Inscrit le: 22 Avr 2016
|Posté le: Jeu 6 Juil - 13:39 (2017) Sujet du message: Mobile Sensing GPS Localization WiFi Mapping Applications
Localization is a fundamental feature of mobile systems such as smartphones, airplanes or self-driving cars. Albeit current smartphones include GPS receivers, accelerometers, and gyroscopes, localization is still difficult in certain environments such as indoors. First we show how WiFi signal strength measurements and motion data recorded with smartphones can be used to create accurate signal strength maps. The maps are created from data collected by a user walking around with a smartphone in the trouser pocket. From this data, we can accurately observe how the person moves by linking sensor measurements to how people walk. Exploiting the signal strength distributions recorded along the way, we show how the error that aggregates from inaccuracies in the motion estimation can be reduced. The signal strength maps are useful for localization indoors when GPS is unavailable. We introduce a GPS receiver design which allows for aggressive duty cycling. This means that the RF front end only has to be turned on for a very short time to collect enough data for localization. From the short signal recordings we compute accurate position fixes efficiently even if the uncertainty on position and time are larger than what can be expected from synchronization and localization using cell towers. The receiver design can be tailored for different applications. For example, short signal recordings can be stored without computing a position right away. This allows for low power and long term tracking as well as instant position annotation for photos or other data. Alternatively, the short signal duration allows for fast initial position estimation for connected devices. Furthermore, we discuss how motion data can be used to infer personal data such as pins or passwords. Our large scale user study reveals that, in uncontrolled environments, touch input on Smartphone screens can be inferred from the same devices' motion data.
bound: 138 pages
publisher: CreateSpace Independent Publishing Platform; 1 edition (April 4, 2017)
isbn: 1545158010, 978-1545158012,
weight: 9.3 ounces (