DFOPS: Deep Learning-Based Fingerprinting Outdoor Positioning Scheme in Hybrid Networks

Many Internet-of-Things (IoT) services rely on location information. This article proposes a deep learning-based fingerprinting outdoor positioning scheme ( $D_{\mathrm{ FOPS}}$ ) for use in scalable environments. The proposed scheme is a hierarchical combination of the support vector machine (SVM) and long short-term memory (LSTM) algorithms. It was applied in a large-scale wireless environment with multiple wireless local area networks (WLANs) and cellular base stations. The results show that the positioning error of the proposed scheme is 42 cm, and the computation time is reduced by 63% compared with conventional methods. Thus, the proposed system can provide promising and reasonable support location-aware services for IoT devices in large-scale wireless environments.