Indoor Navigation Using Virtual Anchor Points

In a series of two papers, we concentrate on indoor location awareness from an unusual perspective. Instead of trying to extract coordinates from Wi-Fi signals, we first treat Wi-Fi time series as atomic objects and use Trajectory Computing for their analysis. This leads to a filter-free, efficient location-aware system predicting the next location from the Wi-Fi signal time series without much lag:

Starting from this application, we analyzed which locations inside these time series allow for high-quality reidentification. These spots are located in signal space, made unique enough and used in order to create event streams out of Wi-Fi streams. As you can see in the following video, these event streams are quite stable in the sense that there are not too many switches between different label assignments. However, the spatial location of these points is often not unique or local. But this is to be expected due to the complex signal propagation.

Still, we think that this approach can be used to augment and greatly simplify indoor location-aware systems based on time series. Instead of having a time series with as many dimensions as there are access points and lots of missing values, we get a sequence of labels for movement in space.

Transform Wi-Fi Signals into Time Series of Labels for Indoor LBS. Tweet It


Below, you find the slides for the presentation held at the European Navigation Conference in Helsinki this year.