The disclosure is related to position estimation with Wi-Fi beacons.
Wireless devices attached to pets, children, or old people allow their loved ones to keep track of them via a mobile device application interface. The wireless devices estimate their position from global positioning system (GPS), cell tower, and/or Wi-Fi beacon reception.
When a wireless device detects one or more Wi-Fi beacons, it can estimate its position based on the locations of the beacons. However, the accuracy of the position estimate depends on how well the locations of the beacons are known.
One way to determine beacon locations is to physically survey them; i.e. visit each beacon site and record beacon identification and location information in a database. Such a survey, while potentially producing high quality beacon location information, is quite tedious. Furthermore, the vast majority of Wi-Fi beacons are not accessible in this manner.
Thus, what are needed are systems and methods for estimating Wi-Fi beacon locations without having to physically survey them.
A method for building a database of Wi-Fi beacon locations is based on estimating locations of three classes of Wi-Fi beacons: (a) beacons at surveyed locations, (b) beacons detected by receivers from known positions as estimated by GPS, and (c) beacons detected by receivers from unknown positions. Case (c) most often occurs when GPS signals are not available, such as when a receiver is indoors.
Even when GPS is not available, Wi-Fi beacon locations may be inferred. Inferred locations usually carry greater estimated location error than surveyed or GPS-derived locations, but uncertain location information is still much more useful than no information. The methods for building a database of Wi-Fi beacon locations described below include methods to keep track of estimated errors in surveyed, GPS-based, or inferred locations.
Whenever a Wi-Fi beacon has a location associated with it, that location may be used in a method to infer locations of other beacons. Multiple rounds of inference and location estimation may be used to build up a database of estimated locations, albeit with increased location error each round. The number of rounds of inference may be limited explicitly (e.g. either a certain number of rounds or until no new location inferences are produced) or implicitly (e.g. when inferred location error exceeds a threshold value).
In this disclosure “Wi-Fi beacon”, “WLAN access point”, “AP”, “wireless access point” and similar terms all have the same meaning, namely a radio beacon designed for wireless connection to a network, most commonly using IEEE 802.11 standards. “GPS” is used as a shorthand to refer to any global navigational satellite system (GNSS) including the NAVSTAR global positioning system, Beidou, Glonass, Galileo, and other satellite navigation systems.
When a receiver detects a set of Wi-Fi beacons it can estimate its own position based on the locations of the beacons.
A wireless device may estimate its own position using methods based on the idea that in order for a device to be able to detect a beacon, the device must be near the beacon. For example, if wireless device 105 detects beacons 110, 115, and 120 with similar signal intensity, then one may conclude that the wireless device is near the centroid of the locations of these beacons rather than being co-located with any one of them. Further, a weighted centroid may represent a more accurate location estimate than an unweighted centroid, particularly when weighting accounts for the uncertainty or error in each beacon's estimated location. In other words, if the location of beacon 110 is known more accurately than that of beacon 115, then there is likely to be less ambiguity in the estimated location of a wireless device detecting beacon 110 than one detecting beacon 115.
A wireless device can obtain a better position estimate the better the locations of the beacons it detects are known, and the better the uncertainties or errors in the beacons estimated locations are known. The method for building a database of Wi-Fi beacon locations described below includes methods for inferring the locations of beacons and determining uncertainty or error in beacon estimated locations.
A first category of beacons is those whose locations are known by survey. These are also referred to as “provisioned” beacons. A set of beacons installed in a commercial building may fall in this category if the location of each beacon is recorded when it is installed. However even provisioned beacons may not have perfect location estimates. Thus estimated error in their surveyed locations is greater than zero.
The first step of a method for building a database of Wi-Fi beacon locations is to load information for known provisioned Wi-Fi beacons into a database. This is shown in
Whenever a wireless device receives signals from a set of Wi-Fi beacons it may generate a measurement report.
In the table of measurement reports in
The next step of a method for building a database of Wi-Fi beacon locations is to estimate the locations of Wi-Fi beacons that are detected by a wireless device when GPS position information is available.
First, measurement reports having GPS position information (GPS FLAG set) are selected from the table of all measurement reports. These reports may be called “GPS measurement reports” or simply “GPS reports”. In
Next, a list of all beacons that appear in GPS reports is constructed. A list of all beacons that appear in GPS reports 1, 3, and 4, for example is the list of AP IDs: {1, 2, 3, 4, 5, 6, 7, 11, 12}.
