Heterogeneous Fingerprints Fusion Indoor Positioning Based On Truth Discovery

With the development of the Internet of Things,indoor positioning has become a hot area.High-precision positioning in complex indoor environments is still the key of indoor positioning.However,it is difficult to eliminate the influence of indoor signal reflection and refraction on positioning accuracy.Nowadays,heterogeneous fusion fingerprints formed by several kinds of indoor wireless signals or other sensor information can increase the diversity of location data features,thereby improving the accuracy and robustness of positioning.And the fusion positioning algorithms can also greatly improve the positioning performance.However,the problem of fusion positioning is that the increase of calculation amount affects the positioning real-time and cost.In addition,the essence of fingerprint-based fusion positioning is to obtain the most reliable positioning estimates from the conflict positioning results.From this perspective,the fusion problem can be solved by the idea of truth discovery.In this paper,the indoor fusion positioning algorithms are studied based on the two kinds of heterogeneous fingerprints represented by WiFi RSSI and geomagnetic field,and the idea of true discovery is also used to realize indoor fusion positioning.The main contents of this paper are as follows:Firstly,the indoor WiFi signal fingerprint and geomagnetic fingerprint characteristics,data processing operations,heterogeneous fingerprint fusion and database construction are described in this paper.Secondly,research on the fusion positioning algorithms based on MMSE criteria and local dynamic fingerprints fusion localization algorithms.Improvement method for the defects of local dynamic fingerprint fusion algorithm is proposed,which fully considers the relationship between the classifier algorithms,and adds the single-point fingerprint sample data set by adding the relevant location fingerprint to calculate the global fingerprint fusion weight matrix.The experimental results show that the global fingerprint fusion algorithm proposed in this paper can achieve better fusion effect.Finally,Location fingerprint based support set expectation maximization through the truth discovery indoor positioning algorithm is proposed in this paper,which proposes the concept of fingerprint quality.By constructing the expectation maximization model,the most reliable target position estimate in the support set is given while evaluating the fingerprint quality.And the algorithm is verified based on the single fingerprint data,heterogeneous fingerprint measured data and simulation data.The experiment results show that the algorithm proposed in this paper achieves high positioning accuracy with all the data tested.