Fluctuations around the mean-field for a large scale Erlang Loss system under SQ(d) load balancing
T. Vasantam, и R. Mazumdar. 31th International Teletraffic Congress (ITC 31), Budapest, Hungary, (2019)
Аннотация
In this paper, we study the fluctuations of the transient and stationary empirical distributions around the mean-field for a large scale multi-server Erlang Loss system that has $N$ servers. Jobs arrive according to a Poisson process with rate $Nłambda$ and each incoming job is dispatched by a central job dispatcher to the server with the minimum occupancy among $d$ randomly chosen servers with ties broken uniformly at random. Previous works have studied the mean-field limit of this model and characterized the asymptotic behavior of the system when $N\toınfty$. In this paper, we focus on quantifying the resulting error when we approximate the transient and stationary behavior of the system when $N$ is large by the mean-field of the system. We obtain functional central limit theorems (FCLTs) by studying the limit of a suitably scaled fluctuation process of the stochastic empirical process of the model with index $N$ around the mean-field limit when $N\toınfty$. We show that for both the transient and stationary regimes, the limiting process is characterized by an Ornstein-Uhlenbeck (OU) process. We also show that the interchange of limits $łim_N\toınftyłim_t\toınfty=łim_t\toınftyłim_N\toınfty$ is valid under the CLT scaling. Finally, we exploit the FCLT to show that the gap between the exact average blocking probability of a job in the system with the number of servers $N$ and the limiting average blocking probability which is a function of the fixed-point of the mean-field, is of the order $o(N^-\frac12)$ and thus establish the accuracy of the mean-field approximation for finite $N$.
%0 Conference Paper
%1 vas19ITC31
%A Vasantam, Thirupathaiah
%A Mazumdar, Ravi R.
%B 31th International Teletraffic Congress (ITC 31)
%C Budapest, Hungary
%D 2019
%K itc itc31
%T Fluctuations around the mean-field for a large scale Erlang Loss system under SQ(d) load balancing
%U https://gitlab2.informatik.uni-wuerzburg.de/itc-conference/itc-conference-public/-/raw/master/itc31/vas19ITC31.pdf?inline=true
%X In this paper, we study the fluctuations of the transient and stationary empirical distributions around the mean-field for a large scale multi-server Erlang Loss system that has $N$ servers. Jobs arrive according to a Poisson process with rate $Nłambda$ and each incoming job is dispatched by a central job dispatcher to the server with the minimum occupancy among $d$ randomly chosen servers with ties broken uniformly at random. Previous works have studied the mean-field limit of this model and characterized the asymptotic behavior of the system when $N\toınfty$. In this paper, we focus on quantifying the resulting error when we approximate the transient and stationary behavior of the system when $N$ is large by the mean-field of the system. We obtain functional central limit theorems (FCLTs) by studying the limit of a suitably scaled fluctuation process of the stochastic empirical process of the model with index $N$ around the mean-field limit when $N\toınfty$. We show that for both the transient and stationary regimes, the limiting process is characterized by an Ornstein-Uhlenbeck (OU) process. We also show that the interchange of limits $łim_N\toınftyłim_t\toınfty=łim_t\toınftyłim_N\toınfty$ is valid under the CLT scaling. Finally, we exploit the FCLT to show that the gap between the exact average blocking probability of a job in the system with the number of servers $N$ and the limiting average blocking probability which is a function of the fixed-point of the mean-field, is of the order $o(N^-\frac12)$ and thus establish the accuracy of the mean-field approximation for finite $N$.
@inproceedings{vas19ITC31,
abstract = {In this paper, we study the fluctuations of the transient and stationary empirical distributions around the mean-field for a large scale multi-server Erlang Loss system that has $N$ servers. Jobs arrive according to a Poisson process with rate $Nłambda$ and each incoming job is dispatched by a central job dispatcher to the server with the minimum occupancy among $d$ randomly chosen servers with ties broken uniformly at random. Previous works have studied the mean-field limit of this model and characterized the asymptotic behavior of the system when $N\toınfty$. In this paper, we focus on quantifying the resulting error when we approximate the transient and stationary behavior of the system when $N$ is large by the mean-field of the system. We obtain functional central limit theorems (FCLTs) by studying the limit of a suitably scaled fluctuation process of the stochastic empirical process of the model with index $N$ around the mean-field limit when $N\toınfty$. We show that for both the transient and stationary regimes, the limiting process is characterized by an Ornstein-Uhlenbeck (OU) process. We also show that the interchange of limits $łim_N\toınftyłim_t\toınfty=łim_t\toınftyłim_N\toınfty$ is valid under the CLT scaling. Finally, we exploit the FCLT to show that the gap between the exact average blocking probability of a job in the system with the number of servers $N$ and the limiting average blocking probability which is a function of the fixed-point of the mean-field, is of the order $o(N^-\frac12)$ and thus establish the accuracy of the mean-field approximation for finite $N$.},
added-at = {2020-04-29T15:29:04.000+0200},
address = {Budapest, Hungary},
author = {Vasantam, Thirupathaiah and Mazumdar, Ravi R.},
biburl = {https://www.bibsonomy.org/bibtex/29676e9417b0be72dc7cad07265a9586c/itc},
booktitle = {31th International Teletraffic Congress (ITC 31)},
interhash = {bc8d69384c2de917ad9ba3ee6e5946c2},
intrahash = {9676e9417b0be72dc7cad07265a9586c},
keywords = {itc itc31},
timestamp = {2020-04-30T18:18:45.000+0200},
title = {Fluctuations around the mean-field for a large scale Erlang Loss system under SQ(d) load balancing},
url = {https://gitlab2.informatik.uni-wuerzburg.de/itc-conference/itc-conference-public/-/raw/master/itc31/vas19ITC31.pdf?inline=true},
year = 2019
}