Publication

Abstract : Millimeter Wave (mmWave) channel study is important in 5G communication to assess novel technological solutions in realistic environments. Thus, it is vital to develop reliable channel models integrating the effects of mmWave atmospheric absorption, foliage loss, and the directionality of high gain antenna systems used in mmWave links. In this paper, a design is presented that combines mmWave channel modeling with machine learning for the efficient management of environment geometry and channel specification in a 5G urban microcell (UMi) street canyon (SC) outdoor channel. Accordingly, we first investigate the channel characteristics of mmWave line of sight (LOS) and non-line of sight (NLOS) directional outdoor links in various reference cases. The analysis is conducted for the backhaul and cellular access cases using a low complexity custom channel model based on ray tracing. Additional modeling components such as antenna directionality, oxygen absorption, and foliage loss modeling are also included to enhance accuracy and generality. It is observed that the estimated channel path loss is largely dependent on SC deployment parameters due to the antenna directionality combined with the small wavelength of mmWaves, and the use of correct values for them is essential to obtain optimal channel performance. Hence, we apply a metaheuristic algorithm called particle swarm optimization (PSO) for the optimal placement of position and deployment parameters of SC channel in a mmWave communication system to yield the best channel path loss.