BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Haoran Wu (University of Waterloo)
DTSTART:20200821T153000Z
DTEND:20200821T160000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/1
DESCRIPTION:Title: Double-Sided Queues with Marked Markovian Arrival Processes and Aband
onment\nby Haoran Wu (University of Waterloo) as part of BIRS workshop
: Canadian Queueing Theorists and Practitioners Conference\n\n\nAbstract\n
In this paper\, we study a double-sided queueing model with marked Markovi
an arrival processes and finite discrete abandonment times. We apply the t
heory of multi-layer Markov modulated fluid flow (MMFF) processes to analy
ze the queueing model. First\, we define three age processes for the queue
ing system and convert them into a multi-layer MMFF process. Then we analy
ze the multi-layer MMFF process to find queueing performance measures rela
ted to the age processes\, matching rates/probabilities\, waiting times\,
and queue lengths for both sides of the queueing system. We obtain a numbe
r of aggregate quantities as well as quantities for individual types of in
puts\, which can be useful for the analysis and design of\, for examples\,
passenger-taxi service systems and organ transplantation systems.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/1/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Qi-Ming He (University of Waterloo)
DTSTART:20200821T160000Z
DTEND:20200821T163000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/2
DESCRIPTION:Title: Bounds on the Mean and Squared Coefficient of Variation of Phase-Type
Distributions\nby Qi-Ming He (University of Waterloo) as part of BIRS
workshop: Canadian Queueing Theorists and Practitioners Conference\n\n\nA
bstract\nWe consider a class of phase-type distributions\, to be called th
e MMPP class of PH-distributions\, and find bounds of their mean and squar
ed coefficient of variation (SCV). As an application\, we have shown that
the SCV of the event-stationary inter-event time of Markov modulated Poiss
on processes (MMPPs) is greater than or equal to unity\, which answers an
open problem about MMPPs\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/2/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Barbara Margolius (Cleveland State University)
DTSTART:20200821T163000Z
DTEND:20200821T170000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/3
DESCRIPTION:Title: Catastrophes and Queueing Systems with Time-Varying Periodic Transiti
on Rates\nby Barbara Margolius (Cleveland State University) as part of
BIRS workshop: Canadian Queueing Theorists and Practitioners Conference\n
\n\nAbstract\nWe study the asymptotic periodic distribution of queues with
time-varying periodic transition rates and catastrophes that occur random
ly according to an exponential distribution with time-varying periodic rat
e. When a disaster occurs\, the system resets\, all customers are lost and
an exponentially distributed period of time elapses before the repair is
complete. Service is governed by a phase distribution. The asymptotic peri
odic distribution of the queue process is analogous to the steady state di
stribution for a system with constant transition rates.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/3/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Suman Thapa (Carleton University)
DTSTART:20200821T180000Z
DTEND:20200821T183000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/5
DESCRIPTION:Title: Construction of New Copulas with Applications to Queueing Models\
nby Suman Thapa (Carleton University) as part of BIRS workshop: Canadian Q
ueueing Theorists and Practitioners Conference\n\n\nAbstract\nIn this pape
r\, we construct the bound copula\, which can reach both Frechet's lower a
nd upper bounds for perfect positive and negative dependence cases. Since
it has a wide range of dependency\, it can be very useful. This new copula
is simple for the computational purpose. It is very difficult to use copu
las such as Archemedes\, Guassian\, t-copula to find the distribution func
tion and the expected value in explicit form. For both copulas\, we derive
the strength of measures of the dependency such as Spearman's rho\, Kenda
ll's tau\, Blomqvist's beta and Gini's gamma\, and the coefficients of the
tail dependency. For the application part\, we analyze the dependency bet
ween two service times to evaluate the mean waiting time and the mean serv
ice time when customers launch two replicas of each task on two parallel s
ervers using the cancel-on-finish policy. We assume that the inter-arrival
time is exponential and the service time is general.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/5/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Myron Hlynka (University of Windsor)
DTSTART:20200821T200000Z
DTEND:20200821T203000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/6
DESCRIPTION:Title: Completing a Task with Interruptions\nby Myron Hlynka (University
of Windsor) as part of BIRS workshop: Canadian Queueing Theorists and Pra
ctitioners Conference\n\n\nAbstract\nAssume that the time to complete a ta
sk without interruption is T. Assume interruptions occur according to a Po
isson process with rate lambda. If a process is interrupted\, it must begi
n again. Let W be the total time to completion. We find the Laplace transf
orm of W. One application is a solution of the classic problem of the time
to cross a one way street assuming Poisson traffic.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/6/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Amir Rastpour (Ontario Tech University)
DTSTART:20200821T203000Z
DTEND:20200821T210000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/7
DESCRIPTION:Title: Algorithms for Queueing Systems with Reneging and Priorities Modeled
as Quasi-Birth-Death Processes\nby Amir Rastpour (Ontario Tech Univers
ity) as part of BIRS workshop: Canadian Queueing Theorists and Practitione
rs Conference\n\n\nAbstract\nAlgorithms for queueing systems with reneging
and priorities modeled as quasi-birth-death processes By Amir Rastpour\,
Ontario Tech. University Abstract: We develop an iterative algorithm for a
class of infinite level-dependent quasi-birth-and-death (LDQBD) systems.
