3.1. Radio Propagation 35Figure 3.1 Schematics of the AROMA 3D-Raytracer. One the left side: ray-launching for variable antenna patterns and on the right side: material intersectionof rays.A RPE of around 3 dBm and a RPE-RMSE of 3. 5 dBm has been evaluated1bycomparing the simulation results of the model against 30 measurements at validationlocations. These values are comparable to the results of the PHOTON model. Butfor using only one initial measurement during training of the free parameters, thisis a very good result.3.1.2 3D-Raytracer ModelsOne of the earliest works on raytracer generated 3D radio propagation Models canbe found in[26]. The described technique was employed more recently in[8]. Thepresented raytracer simulates the physical effects of reflection and refraction. Dif-ferent antenna patterns are employed. The 3D-scene is imported from CAD outputfiles or synthesized by interpreting a 2D raster image containing a floor plan. Wallsor other anticipated solid structure from the 2D floor plan are extruded into theZ-axis with equal height. Different materials are assigned to the different parts ofthe 3D-Scene. Instead of applying some training procedure, the material parameterswere taken from literature like[30] and[20]. This is an unique approach comparedto the other presented systems.The evaluation of the system has been conducted on a 4-room scenario of around80 m2. 3 APs were placed and 3 averaged measurements at 42 locations are after-wards used for validation. A RMS-RPE of 2. 28 dBm is reported. The RMS-RPEdegrades to 2. 83 dBm, if the raytracer uses only the 2D mode. Comparing the per-formance over different sources of material parameters, the parameters in[30] haveled to the best result.A sophisticated raytracer driven radio propagation Model is given by the AROMAframework[9]. The implemented 3D-raytracer is able to simulate the physical ef-fects of reflection, refraction and diffraction. The diffraction model is based on1The reported value of 0. 65 dBm MSE has been corrected to 3. 5 dBm to be comparable to theother systems. Restoring the correct values was possible since a known number of 30 locationswere used.
Diplomarbeit
Indoor Localization of Mobile Devices Based on Wi-Fi Signals Using Raytracing Supported Algorithms
Einzelbild herunterladen
verfügbare Breiten