We analyze the asymptotics of waiting time distributions in the two-class accumulating priority queue with general service times. The accumulating priority queue was suggested by Kleinrock in the 60s-he coined it time-dependent priority-to diversify waiting time objectives of different classes in a paramaterized way. It also avoids the typical starvation problem of regular priority queues. All customers build up priority linearly while waiting in the queue but at a class-dependent rate. At a service opportunity epoch, the customer with highest priority present is served. Stanford and colleagues recently calculated the Laplace-Stieltjes Transform (LST) of the waiting time distributions of the different classes, but only invert these LSTs numerically. In this paper, we analytically calculate the asymptotics of the corresponding distributions from these LSTs. We show that different singularities of the LST can play a role in the asymptotics, depending on the magnitude of service differentiation between both classes.
机构:
Scuola Super Sant Anna, TeCIP Inst, Via G Moruzzi 1, I-56124 Pisa, PI, Italy
Scuola Super Sant Anna, Dept Excellence Robot & AI, Via G Moruzzi 1, I-56124 Pisa, PI, ItalyScuola Super Sant Anna, TeCIP Inst, Via G Moruzzi 1, I-56124 Pisa, PI, Italy
Casini, Daniel
Pazzaglia, Paolo
论文数: 0引用数: 0
h-index: 0
机构:
Saarland Univ, Saarland Informat Campus E1-3, D-66123 Saarbrucken, GermanyScuola Super Sant Anna, TeCIP Inst, Via G Moruzzi 1, I-56124 Pisa, PI, Italy
Pazzaglia, Paolo
Biondi, Alessandro
论文数: 0引用数: 0
h-index: 0
机构:
Scuola Super Sant Anna, TeCIP Inst, Via G Moruzzi 1, I-56124 Pisa, PI, Italy
Scuola Super Sant Anna, Dept Excellence Robot & AI, Via G Moruzzi 1, I-56124 Pisa, PI, ItalyScuola Super Sant Anna, TeCIP Inst, Via G Moruzzi 1, I-56124 Pisa, PI, Italy