Thesis (Diplom) 
Indoor Localization of Mobile Devices Based on Wi-Fi Signals Using Raytracing Supported Algorithms
Place and Date of Creation
Turn right 90°Turn left 90°
  
  
  
  
  
 
Download single image
 

6.1. Radio Propagation Model 77MaterialConcreteLightWallsDoorsGlassDoorIronDoorFascadeGlassWindowCupboardTableHardwareRailingNvertex334418225284801201432137236832848240Ntriangle584929427926841802144204654049272408 V in m3152317421212728121333avg. Thickness in m0.30.150.030.020.050.40.05n/an/an/an/aTable 6.1 Properties of the UMIC scene geometry with respect to the differentmaterial classes.All these materials are listed in table 6.1 with corresponding vertex and trianglecounts for the UMIC scene. The enclosed volume V for each material has only beenapproximated from a voxelized representation. Since the voxel-representation tendsto overestimate the volume at the boundaries of wall-like objects, the volume wasadapted with respect to the empirically determined average thickness of the sceneobjects for each material.6.1.1.2 AccesspointsAs the primary information source for the localization system, 22 APs were em-ployed in the evaluation. 9 of these APs are part of the EDUROAM infrastructure,that is provided as a service of the university and are therefore stationary in theexperimental setup. 7 APs are ASUS models and were placed to cover the floorsevenly. 4 APs of a different ASUS model were part of an experiment of anotherresearch group and are also assumed as stationary. The last two APs are LinksysWRT54G models, that were salvaged from the harware pool of the author. Theywere placed near the stairways since the other APs had somewhat neglected thisarea.As can be seen in figure 6.1, the coverage density of the APs is pretty high. Mostspace is covered by at least 6 APs. As is concluded later in the evaluation, loweringAP density below that threshold leads to much higher errors in the localizationalgorithms. Further steps to optimizing the setup can be conducted by relocatingthe APs with the target to maximize the average AP density. Other optimizationtargets for a guided relocalization could try to maximize the variety of the signals,as this is the primary information source for the localizer. These approaches canprobably be automated with the help of the raytracer.