Print Email Facebook Twitter Google Timeline Geolocation Accuracy Title Google Timeline Geolocation Accuracy Author Macarulla Rodriguez, Andrea Contributor Tiberius, C.C.J.M. (mentor) Van Bree, R. (mentor) Janssen, G. (mentor) Hansen, R. (mentor) Faculty Civil Engineering and Geosciences Department Geoscience and Remote Sensing Date 2017-06-01 Abstract Google Location History Timeline could be used in the future to track mobile devices of users with a Google account. The Department of Forensic Digital Technology in the Netherlands Forensic Institute might consider using it as available data for evidence in its investigations. A part of this research is to assess the accuracy of the locations given by Google Location History Timeline. Google informs that any registered mobile device was at a certain time at a certain position, and provides a measure of the accuracy. To study the veracity of the information provided by Google, a series of experiments were carried out. During these experiments the true position was recorded with a reference GPS device with a superior order of accuracy. Subsequently, the accuracy values given by Google were studied and analyzed based on various parameters, such as the configuration of mobile device connectivity, speed of movement, environment, traffic density and weather. The distance between Google provided position and actual position (determined with a more precise device) is computed and called Google Error. Then this error was compared with the Google provided accuracy to have a measure of Google data quality. Additionally, linear least squares multivariate models were developed with the purpose of calculating the precision that Google would provide a priori together with its positioning error. When studying the variability of the Google accuracy and Google error, in the experiments it was found that these variables are dependent on the configuration of the mobile device, the environment, and the means of transport, but weather and traffic have no influence on these variables. To quantify the performance of the values provided by Google, a Hit is defined as the observation in which the actual error committed by Google is less than the accuracy provided. The configuration that has the largest Hit rate is the GPS connection, with a 52% success. Then 3G and 2G go with 38% and 33%respectively. The WiFi connection only has a Hit rate of 7%. Regarding the means of transport, when the connection is 2G or 3G, the worst results are in Still with a Hit rate of 9% and the best in Car with 57%. For predicting values for Google accuracy and Google error, six multivariate linear models were defined. The model input variables were the distances and angles from the position of the device to the three nearest cell towers, and the categorical (non-numerical) variables of Environment and means of transport. The signal strength received by the device from the base stations were treated as possible input variables too, but not sufficient correlation was obtained, on top these models would not be useful to study future forensic cases, since these measures are not usually available. To evaluate the utility of a model, a Model Hit is defined when the actual observation is within the 95% confidence interval provided by the model. The model that shows the best results was the one that predicted the accuracy when the used network is 2G, with 76% of Model hits. The next one had only a 23% success (accuracy 3G). As a conclusion from the performed experiments the assurance of Google providing the correct position can not be given. The accuracy radius Google provides when using exclusively telephony networks (2G or 3G) is over bounding the actual position error only in about 35% of the experiments; in the other 65% of the experiments, the actual error is larger than the given accuracy radius. For an accuracy measure to be of practical meaning, the confidence level should be much larger, for instance 95%. Even when using WiFi and GPS, Google gives accurate locations but the accompanying accuracy measure is too optimistic (small radii) and hit rates are very low. The linear models developed in this thesis gave results which were not satisfactory enough yet. Further research in the parameters involved and a major collection of data is required. Subject Wilkinsonaccuracycasebase stationAndroidCellCell IDCell TowercoordinatesGNSSGoogle accountGoogle accuracyGoogle MapsGoogle TimelineGPSsignalpower interpolationleast squareslinear modellocation historymulti linear modelNFIpositionsignal strenghtWiFi To reference this document use: http://resolver.tudelft.nl/uuid:d8653d95-b2ec-48c5-9e28-0e69758d9053 Part of collection Student theses Document type master thesis Rights (c) 2017 A. Macarulla Rodriguez Files PDF report_20170518.pdf 5.37 MB Close viewer /islandora/object/uuid:d8653d95-b2ec-48c5-9e28-0e69758d9053/datastream/OBJ/view