Perfect sampling of Jackson queueing networks
نویسندگان
چکیده
We consider open Jackson networks with losses with mixed finite and infinite queues and analyze the efficiency of sampling from their exact stationary distribution. We show that perfect sampling is possible, although the underlying Markov chain may have an infinite state space. The main idea is to use a Jackson network with infinite buffers (that has a product form stationary distribution) to bound the number of initial conditions to be considered in the coupling from the past scheme. We also provide bounds on the sampling time of this new perfect sampling algorithm for acyclic or hyperstable networks. These bounds show that the new algorithm is considerably more efficient than existing perfect samplers even in the case where all queues are finite. We illustrate this efficiency through numerical experiments. We also extend our approach to non-monotone networks such as queueing networks with negative customers. Key-words: Perfect Simulation, Markov Chain, Jackson networks, bounding process ∗ Inria, Département d’Informatique de l’ENS (DI ENS) Paris, France † Inria, Univ. Grenoble Alpes, France ‡ Université Joseph Fourier, Univ. Grenoble Alpes, France ha l-0 08 51 33 1, v er si on 1 13 A ug 2 01 3 Simulation parfaite de réseaux de Jackson Résumé : On considère les réseaux de Jackson avec perte comportant des files finies et infinies, et l’on s’intéresse à l’efficacité des techniques d’échantillonnage de leur distribution stationnaire exacte. Nous démontrons que la simulation parfaite est possible même si la chaîne de Markov sous-jacente a un espace d’états potentiellement infini. L’idée principale est d’utiliser un réseau de Jackson aux files infinies (qui admet une distribution de forme-produit) pour borner les conditions initiales à considérer dans l’algorithme de simulation parfaite. Nous donnons également des bornes sur le temps d’échantillonnage de ce nouvel algorithme dans le cas des réseaux acycliques, ainsi que pour des réseaux hyperstables. Ces bornes prouvent que le nouvel algorithme est considérablement plus efficace que les échantillonneurs parfaits acuels, même dans le cas où toutes les files sont finies. Nous illustrons cette efficacité par des expériences numériques. Enfin, nous généralisons notre approche au cas des réseaux non-monotones comme les réseaux aux clients négatifs. Mots-clés : Simulation parfaite, réseaux de Jackson, chaînes de Markov, processus bornant ha l-0 08 51 33 1, v er si on 1 13 A ug 2 01 3 Perfect Sampling of Jackson Queueing Networks 3
منابع مشابه
2005: Approximate/Perfect Samplers for Closed Jackson Networks
In this paper, we propose two samplers for the productform solution of basic queueing networks, closed Jackson networks with multiple servers. Our approach is sampling via Markov chain, but it is NOT a simulation of behavior of customers in queueing networks. We propose two of new ergodic Markov chains both of which have a unique stationary distribution that is the product form solution of clos...
متن کاملMATHEMATICAL ENGINEERING TECHNICAL REPORTS Polynomial-time Randomized Approximation and Perfect Sampler for Closed Jackson Networks with Single Servers
In this paper, we propose the first fully polynomial-time randomized approximation scheme (FPRAS) for basic queueing networks, closed Jackson networks with single servers. Our algorithm is based on MCMC (Markov chain Monte Carlo) method. Thus, our scheme returns an approximate solution, of which the size of error satisfies a given error rate. We propose two Markov chains, one is for approximate...
متن کاملPolynomial-time Perfect Sampler for Closed Jackson Networks with Single Servers
In this paper, we propose a sampler for the product-form solution of basic queueing networks, closed Jackson networks with single servers. Our approach is sampling via Markov chain, but it is NOT a simulation of behavior of customers in queueing networks. We propose a new ergodic Markov chain whose unique stationary distribution is the product form solution of a closed Jackson Network, thus we ...
متن کاملImportance sampling for Jackson networks
Rare event simulation in the context of queueing networks has been an active area of research for more than two decades. A commonly used technique to increase the efficiency of Monte Carlo simulation is importance sampling. However, there are few rigorous results on the design of efficient or asymptotically optimal importance sampling schemes for queueing networks. Using a recently developed ga...
متن کاملAdaptive Importance Sampling Simulation of Queueing Networks
In this paper, a method is presented for the efficient estimation of rare-event (overflow) probabilities in Jackson queueing networks using importance sampling. The method differs in two ways from methods discussed in most earlier literature: the change of measure is state-dependent, i.e., it is a function of the content of the buffers, and the change of measure is determined using a cross-entr...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Queueing Syst.
دوره 80 شماره
صفحات -
تاریخ انتشار 2015