58 4. DesignFigure 4.9 Localization result of a tracked path visualized with the VTK 3D-enginethat is used in the Fat Client. The path start in the center of the first floor and endssomewhat right on the ground floor. At the end of the path the localization errorsare up to 5 m.inspecting the various properties of the models during runtime. Starting with theinitialization phase, the Fat Client receives all needed configuration data for a scenefrom the server over HTTP. The configuration contains:1. 3D-geometry of the building in form of a.obj-file.2. Positions and associated information of all APs.3. The location of the measurement points and the collected measurements fromthe different devices.4. A corpus of location annotated measurements, also called tracked paths, forevaluation of the localizer performance.5. A list of the optimization runs containing material parameters trained on thecluster nodes.With an initialized system, the user gets the ability to select a raytracer configurationand the material parameters of an optimization run to request the simulation ofradio propagation models from the Server. The Server distributes these requestedjobs over the cluster nodes and receives all simulation results that are then furtherredistributed to the Fat Client. This makes the Fat Client independent of a localCUDA environment that is needed for running the GPU-driven raytracer.With available radio propagation models for each AP, the localization engine isthen used to evaluate and visualize localization result sequences sT1for positionannotated measurements from the collected corpus of tracked paths. New paths
Diplomarbeit
Indoor Localization of Mobile Devices Based on Wi-Fi Signals Using Raytracing Supported Algorithms
Einzelbild herunterladen
verfügbare Breiten