Improving crowd-sourced Wi-Fi localization systems using Bluetooth beacons

Crowd-sourced Wi-Fi-based localization systems utilize user input for RF scene analysis and map construction. Such systems reduce the deployment cost and privacy concerns that expert-based site survey systems can create. However, the main bottleneck of such crowd-sourcing localization systems is a bootstrapping stage, where lack of contributions by users results in no accuracy guarantee and frequent unnecessary prompting for users' input, even for explored areas. In this paper, we propose a crowd-sourcing localization system that uses both Wi-Fi scene analysis and Bluetooth beacons to address the insufficient contribution challenge. After prompting for user input, the mobile device not only submits Wi-Fi fingerprint to a map server, but also enables Bluetooth beacons to disseminate/share its location and fingerprint information to quickly populate the signal map. Then, subsequent user devices entering the area can discover the Bluetooth beacons and are able to instantly obtain room-level location information without causing unnecessary prompting to users. We implement our proposed system in the Linux OS and evaluate the prototype extensively through both experiments and simulation. Our evaluation results show that using Bluetooth beacons help to improve signal map growth, while maintaining reasonable localization accuracy.