Accurate estimation of queuing delays is crucial for designing and optimizing communication networks, particularly in the context of Deterministic Networking (DetNet) scenarios. This study investigates the approximation of Internet queuing delays using an M/M/1 envelope model, which provides a simple methodology to find tight upper bounds of real delay percentiles. Real traffic statistics collected at large Internet Exchange Points (like Amsterdam and San Francisco) have been used to fit polynomial regression models for transforming packet queuing delays into the M/M/1 envelope models. We finally propose a methodology for providing delay percentiles in DetNet scenarios where tight latency guarantees need to be assured.

翻译：准确估计排队延迟对于设计和优化通信网络至关重要，特别是在确定性网络（DetNet）场景中。本研究探讨了使用M/M/1包络模型对互联网排队延迟进行近似的方法，该模型提供了一种获取实际延迟百分位数紧密上界的简单方法。我们利用在大型互联网交换点（如阿姆斯特丹和旧金山）收集的实际流量统计数据，拟合了多项式回归模型，用于将数据包排队延迟转换为M/M/1包络模型。最终，我们提出了一种在需要确保严格延迟保证的DetNet场景中提供延迟百分位数的方法论。