Next, a set of steps is performed for each beacon in the list of AP IDs just mentioned. In other words, a set of steps will be performed for beacon 1, then the same set of steps will be performed for beacon 2, then beacon 3, etc. Here we use beacon 3 (AP ID 3) as an example.
For each beacon, first obtain positions and estimated errors from each measurement report that includes the beacon. Using beacon 3 as an example, inspection of
Suppose, for example, that reports 1 and 3 include the following data:
Next, a weighting factor is computed for each GPS position. The weighting factor, wi, for the position reported in the ith measurement report may be computed as:
Here {circumflex over (σ)}i is the estimated error of the ith position estimate, for example the radius of a circle having a 67% likelihood of containing the wireless device and centered on the ith GPS-based position estimate.
Note that the sum of all weighting factors for one beacon is one:
Σwi=1
In this example, w1=⅘ and w2=⅕.
Next, a weighted location estimate is computed by weighting each GPS-based position by its weighting factor. In this example, the weighted location estimate is:
Next, the estimated location error of the weighted location estimate is computed according to:
σm2=Σwi2·{circumflex over (σ)}i2
Here, {circumflex over (σ)}m is the estimated error of the weighted location estimate. In this example, the estimated error of the weighted location estimate is 8.9 m.
Finally, the weighted location estimate and location error estimate for the Wi-Fi beacon is stored in a database. In this example, the following information is stored:
The procedure just described is repeated for each beacon that appears in a GPS measurement report. In this way estimated location and estimated location error may be computed for any Wi-Fi beacon that is detected by a wireless device when that device is also able to simultaneously estimate its own position via GPS.
The next step of a method for building a database of Wi-Fi beacon locations is to estimate the locations of Wi-Fi beacons that are detected by a wireless device when GPS position estimates are not available.
The overall procedure involves first inferring Wi-Fi beacon locations and then second using inferred locations to estimate locations of other Wi-Fi beacons. This procedure involves an “infer process” illustrated in
The infer process is based on the idea that a detected beacon having an unknown location is likely to be near other, simultaneously-detected beacons having known locations. For example, measurement report #2 in
The infer process proceeds as follows. First, measurement reports not having GPS position information (GPS FLAG not set) are selected from the table of all measurement reports. These reports are called “non-GPS measurement reports” or simply “non-GPS reports”. In
Next, a set of steps is performed for each non-GPS report. First, for each non-GPS report, a list is constructed of all beacons that not only appear in the report, but also are already in the database. If no such beacons exist, then the process skips to the next (non-GPS) measurement report.
Next, the location estimate and location error estimate for each beacon in the list just constructed is retrieved from the database. These locations and location error estimates are now used to compute an inferred position estimate and an inferred position error estimate for the measurement report.
The procedure for computing an inferred position and error estimate is similar to the procedure outlined above for computing a Wi-Fi beacon location and error estimate given a set of GPS position and error estimates. Here it is beacon location and error estimates retrieved from the database that are inputs for computing weighting factors, weighted location estimates, and error estimates.
The result is that a measurement report that had no position estimate may now have an inferred position estimate and an error estimate for that inferred position. This information is used to update the table of measurement reports. The infer process loops over non-GPS reports until no new inferred positions are generated.
The locate process is similar to the process described above (i.e. STEP 2) for estimating Wi-Fi beacon locations based on GPS measurement reports, except that it uses inferred position estimates rather than GPS position estimates.
The locate process starts with construction of a list of all Wi-Fi beacons that appear in measurement reports that have an inferred location estimate and are not in the database. Next a set of steps is performed for each beacon in the list. The steps are the same as the steps described above for beacons in a list of beacons that appear in GPS reports. This time, however, the list that is processed is a list of beacons that have inferred locations rather than GPS locations. The result is an inferred location and an estimate of the error in the inferred location for each beacon in the list.
The infer/locate loop may proceed as long as inferred location estimates can be computed for measurement reports not having GPS position estimates. When no additional measurement reports are able to be assigned an inferred position estimate, then no additional beacon location estimates are computed.
The methods described above are useful for building a database of Wi-Fi beacon locations. The information contained in the database allows wireless devices to estimate their position based on detecting Wi-Fi access points even when GPS position estimates are not available.
