Augmented continuous wavelet transform features for deep learning-based indoor localization using WiFi RSSI Data
MetadataShow full item record
Localization in indoor environments is currently one of the challenges in navigation research. The conventional global positioning system (GPS) is affected by weak signal strengths due to high levels of signal interference and fading in indoor environments. Therefore, new positioning solutions tailored for indoor environments need to be developed. In this paper, we propose a deep learning approach for indoor localization. However, the performance of a deep learning system depends on the quality of the feature representation. This paper introduces two novel feature set extractions based on the continuous wavelet transforms (CWT) of the received signal strength indicators' (RSSI) data. The two novel CWT feature sets contain augmented data generated from the CWT of RSSI data with additive white Gaussian noise. The first feature set is an image feature set while the second feature set is a numerical feature set composed of the power spectral densities (PSD) of the CWT that is dimensionally equalized using the principal component analysis (PCA). These proposed images and numerical data feature set were both evaluated using the CNN and ANN model respectively with the goal of identifying the room that the human subject is in and estimating the precision location of a human subject in an indoor environment. Extensive experiments were conducted to generate the proposed augmented CWT feature set and numerical CWT PSD feature set using two analyzing functions namely the Morlet and Morse. The feature sets were both tested to predict a room and the precise position a human subject is at. For validation purposes, the two proposed feature sets were validated against each other and other existing feature set formulations. The accuracy, precision and recall results show that the proposed feature sets perform better than conventional feature sets used to validate the study. Similarly, the mean displacement error generated by the proposed feature set predictions is less than that of the conventional feature sets used in indoor localization. More particularly, the proposed augmented CWT-image feature set out performs the augmented CWT-PSD numerical feature set. The results also show that the Morse-based feature sets trained with CNN produce the best indoor positioning results compared to all Morlet and ANN-based feature set formulations.