With the fast growing demand of location-based services in indoorenvironments, indoor positioning based on fingerprinting has attracted a lot ofinterest due to its high accuracy. In this paper, we present a novel deeplearning based indoor fingerprinting system using Channel State Information(CSI), which is termed DeepFi. Based on three hypotheses on CSI, the DeepFisystem architecture includes an off-line training phase and an on-linelocalization phase. In the off-line training phase, deep learning is utilizedto train all the weights of a deep network as fingerprints. Moreover, a greedylearning algorithm is used to train the weights layer-by-layer to reducecomplexity. In the on-line localization phase, we use a probabilistic methodbased on the radial basis function to obtain the estimated location.Experimental results are presented to confirm that DeepFi can effectivelyreduce location error compared with three existing methods in tworepresentative indoor environments.
展开▼