The above description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the principles defined herein may be applied to other embodiments without departing from the scope of the disclosure. Thus, the disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Number | Name | Date | Kind |
---|---|---|---|
7769396 | Alizadeh-Shabdiz et al. | Aug 2010 | B2 |
8897803 | Pylappan et al. | Nov 2014 | B2 |
9037162 | Morgan et al. | May 2015 | B2 |
9100778 | Stogaitis et al. | Aug 2015 | B2 |
9119168 | Huang et al. | Aug 2015 | B2 |
20060095348 | Jones et al. | May 2006 | A1 |
20060095349 | Morgan et al. | May 2006 | A1 |
20060106850 | Morgan et al. | May 2006 | A1 |
20060200843 | Morgan et al. | Sep 2006 | A1 |
20060217131 | Alizadeh-Shabdiz et al. | Sep 2006 | A1 |
20060240840 | Morgan et al. | Oct 2006 | A1 |
20070004427 | Morgan et al. | Jan 2007 | A1 |
20070004428 | Morgan et al. | Jan 2007 | A1 |
20070167173 | Halcrow et al. | Jul 2007 | A1 |
20070202888 | Brachet et al. | Aug 2007 | A1 |
20070258408 | Alizadeh-Shabdiz et al. | Nov 2007 | A1 |
20070258409 | Alizadeh-Shabdiz et al. | Nov 2007 | A1 |
20070258420 | Alizadeh-Shabdiz et al. | Nov 2007 | A1 |
20070258421 | Alizadeh-Shabdiz et al. | Nov 2007 | A1 |
20070259624 | Alizadeh-Shabdiz et al. | Nov 2007 | A1 |
20080008117 | Alizadeh-Shabdiz | Jan 2008 | A1 |
20080008118 | Alizadeh-Shabdiz | Jan 2008 | A1 |
20080008119 | Alizadeh-Shabdiz | Jan 2008 | A1 |
20080008120 | Alizadeh-Shabdiz | Jan 2008 | A1 |
20080008121 | Alizadeh-Shabdiz | Jan 2008 | A1 |
20080108371 | Alizadeh-Shabdiz et al. | May 2008 | A1 |
20080132170 | Alizadeh-Shabdiz et al. | Jun 2008 | A1 |
20080139217 | Alizadeh-Shabdiz et al. | Jun 2008 | A1 |
20080176583 | Brachet et al. | Jul 2008 | A1 |
20080188239 | Carlson et al. | Aug 2008 | A1 |
20080188244 | Mazlum et al. | Aug 2008 | A1 |
20080188245 | Alles et al. | Aug 2008 | A1 |
20080248741 | Alizadeh-Shabdiz | Oct 2008 | A1 |
20080248808 | Alizadeh-Shabdiz | Oct 2008 | A1 |
20090075672 | Jones et al. | Mar 2009 | A1 |
20090149197 | Morgan et al. | Jun 2009 | A1 |
20090154371 | Alizadeh-Shabdiz et al. | Jun 2009 | A1 |
20090175189 | Alizadeh-Shabdiz et al. | Jul 2009 | A1 |
20090252138 | Alizadeh-Shabdiz | Oct 2009 | A1 |
20090303112 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090303113 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090303114 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090303115 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090303119 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090303120 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090303121 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090310585 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090312035 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20090312036 | Alizadeh-Shabdiz | Dec 2009 | A1 |
20100052983 | Alizadeh-Shabdiz | Mar 2010 | A1 |
20110012780 | Alizadeh-Shabdiz | Jan 2011 | A1 |
20110012784 | Alizadeh-Shabdiz | Jan 2011 | A1 |
20110021207 | Morgan et al. | Jan 2011 | A1 |
20110035420 | Alizadeh-Shabdiz et al. | Feb 2011 | A1 |
