Analysis of abstract simulation via stochastic differential equation models
Yujing Wu, Weibo Gong, Don Towsley · 2004
We study a queue where the rapidly varying component of input traffic is treated as a noise. The queue length statistics are compared between the cases with and without noise smoothing. This relates to abstract simulation and traffic modeling. We found that the system load and traffic burstiness at large timescales have critical impacts on the evaluation errors caused by such smoothing.