Research On Automatic Update Algorithm Of Wi-Fi Fingerprint Map For Indoor Location

With the acceleration of social modernization,human beings spend more than 80% of their time in the indoor environment,and the demand for location-based services is increasing,leading the indoor positioning market to be booming.In the past two decades,various indoor positioning technologies have emerged one after another.With the large-scale deployment of wireless network and the rapid development of smart terminal,positioning method based on Wi-Fi location fingerprint have becoming the first choice for current indoor positioning service.In the location fingerprint positioning method,the “fidelity” of the fingerprint map is closely related to the positional accuracy.In the long-term operation process,when the location of the AP in the indoor environment has changed,the fingerprint map can't truly reflect the distribution of signal strength in the indoor environment,and therefore can't provide satisfactory positional accuracy.A direct and effective fingerprint map calibration method is to update the initial fingerprint map.In the fingerprint map update work,first,identifying the AP whose location has changed.Second,collecting the user's motion information and RSS vector to generate a motion trajectory that containing the RSS data.Third,locating the position coordinates of multiple sampling points on the trajectory in the fingerprint map.Finally,the RSS vector of the sampling point on the track is used as the fingerprint of the corresponding position in the fingerprint map.However,the acceleration data under the user's fast motion contains more noise,and the existing pedometer algorithm has' t sufficient ability to eliminate noise,and it is easy to regard noise as a normal step.At the same time,users have a variety of modes for carrying smartphones during exercise,and the acceleration data on the three coordinate axes in different modes will have obvious differences,so it is necessary to select the user acceleration data on the appropriate coordinate axis.In addition,the time between two adjacent RSS samplings needs to meet the minimum sampling time threshold,otherwise the RSS values of some APs at the sampling point may actually be outdated data.In response to these problems,this paper proposes a step-counting algorithm for detecting peaks and a fingerprint updating algorithm with Altered APs(FUAA).Specifically,the main research contents of this article are as follows:(1)In order to reduce the error caused by the user's fast motion on the step count and to support the four modes of the user carrying the smart phone in real time,this paper proposes a step-counting algorithm with peak and trough detection to distinguish modes.For the mode change of the smart phone carried by the user,the pedometer algorithm uses a finite state machine to perform mode conversion.The pedometer algorithm first detects the mode change of the user carrying the smartphone from the angular velocity data,then recognizes the four modes of the current user carrying the smartphone according to the components of the acceleration of gravity on the three coordinate axes,and can select the appropriate user acceleration data on the axis in different modes.Then uses the sliding window to detect the possible positions of the peaks and troughs,and use the peak and trough detection step-counting algorithm to eliminate all false peaks and false troughs,obtaining the true peaks and troughs,and completing the step counting of the user's movement.Experiments have proved that the step-counting algorithm can accurately count the steps of the user's movement and updating the fingerprint map automatically.(2)In order to reduce the positioning error caused by the change of AP,the fingerprint map can't truly reflect the distribution of signal strength in the current positioning area,we use the idea of group intelligence to collect the movement data and fingerprint information of ordinary users in daily life.Proposing a fingerprint map automatic update algorithm FUAA which can adapt to AP location changes.For APs whose locations have changed,FUAA performs RSS subset sampling,acquiring multiple RSS subsets from an RSS vector collected in real time,and using the location change AP recognition algorithm based on DBSCAN clustering for identification.Subsequently,FUAA uses the anti-fingerprint misalignment RSS collection algorithm to collect the RSS vectors of multiple sampling points on the user's motion trajectory.So the collected RSS data is the latest,and then uses the trajectory matching algorithm to find all the sampling points on the trajectory in the fingerprint map.The RSS vector difference is a set of reference points.Finally,FUAA uses the inverse distance-weighted interpolation algorithm to estimate the RSS value of the AP who has changed at the remaining reference point positions in the fingerprint map to complete the update of the fingerprint map.Experiments show that FUAA can effectively cope with AP changes,reducing the positioning error caused by the deviation of the fingerprint map from the real environment,and laying a foundation for the fingerprint map to provide positioning services in the long-term operation.