Applying Long Short-Term Memory (LSTM) Mechanisms for Fingerprinting Outdoor Positioning in Hybrid Networks

Recently, Location Based Services (LBSs) becomes an important technology to enhance the applicability of Internet-of-Things (IoT) to provide better services in wireless environments. The Global Positioning System (GPS) is not always optimal for urban and suburban areas to provide positioning services. Because GPS is easily affected by signal fluctuations and shadowing effects in a complex and dynamic environments, consumes much power and it is hardware dependent. In this paper, we present an accurate Fingerprinting Outdoor Positioning Scheme (FOPS) that contains Wi-Fi and Orthogonal Frequency Division Multiplexing (OFDM) signal values as a dataset using Long Short-Term Memory (LSTM) network approach. To select the most representative features from Wi-Fi data, we apply Linear Discriminant Analysis (LDA) techniques as preprocessing stage to optimize positioning services. The experimental results revealed that, the proposed system achieves positioning results with error no more than 1.7 m. Moreover, the positioning time improved due to the integrations of LDA and LSTM algorithms. Therefore, the proposed system provides a promising positioning services of the IoT devices in wireless environments.