AODC: Automatic Offline Database Construction for Indoor Localization in a Hybrid UWB/Wi-Fi Environment

With the rapid development of mobile terminals and the ever-expanding deployment of wireless infrastructures, the requirement of location-based services (LBS) has permeated majority industries. However, fingerprint-based localization method has a bottleneck, namely, the offline database construction is time-consuming and labor-intensive, which hampers its implementation and adaption. Moreover, the mismatch of fingerprint due to various environmental interferences during the online localization phase is another critical issue to be addressed. In view of this, we consider the fingerprint-based indoor localization in a hybrid UWB/WiFi environment, which integrates UWB and Wi-Fi to speed up the construction of offline database, while maintaining a meter-level localization accuracy. Specifically, the system does not require manual labeling efforts for reference points (RP). Instead, we propose a heterogeneous data synchronization scheme (HDSS) to integrate the RSSI data obtained by the Wi-Fi device and the corresponding coordinates obtained by UWB device. Then, an automatic radio map generation scheme (ARMGS) is proposed, which can automatically generate the fingerprints profiles of RPs. Furthermore, for the online localization, unlike traditional approaches, which estimate locations only based on features of signal space, we proposed a dual-domain constraints localization algorithm (DCLA), which takes the physical space into consideration at the same time based on mean shift clustering algorithm. Finally, we implement the prototype of AODC system and carry out extensive real-world experiments. The experimental results validate the effectiveness of the proposed solutions.