability.A CNN positioning model is designed.Taking the pre-processed RSSI features and the incident-angle features of the photodiode(PD)array as inputs,deep features are extracted through convolutional and pooling layers,and the positioning coordinates are output via the fully-connected layer,realizing the nonlinear mapping between features and positions.Experiments are carried out in an indoor scenario of 4.5 m×3.6 m×3 m.The results show that the average positioning error of this fusion positioning method is 3.6 cm,and the maximum positioning error is 8.8 cm.Compared with the traditional RSSI positioning method and the single-CNN positioning method,the positioning accuracy is improved by 41.9%and 18.2%,respectively.In the scenario with enhanced multipath interference,the increase in positioning error is only 29.0%,and the anti-interference performance is significantly better than that of the comparative methods.This method does not require additional hardware and can reuse the existing light emitting diode(LED)system.It takes into account both positioning accuracy and engineering practicality,and is suitable for indoor positioning scenarios such as smart offices and intelligent warehouses.