基于分层自适应随机森林的无源RFID相对定位方案OA

For high-precision and low-cost relative localization of passive RFID tags in three-dimensional space,a relative positioning scheme based on the packet reception rate(PRR)measurement and the hierarchical self-adaptive random forest algorithm is proposed.As the beginning step,the PRRs of passive tags at observation points around the target area are collected to build the offline fingerprint database.Afterwards,two lightweight online models are obtained for matching the PRR measurement and the relative position,based on the hierarchical decision tree model and the further extended hierarchical self-adaptive random forest model,respectively.Meanwhile the hyper-parameters of the proposed models are optimized by the random grid search strategy.Practical experiments carried out in libraries and indoor corridors show that the accuracy can reach 98.04%in static environments and 91.1%in dynamic environments,outperfor-ming the existing other machine learning algorithms.