Improving Ray Tracing Based Radio Propagation Model Performance Using Spatial Acceleration Structures

Ray tracing based propagation models are suited to simulate wireless networks in complex environments and can stand in for physical measurements if they are inaccessible. However, due to the associated computational cost, engineers have to trade off the size of modelled scenarios, level of detail and time resolution to keep simulation times feasible. We present a Monte Carlo ray tracing model that is implemented entirely on the GPU. State of the art spatial acceleration structures are used to optimize the underlying ray tracing routines. The model is verified and the implementations performance is measured in five demo scenes. We can report interactive frame rates in scenes with up to 7M triangles, 2M rays and multiple bounces.