Collaborative localization of mobile users with Bluetooth

Abstract

Location awareness is a key requirement for many pervasive applications. Collaborative localization can improve accuracy and coverage indoors and improve power consumption by duty-cycling GPS outdoors. We use Bluetooth for collaborative localization of mobile personal devices. Specifically, we embed information in Bluetooth device names to improve latency of information exchange between participating nodes. We identify and demonstrate on real hardware two problems in the Bluetooth stack that negatively impact localization accuracy: a) device name caching that introduces significant device-specific delays in transmitting information between nodes, and b) poor accuracy of time synchronization in modern mobile devices. Our solution is to append additional time information to the device name and track time offsets between nodes. We verify experimentally that this helps to both detect outliers and correct for time-synchronization errors and thus mitigate localization errors.