The statistical characteristics of the propagation environment and traffic arrival process are known to affect the user performance in 5G/6G millimeter wave (mmWave) and subterahertz (sub-THz) systems. While the former topic has received considerable attention recently, little is known about the impact of traffic statistics. In this study, we characterize the effects of correlation and variability in the session arrival process on the performance of 5G/6G mmWave/sub-THz systems. To this end, we use the tools of stochastic geometry and queuing theory to model the service process at base stations (BS) and specifics of the mmWave/sub-THz radio part. The metrics considered include the system resource utilization and session loss probability. Our results show that the normalized autocorrelation function (NACF), coefficient of variation (CoV), and variance of the resource request distribution have a significant impact on the considered parameters. For the same arrival rate, high values of lag-1 NACF and CoV may lead the system out of the operational regime, affecting the loss probability and resource utilization by up to an order of magnitude. Even a slight deviation from the uncorrelated Poisson process decreases the utilization by 10-20% and increases the session loss probability multiple times. Radio and environmental characteristics may further increase the variability in resource request distribution and decrease resource utilization. In general, the use of the commonly accepted Poisson assumption leads to a severe underestimation of the actual performance of 5G/6G mmWave/sub-THz systems. Therefore, both traffic arrival and propagation statistics are equally important for accurate performance assessment of such systems.

翻译：暂无翻译