AutLoc: Deep Autoencoder for Indoor Localization with RSS Fingerprinting

Abstract

Wi-Fi based indoor localization has attracted great interest due to its ubiquitous access in many indoor environ- ments. However, the accuracy is deteriorated by the complex indoor propagation environments, which result in variable received signal strength (RSS). In this paper, we propose to utilize an autoencoder to improve the accuracy of indoor localization by preprocessing the noisy RSS. The AutLoc system includes an offline training phase and an online localization phase. In the offline training phase, we train the deep autoencoder to denoise the measured data and then build the RSS fingerprints according to the trained weights. In the online localization phase, we adopt three machine learning algorithms, which are random forest regression, multi-player perceptron classification and multi-player perceptron regression, to estimate the location. Averaging over the results of three algorithms, we then obtain the final estimated location. Simulation results justify superiority of the proposed AutLoc system over previous indoor location schemes in vast scenarios.