Heavy-traffic asymptotics for the single-server queue with random order of service (Q2488203)

The paper investigates the GI/GI/1 queue with random order of service (ROS). It considers the behavior of the steady-state waiting time \(W^{\text{ROS}} \) when the system is in heavy traffic (\(\rho \to 1\)). It is known that, under the assumption of Poisson arrivals, there exists a scaling function \(\Delta (\rho )\) as \(\rho \to 1\) such that \(\Delta (\rho )W^{\text{ROS}} \mathop \to \limits^{\text{d}} YW^{\text{FCFS}} \). Here, \(W^{\text{FCFS}} \) is the corresponding heavy-traffic limit of the workload (which is equal to the waiting time in FCFS) and \(Y\) is an exponential variable with mean 1, which is independent of \(W^{\text{FCFS}} \). The paper presents an insightful proof of this statement, which does not need the assumption of Poisson arrivals.