Diplomarbeit 
Indoor Localization of Mobile Devices Based on Wi-Fi Signals Using Raytracing Supported Algorithms
Entstehung
Einzelbild herunterladen
 

2.1. Radio Propagation Model 11Figure 2.3 Signal Strength Map for one AP from a raytracing generated radiopropagation model. This is an example of the output from the PHOTON raytracerfor an AP of the UMIC scene on a voxel size of 0. 2 mevaluation of these parameters works as follows if N is the number of PHOTONshitting a material:1.(1- ) N PHOTONs are absorbed, therefore N PHOTONs remain for trans-mission.2.(1- ) N PHOTONs are refracted and continue to travel at the same directionand the remaining N PHOTONs are reflected and travel in the invertedinput direction.After simulating the paths of all PHOTONs by applying the BRDF recursively,the paths are drawn into the 3D space as voxelized lines. These lines are furthersmoothed through an anti-aliasing step. The sum of the individual power valuesof all overlapping lines determines the resulting power value of a voxel thereforehandling multipath effects as in[11]. A final smoothing step is given by applying athree dimensional Gaussian filter.The raytracer has been developed for the simulation of radio waves emitted by GSMbase stations in outdoor intracity scenarios. Experiments have shown an averageerror between measured and predicted signal strength within 6- 8 dBm.The free parameters of the PHOTON raytracer can be estimated by OptimizationAlgorithms. The author of the raytracer has successfully employed the Levenberg-Marquardt algorithm. But in the context of this thesis, Genetic Algorithms haveshown lower errors in the range of 3- 5 dBm for the UMIC indoor scenario.Another interesting possibility of the PHOTON raytracer is the modelling of antennapatters by using Spherical Harmonics[22]. Spherical Harmonics can be compactly