Indoor wireless positioning method based on variability feature, system and device therefor
An indoor wireless positioning method includes receiving initial received signal strength indicator (RSSI) information from a reception unit; searching for a fingerprint matching measured initial RSSI information in a fingerprint database; determining a first location corresponding to the retrieved fingerprint to be an initial location of the reception unit; receiving additional RSSI information from the reception unit; extracting features of variability between the additional RSSI information and the initial RSSI information; searching for variability fingerprints matching the initial location and extracted RSSI variability features in a variability fingerprint database; and updating a second location to a current location of the reception unit when the second location corresponding to the retrieved variability fingerprint is different from the initial location, wherein the second location is a location in a candidate area within a preset distance range centered on the first location.
1 . An indoor wireless positioning method comprises:
receiving initial received signal strength indicator (RSSI) information from a reception unit;
searching for a fingerprint matching measured initial RSSI information in a fingerprint database;
determining a first location corresponding to the retrieved fingerprint to be an initial location of the reception unit;
receiving additional RSSI information from the reception unit;
extracting features of variability between the additional RSSI information and the initial RSSI information;
searching for variability fingerprints matching the initial location and extracted RSSI variability features in a variability fingerprint database; and
updating a second location to a current location of the reception unit when the second location corresponding to the retrieved variability fingerprint is different from the initial location,
wherein the second location is a location in a candidate area within a preset distance range centered on the first location.
2 . The method according to claim 1 , further comprising generating the fingerprint database,
wherein the generating of the fingerprint database includes:
virtually forming a plurality of vertical and horizontal grid lines on an indoor space;
obtaining RSSIs, which are transmitted from a plurality of access points (APs) disposed at different locations, for each intersection of the vertical and horizontal grid lines; and
constructing the fingerprint database by linking location information of each intersection with the RSSI obtained for each intersection.
3 . The method according to claim 2 , further comprising generating a variability fingerprint database,
wherein the generating of the variability fingerprint database includes:
collecting the RSSI transmitted from the plurality of APs for a predetermined period for each location at the intersection of the vertical and horizontal grid lines;
extracting variability features of the collected RSSIs for each intersection using a machine learning-based feature extraction algorithm; and
constructing the variability fingerprint database by linking the location information of each intersection and variability features extracted for each intersection.
4 . The method according to claim 3 , wherein the searching for the variability fingerprints in the variability fingerprint database includes:
extracting candidate variability fingerprints corresponding to the intersections included in a preset distance range from the initial location; and
searching for variability features that match the RSSI variability features among the candidate variability features.
5 . The method according to claim 3 , wherein the variability features are extracted based on at least one category among a maximum variation value, a minimum variation value, a variation range, and a variation speed of the RSSI information collected for each intersection.
6 . The method according to claim 3 , further comprising constructing a variability feature classification model through machine learning the variability features extracted for each intersection,
wherein the extracting of the variability features is to classify the variability features between the initial RSSI information and the additional RSSI information after inputting the initial RSSI information and the additional RSSI information to the variability feature classification model.
7 . The method according to claim 3 , wherein the plurality of vertical and horizontal grid lines are virtually formed in a dynamic manner based on at least one among floating population of the indoor space, structural features, and positioning time.
8 . The method according to claim 3 , wherein:
as an average floating population of the indoor space increases, the number of virtually formed vertical and horizontal grid lines increases and the spacing therebetween becomes narrower; and
as the average floating population of the indoor space decreases, the number of the plurality of virtually formed vertical and horizontal grid lines decreases and the spacing therebetween becomes wider.
9 . The method according to claim 3 , wherein, based on structural features:
as the variability of the indoor space is predicted to be higher, the number of virtually formed vertical and horizontal grid lines increases and the spacing therebetween becomes narrower; and
as the variability of the indoor space is predicted to be lower, the number of virtually formed vertical and horizontal grid lines decreases and the spacing therebetween becomes wider.
10 . The method according to claim 3 , wherein, based on a positioning time:
as the variability is predicted to be higher, the number of virtually formed vertical and horizontal grid lines increases and the spacing therebetween becomes narrower; and
as the variability is predicted to be lower, the number of virtually formed vertical and horizontal grid lines decreases and the spacing therebetween becomes wider.
11 . The method according to claim 1 , further comprising maintaining the initial location as the current location of the reception unit and recognizing the additional RSSI information as the initial RSSI information when the second location stored in correspondence with the retrieved variability fingerprint and the initial location are the same.