Online co-localization in indoor wireless networks by dimension reduction

This paper addresses the problem of recovering the locations of both mobile devices and access points from radio signals that come in a stream manner, a problem which we call online co-localization, by exploiting both labeled and unlabeled data from mobile devices and access points. Many tracking systems function in two phases: an offline training phase and an online localization phase. In the training phase, models are built from a batch of data that are collected offline. Many of them can not cope with a dynamic environment in which calibration data may come sequentially. In such case, these systems may gradually become inaccurate without a manually costly re-calibration. To solve this problem, we proposed an online co-localization method that can deal with labeled and unlabeled data stream based on semi-supervised manifold-learning techniques. Experiments conducted in wireless local area networks show that we can achieve high accuracy with less calibration effort as compared to several previous systems. Furthermore, our method can deal with online stream data relatively faster than its two-phase counterpart.