Research On Indoor Positioning Technology Of Fusion Based On WiFi Signals And Magnetic Field

With the rapid economic development,various high-rise buildings have been built up.In addition,indoor space has been expanding,and indoor terrain has become more and more complicated.So the demand of indoor positioning also becomes more urgent.As the existing single positioning technology has disadvantages in positioning cost,positioning accuracy and positioning stability,the combination of multiple positioning technology has become the main direction of indoor positioning.Considering the strong-discernibility WiFi signals in the global area and the stable magnetic field,this paper studies the indoor positioning technology of fusion based on WiFi signals and magnetic field.The main research contents are as follows:First,indoor WiFi signals and magnetic field are analyzed.The characteristics of indoor WiFi signals and magnetic field are studied from the five aspects,such as the signal stability in time dimension,differences in spatial dimension,terminal direction,terminal type and height.The experimental results show that the WiFi signals have difference in spatial dimension and can be distinguished between different locations.However,WiFi signals are time-variant and affected by terminal direction,terminal type,and height.In contrast,magnetic field is relatively stable compared with WiFi signals in the same environment.But magnetic field has low discernibility and is also affected by terminal type.Second,WiFi fingerprint database is reconstructed based on compressive sensing.In this method,the variation of WiFi signals recollected at a few locations are used as observation signals.A measurement matrix is constructed through the Euclidean distance and the difference of WiFi signals between the locations.An orthogonal matching persuit algorithm is proposed to recover the WiFi signals at the non-resampled locations.The effectiveness of the reconstruction algorithm is verified from three aspects,such as the number of resampled locations,the spacing of resampled locations,and the terminal type.Third,indoor positioning algorithm based on Hidden Markov Model is studied.In the step detection,the dynamic threshold of acceleration signal is detected,and then steps exceeding the acceleration threshold are re-detected by combining the magnetic field.The error of the step-counting is reduced caused by the body shaking,and the robustness of the step-counting is improved under different walking speeds.For the low discernibility of magnetic field,the magnetic field is serialized so that the geomagnetic information is enriched,which can better distinguish different locations.In addition,the the serialized geomagneticsignals are matched by a cosine similarity to overcome the difference of geomagnetic signals on different terminals.Finally,Hidden Markov Model is proposed to combine the strong-discernibility WiFi signals with the stable magnetic field.In an initial probability distribution matrix,WiFi signals are used for the coarse positioning.The serialized geomagnetic signals are added to an observation probability matrix.A forward algorithm uses the three elements of the model to iterate constantly to get the pedestrian's position.Experimental results show that the proposed algorithm improves the positioning stability and positioning accuracy.