Research On Key Technologies Of Crowdsourcing-based Indoor Localization System

For indoor location systems based on WLAN infrastructure, fingerprints matching is adopted by a large family of indoor localization schemes, where collecting fingerprints is inevitable but all consuming. Recent studies show that the crowdsourcing scheme could be utilized to perform fingerprints collecting with desirable performance price ratio. However, new challenges have raised. Firstly, the key problem is devices diversity problem, as crowdsourcing users employ diverse kinds of devices to collectfingerprints, which results in relative difference between pairwise of fingerprints. Secondly, in order to ensure crowdsourcing users’ initiative for participating in fingerprints collecting, novel techniques of auto constructing fingerprints’ map need to be proposed under crowdsouring model in practical indoor location system.For the problem of device diversity, since the fingerprints are collected from diverse devices, it is a great challenge to determine the reliability of the fingerprints and develop a corresponding localization scheme. This paper employs a suit of novel techniques to provide a total solution for this problem: 1) A novel fingerprint processing algorithm named Refined Relative Relationship(RE3) based on RSSI relative relationship is proposed to overcome the heterogeneous devices problem. 2) To resolve the RSSI series’ mismatching problem, a RSSI series processing algorithm named Dynamic RSSI Wrapping(DRW) is introduced to the RE3 algorithm. 3) In order to verify RE3’s extensive ability to adapt with existing localization algorithms, this paper combines RE3 with clustering-based localization algorithms. Thus, for the purpose of comparing clustering results derived from diverse devices, an algorithm based on optimal transform model is proposed to compare clusterings among different devices in practical indoor location system. Experimental results indicate that the proposed practical indoor localization system has competitive location accuracy symbiosis with crowdsourcing fingerprints collecting, which limits error difference among diverse devices below 1.2m and limits such difference below 1.4m with new added devices.For the problem of auto fingerprints collecting in crowdsourcing-based indoor location system, a novel approach based on crowd paths to solve this problem is presented, which collects and constructs automatically fingerprints database for anonymous buildings through common crowdsourcing customers. However, the accuracy degradation problem may be introduced as crowdsourcing customers are not professional trained and equipped. Therefore, this paper defines two concepts: fixed landmark and hint landmark, to rectify the fingerprint database in the practical system, in which common corridor crossing points serve as fixed landmark and cross point among different crowd paths serve as hint landmark. Based on training data collected from crowdsourcing users, short range approximation around fixed landmarks and fuzzy logic decision technology is applied for searching hint landmarks in crowd traces space. Besides, the particle filter algorithm is also introduced to smooth the sample points in crowd paths.We implemented the approach on off-the-shelf smartphones and evaluate the performance. Experimental results indicate that the approach can availably construct Wi-Fifingerprints database without reduce the localization accuracy.