Analyzing the quality of crowd sensed WiFi data

Abstract

The widely extended WLAN infrastruc- ture has often been used as geographic landmark to support localization applications and human mobility studies. In these applications a WiFi network made of static nodes seems to be always assumed. Here, evidence is presented to show that this is hardly ever the case. Several independently collected datasets were analyzed to show that dynamic, moving Access Points are often present and could have a significant negative impact in this kind of applications. Additionally other irregularities in these traces are also exposed. The possible impact of these irregularities is evaluated in one specific application, Proximity Maps. The node degree distribution of Proximity Maps is studied and the influence of the proposed solution on the degree distribution is analyzed. Finally some possible simple solutions to mitigate the problem are presented.