Efficient Ray Tracing Simulation

Ray Tracing (RT) is a geometrical optics approach which determines most propagation paths that can propagate from transmitter (Tx) to receiver (Rx). It has been proved to be an accurate and versatile propagation prediction tool for both indoor and outdoor scenarios. As other deterministic channel models, RT requires exact knowledge of the geometrical and electromagnetic description of the relevant environment. Hence, the accuracy of RT comes at the cost of high computational complexity, which directly scales with the number of propagation paths considered. To date, the developed RT tool includes not only specular reflection, transmission through dielectric blocks and diffraction, but also diffuse scattering mechanisms. The large number of paths involved in RT makes the computation workload very heavy. In order to accelerate the execution of a current three dimensional (3–D) RT model, which is implemented in MATLAB, the code is optimized and time–consuming functions are converted to MEX functions by using MATLAB Coder. The speeding up effort in the thesis focuses on the reflection and diffuse scattering calculations. Compared with the original code, the simulation time of the revised Matlab code is significantly reduced. For example, about ninety percent time consumption is saved relative to the original RT for an 8–block scenario. The diffuse scattering components constitute a high proportion of all the propagation paths, which is closely related to the search for the diffuse scattering tiles. For reducing the simulation time of the original RT further, an efficient approach to generate diffuse scattering tiles is explored. In the original RT model, each surface is recursively divided until all segments fulfill the far-field condition. A substantial number of low-energy scattering paths might be produced due to the over segmentation problem. Moreover, the other approaches mentioned by relevant papers are usually specific to certain cases. We proposed an effective approach, which can be used as a general procedure to subdivide rough surfaces. Scattering tiles are generated on each surface with full consideration of the random characteristic of diffuse components. Besides, the proper tile size can be determined based on the system bandwidth, which makes this approach site independent and widely applicable in different scenarios. Another approach to calculate the tile size based on the far–field condition is also provided and proved to be efficient as well. Simulation results of the proposed algorithm are compared with the original RT model for the purpose of validation. Parameters such as power delay profile(PDP), delay spread(DS), angle spread(AS) are analyzed as the criteria for the performance assessments. With the proposed algorithm, lower computational complexity and less simulation time can be achieved without degradation of the prediction accuracy.