20110045840 | Alizadeh-Shabdiz et al. | Feb 2011 | A1 |
20110058495 | Alizadeh-Shabdiz et al. | Mar 2011 | A1 |
20110074626 | Alizadeh-Shabdiz et al. | Mar 2011 | A1 |
20110080317 | Alizadeh-Shabdiz et al. | Apr 2011 | A1 |
20110080318 | Alizadeh-Shabdiz et al. | Apr 2011 | A1 |
20110080841 | Alizadeh-Shabdiz et al. | Apr 2011 | A1 |
20110093443 | Alizadeh-Shabdiz et al. | Apr 2011 | A1 |
20110164522 | Alizadeh-Shabdiz et al. | Jul 2011 | A1 |
20110177831 | Huang | Jul 2011 | A1 |
20110235623 | Alizadeh-Shabdiz et al. | Sep 2011 | A1 |
20110287783 | Alizadeh-Shabdiz et al. | Nov 2011 | A1 |
20110298659 | Alizadeh-Shabdiz | Dec 2011 | A1 |
20110298660 | Alizadeh-Shabdiz | Dec 2011 | A1 |
20110298663 | Alizadeh-Shabdiz | Dec 2011 | A1 |
20110298664 | Alizadeh-Shabdiz | Dec 2011 | A1 |
20110306358 | Alizadeh-Shabdiz et al. | Dec 2011 | A1 |
20110306359 | Alizadeh-Shabdiz et al. | Dec 2011 | A1 |
20110306360 | Alizadeh-Shabdiz et al. | Dec 2011 | A1 |
20110306361 | Alizadeh-Shabdiz et al. | Dec 2011 | A1 |
20110317579 | Jones et al. | Dec 2011 | A1 |
20120007775 | Alizadeh-Shabdiz | Jan 2012 | A1 |
20120100872 | Alizadeh-Shabdiz et al. | Apr 2012 | A1 |
20120108260 | Alizadeh-Shabdiz | May 2012 | A1 |
20120112958 | Alizadeh-Shabdiz et al. | May 2012 | A1 |
20120178477 | Morgan et al. | Jul 2012 | A1 |
20120196621 | Alizadeh-Shabdiz et al. | Aug 2012 | A1 |
20120230313 | Alizadeh-Shabdiz | Sep 2012 | A1 |
20120280866 | Alizadeh-Shabdiz | Nov 2012 | A1 |
20120309420 | Morgan et al. | Dec 2012 | A1 |
20130072227 | Morgan et al. | Mar 2013 | A1 |
20130076569 | Alizadeh-Shabdiz | Mar 2013 | A1 |
20130184003 | Alizadeh-Shabdiz et al. | Jul 2013 | A1 |
20130217390 | Alizadeh-Shabdiz et al. | Aug 2013 | A1 |
20130241776 | Alizadeh-Shabdiz | Sep 2013 | A1 |
20130260797 | Jones et al. | Oct 2013 | A1 |
20130288708 | Alizadeh-Shabdiz | Oct 2013 | A1 |
20140009339 | Sutton et al. | Jan 2014 | A1 |
20140049424 | Alizadeh-Shabdiz | Feb 2014 | A1 |
20140080510 | Morgan et al. | Mar 2014 | A1 |
20140135033 | Alizadeh-Shabdiz | May 2014 | A1 |
20140160963 | Alizadeh-Shabdiz et al. | Jun 2014 | A1 |
20140187169 | Alizadeh-Shabdiz et al. | Jul 2014 | A1 |
Entry |
---|
Cheng, Y.C. et al. “Accuracy Characterization for Metropolitan-scale Wi-Fi Localization,” Intel Research, Intel Corporation, 13 pages, Jan. 2005. |
Koo, J. et al. “Localizing WiFi Access Points Using Signal Strength,” IEEE, 15(2), pp. 187-189, Feb. 2011. |
Koo, J. et al. “Unsupervised Locating of WiFi Access Points Using Smartphones,” IEEE, 42(6), pp. 1341-1353, Nov. 2012. |
Koo, J. et al. “Autonomous Construction of a WiFi Access Point Map Using Multidimensional Scaling,” Proceedings of the Pervasive Computing—9th International Conference, Pervasive 2011, vol. 6696, pp. 115-132, Jun. 2011. |
LaMarca, A. et al. “Self-Mapping in 802.11 Location Systems,” UbiComp 2005 Proceedings of the 7th International Conference on Ubiquitous Computing, vol. 3660, pp. 87-104, Sep. 2005. |
Sapiezynski, P. et al. “Opportunities and Challenges in Crowdsourced Wardriving,” Proceedings of the 2015 ACM Conference on Internet Measurement Conference, pp. 267-273, Oct. 2015. |