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：传播环境的统计特性与业务到达过程的统计特性已知会影响用户在5G/6G毫米波（mmWave）及亚太赫兹（sub-THz）系统中的性能。尽管前一个主题近来已受到相当多的关注，但关于流量统计特性的影响却知之甚少。在本研究中，我们刻画了会话到达过程中的相关性与可变性对5G/6G毫米波/亚太赫兹系统性能的影响。为此，我们运用随机几何与排队论工具对基站（BS）的服务过程以及毫米波/亚太赫兹无线部分的具体特性进行建模。所考虑的指标包括系统资源利用率和会话丢失概率。我们的结果表明，资源请求分布的归一化自相关函数（NACF）、变异系数（CoV）以及方差对所考察的参数有显著影响。在相同到达率下，高滞后1阶NACF和高CoV值可能导致系统脱离正常运行状态，使丢失概率和资源利用率产生高达一个数量级的变化。即使与无相关的泊松过程存在轻微偏离，也会使利用率降低10-20%，并使会话丢失概率增加数倍。无线与环境特性可能进一步增加资源请求分布的可变性并降低资源利用率。总体而言，使用普遍接受的泊松假设会导致严重低估5G/6G毫米波/亚太赫兹系统的实际性能。因此，对于此类系统的准确性能评估，流量到达统计与传播统计同等重要。