With  the  growing need  of  mobility, efficient  indoor  navigation system  are
gaining more and more importance. Positioning systems are becoming  increasingly
vital to today’s pervasive  wireless technology. Even though  Global Positioning
System (GPS) based  localization is dominant  in outdoor navigation,  its signal
structure and signal levels are barely sufficient in indoor environment. Present
day  indoor navigation  systems sometimes  relay on  high sensitivity  (HS)  GPS
receivers  augmented with  dead reckoning  (DR) sensors  or mostly  on  wireless
technologies, such as  Wireless LAN (WLAN),  Infra Red (IR),  or Bluetooth, that
are existing  as part  of the  communication infrastructure.  The resulting cost
advantage compared  to pure  localization systems  has drawn  much attention  to
them. However, the accuracy and reliability in position estimates obtained using
such systems is largely erroneous for indoor navigation.

Even  though  Inertial  Measurement  Units  (IMU)  and  WLAN  infrastructure are
separately used  for indoor  positioning, not  much work  is done in integrating
both. The object of this paper is to integrate MEMS IMU measurements with a WLAN
received signal  strength (RSS)  based positioning  platform to  obtain reliable
position estimates, combining short-term accuracy of inertial sensors with  long
-term stability  of absolute  WLAN positioning.  Here, HSGPS  is replaced by the
WLAN positioning platform to update the position estimates provided by IMU.

In this paper we propose an architecture and localization algorithms to fuse the
WLAN position estimates with the IMU raw data using a simple two-step  procedure
and a more sophisticated one  using Kalman filters. The WLAN  position estimates
are periodically obtained by location fingerprinting using neural networks,  and
used to update the dead reckoning navigation of the IMU. Real-world test results
conducted in  an office  environment compare  RSS-based, DR-based  and the  new,
hybrid WLAN/IMU  positioning, and  show the  reduction of  position error of the
hybrid system in addition to increased reliability.