high performance computing on graphics processing units: hgpu.org

Radar producers, which are mostly in defense industry, need radar environment simulator to test their products during the development. Such a simulator helps them to be able to get rid of costly field tests. For developing a radar environment simulator, radio wave propagation should be modeled. However, this is a computationally expensive and time consuming process. Improving the performance of propagation modeling contributes to the radar development work. Ray tracing is one of the several electromagnetic wave propagation techniques. It enables calculation of total range, delay and power of radio waves on each point of the field. In this study, we have developed a radio wave propagation modeling application using a parallel implementation of the ray tracing method. Reflection, refraction and free space path loss properties of radio waves are implemented. Predefined atmosphere types that affect the refraction and surface types that affect the reflection are included for user selection. Moreover, the user has the chance of defining special atmosphere and surface types. Our application works on two-dimensional (2D) maps. It also has the ability of converting three-dimensional (3D) maps to 2D slices and working on them. We have developed and accelerated the application using GPU computing and parallel programming concepts. We have run the proposed method sequential and parallel on CPU and parallel on GPU. We have compared and analyzed time measurements of the application on different domains. We have achieved up to 18.14 speedup values between high specification CPU and GPU cards within the scope of this thesis.