Machine Learning for WLAN Positioning

  • Teemu Roos | Helsinki Institute for Information Technology HIIT

The need for special-purpose indoor positioning systems arises from the failure of established technologies, such as GPS, in indoor scenarios. Recently, the interest in positioning based on WLAN networks has grown. We discuss the basics of WLAN positioning, focusing on machine learning approaches where signal strength measurements are associated to geographic coordinates by applying classification and regression techniques. We also present recent work on semi-supervised learning for WLAN positioning where the costly training phase is simplified by exploiting easily obtainable unlabeled signal strength measurements whose position needs not be recorded.

    • Portrait of Jeff Running

      Jeff Running

Watch Next