Design Of Infrastructure-free WiFi Indoor Localization

With the development of modern life, localization has become anecessity in every field. For outdoor localization, GPS (GlobalPositioning System) plays a leading role. But for indoor localization,GPS is not a good choice because of the geographic environment. With thedevelopment of wireless network, WiFi based localization is becoming moreand more popular for indoor environment.In traditional positioning systems, they depend on additionalinfrastructure, which is expensive and needs the support of other softwareto provide positioning services. Signal-Propagation-Model localizationgets rid of the hardware obstacle to some extent, but the result is notsatisfactory due to the complexity of the signal propagation. RSS-basedWiFi location is easier but requires a great amount of time and labor costto build the RSS database. Although some RSS-based systems have cut downthe work of database construction by GPS and sensors, it needs too muchuser participation and thus becomes hard to operate in the real world.To overcome the above weakness, this paper proposes a RSS-basedlocalization algorithm without infrastructure support. It is based onlyon WiFi and does not require any additional infrastructure. It will notinvolve any user participation and is easy to use. Instead of static signalcollection, it utilizes data mining to dynamically build the database.After matching the signal data and geographic space by bipartite graph match, it can support high accurate localization with a little manual help.Moreover, it has a mechanism of cleaning data, which makes it adaptiveto the environment. The experiment result shows that the degree ofaccuracy can reach up to90%and the system is hardly affected by theenvironment. Of course, the algorithm has shortness. It has to depend onsome manual help to ensure the matching correctness between virtual roomand geographic room and can not work automatically.