CRCLoc: A Crowdsourcing-Based Radio Map Construction Method for WiFi Fingerprinting Localization
Xiaoqian Du,Xuewen Liao,Minmin Liu,Zhenzhen Gao
DOI: https://doi.org/10.1109/jiot.2021.3135700
IF: 10.6
2021-01-01
IEEE Internet of Things Journal
Abstract:The WiFi-based fingerprint indoor-positioning system has attracted increasing interest from industry and academia, benefiting from the widespread deployment of the wireless local area network (WLAN) infrastructure. However, with the expansion of application scenarios, this system suffers from labor-intensive work for received signal strength (RSS) fingerprint construction. In this article, we propose a crowdsourcing-based radio map construction method and trajectory matching algorithm to solve the time-consuming preliminary fingerprint data collection in the offline phase, where the tedious collection work is replaced by the massive crowdsourcing sensor data. First, the latent information of the map is extracted by using some image processing methods, and all possible routes are obtained simultaneously with the depth-first traversal method. Then, considering the strict restrictions on walking in indoor environments, the crowdsourcing trajectories are produced by matching the result of pedestrian dead reckoning (PDR) with the candidate routes based on the Shape Context algorithm. Further, a crowdsourcing trajectory can be represented by uniformly distributed reference points based on step detection, and each point corresponds to a unique RSS of access points (APs). Since such a fingerprint construction method can be performed while smartphone holders are moving and not aware of their actual positions, it can be referred to the dynamic construction method of the radio map. The real scenario experiments reveal that the proposed solution can significantly reduce the time and manpower consumption to build the radio map. Moreover, compared with traditional schemes, our crowdsourcing-based radio map construction method for the WiFi fingerprinting localization (CRCLoc) system can improve the accuracy and robustness of the indoor-positioning results.