On solving device diversity problem via fingerprint calibration and transformation for RSS-based indoor localization system

The device diversity problem is an important issue in indoor fingerprinting localization systems based on received signal strength (RSS). Since different mobile devices are usually equipped with different brands and types of radio transceivers, the RSS samples collected at the same location for the same access point (AP) may be much different. In this paper, we propose an inter-device calibration scheme together with a threshold-based AP selection for offline training fingerprint composition. Furthermore, we propose to use transformed fingerprints for online positioning to further reduce the negative impact of the device diversity problem. We conduct field measurements and experiments to examine the localization performance. Results validate the effectiveness of the proposed scheme in terms of lower localization errors, compared with the peer schemes.