Parallel GPU Processing for Fast Radio Signal Propagation Computation in GRASS-RaPlaT

Radio propagation simulation tools are important for prediction and verification of the radio signal coverage by individual transmitters or transmitter networks such as mobile phone cellular networks. In the case of a large geographic area with a relative high resolution, the simulation can become computationally demanding, taking a considerable amount of time to accomplish. Parallel processing can be used to speed up the computation and shorten the response time. We used the GPGPU (General-Purpose Computing on Graphics Processing Units) approach for the open source GRASS-RaPlaT radio propagation tool. By using OpenCL, we modified the existing radio propagation model modules for GPU (Graphics Processing Unit) execution, achieving multiple times processing speedup for computationally intensive modules. In the article, we present our GPU parallelization approach and analyze the results and conditions that must be fulfilled to successfully employ GPU computation and achieve a considerable computation speedup. Key-Words: Radio propagation simulation, GRASS, RaPlaT, parallel processing, GPU computing, OpenCL