S&M3022 Research Paper of Special Issue
Published: August 2, 2022
Indoor Fingerprinting Positioning System Using Deep Learning with Data Augmentation [PDF]
Luomeng Liu, Qianyue Zhao, Shoma Miki, Jumpei Tokunaga, and Hiroyuki Ebara
(Received April 1, 2022; Accepted May 27, 2022)
Keywords: indoor positioning, fingerprinting, residual network, data augmentation, deep learning
We propose an indoor positioning system based on deep learning and fingerprinting. On the mobile side, we designed an Android application with received signal strength information (RSSI) signal reading, database storage, and real-time online positioning module functions. In addition, we placed a trained neural network model on the built server to achieve real-time positioning using the developed Android application. The deep learning framework of this paper uses a residual network (ResNet) and a data augmentation technique called mean and uniform random numbers in the preparation of the dataset. By using this data augmentation method, we significantly reduced the collection time of the dataset and increased the test accuracy of the neural network from 20.4% before the augmentation to 97.5% after the augmentation.Corresponding author: Hiroyuki Ebara
This work is licensed under a Creative Commons Attribution 4.0 International License.
Cite this article
Luomeng Liu, Qianyue Zhao, Shoma Miki, Jumpei Tokunaga, and Hiroyuki Ebara, Indoor Fingerprinting Positioning System Using Deep Learning with Data Augmentation, Sens. Mater., Vol. 34, No. 8, 2022, p. 3047-3061.