The class of queueing systems that we focus on includes the Erlang A syste
m with two priority classes where customers from both classes are impatien
t\, and they can have different arrival\, service\, and abandonment rates.
Our algorithm provides upper and lower bounds for stationary probabilitie
s and automatically proceeds until a pre-specified error tolerance is achi
eved. Our algorithm exploits an approach that can be used to obtain elemen
t-wise bounds for the rate matrix of any LDQBD system within any desired e
rror tolerance. We generate a wide range of instances and perform numerica
l analyses on them. We report numerical results and discuss algorithm limi
tation\, accuracy and speed.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/7/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Douglas Down (McMaster University)
DTSTART:20200821T210000Z
DTEND:20200821T213000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/8
DESCRIPTION:Title: Size-based Scheduling with Estimation Errors\nby Douglas Down (Mc
Master University) as part of BIRS workshop: Canadian Queueing Theorists a
nd Practitioners Conference\n\n\nAbstract\nWhen job sizes are known\, Shor
test Remaining Processing Time (SRPT) is known to be an optimal (in a very
strong sense) scheduling policy for a single server queue under general a
ssumptions on underlying random variables. However\, the performance of SR
PT is known not to be robust to errors in processing time estimates. For a
popular error model\, we characterize the optimal policy using a Gittins
Index approach and discuss its properties. The implementability of the pol
icy is studied\, with the structure of the optimal policy guiding heuristi
c policies that appear to perform well. Time permitting\, issues for multi
ple server queues will be highlighted.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/8/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Peter Taylor (University of Melbourne)
DTSTART:20200821T220000Z
DTEND:20200821T230000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/9
DESCRIPTION:Title: Admission Policies for Complex Resource Allocation Problems\nby P
eter Taylor (University of Melbourne) as part of BIRS workshop: Canadian Q
ueueing Theorists and Practitioners Conference\n\n\nAbstract\nThere are ma
ny applications where users of different types arrive to a finite set of r
esources and request temporary use of subsets of these resources. The mana
ger of the resources is entitled to charge for their use and might incur s
ome costs in making them available.\n\nIn this context\, the manager has a
n admission control problem. Given the current state of allocation\, can t
hey admit a user of a particular type? If they can\, should they\, or wait
for a more lucrative user to arrive in the future? In the situation where
a user is indifferent between different sets of resources\, which ones sh
ould they choose?\n\nIn this talk I shall discuss my long history of engag
ing with different versions of this problem\, culminating with some curren
t work with Jing Fu and Bill Moran in which we are looking at a restless m
ulti-armed bandit formulation.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/9/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Winfried Grassmann (University of Saskatchewan)
DTSTART:20200822T143000Z
DTEND:20200822T153000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/10
DESCRIPTION:Title: Queueing Theory in a World where most Queueing Problems are Solved b
y Simulation\nby Winfried Grassmann (University of Saskatchewan) as pa
rt of BIRS workshop: Canadian Queueing Theorists and Practitioners Confere
nce\n\n\nAbstract\nMonte Carlo simulation is one of the most successful te
chniques\, not only in operations research and performance evaluation\, bu
t in science in general. One reason for this extraordinary success is its
flexibility. In contrast\, most queueing models are rather specialized. In
this talk\, we suggest methods to make queueing theory more flexible. In
particular\, we suggest an event-based approach\, which provides great fle
xibility for the modeller. We also show how to convert such event-based mo
dels into Markov chains\, which can then be solved by classical numerical
methods. The suggested method is particularly suited for small models\, wh
ere its execution times are much lower than Monte-Carlo simulation. For la
rger problems\, the curse of dimensionality takes over\, and the execution
times based on classical numerical methods increase exponentially. This m
eans that for complex models\, simulation finds numerical solutions with l
ess computer time than classical numerical methods.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/10/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Javad Tavakoli (University of British Columbia - Okanagan)
DTSTART:20200822T153000Z
DTEND:20200822T160000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/11
DESCRIPTION:Title: The Distribution of the Line Length in a GI/G/1 Queue Using Distribu
tion Little Laws and Roots Methods\nby Javad Tavakoli (University of B
ritish Columbia - Okanagan) as part of BIRS workshop: Canadian Queueing Th
eorists and Practitioners Conference\n\n\nAbstract\nIn this talk\, we prov
ide a method\, called L\, based on distributional law of Little (DLL) to d
etermine the equilibrium distribution of the number of elements in a discr
ete-time GI/G/1 queueing system. We also clarify a number of issues\, and
provide a number of new results. We assume that the inter-arrival times ra
nge from 1 to g+1 and the service times from 1 to h+1. The majority of aut
hors have formulated the system in question as a quasi birth and death pro
cess\, which can be solved by the matrix iterative methods pioneered by Ne
uts\, methods that are cubic in the number of phases\, and in fact\, if g=
h\, all matrix analytic methods we found in literature are cubic in g\, or
worse. In contrast\, our method L\, which finds the distribution of the n
umber of elements in the system in quadratic time. This implies that for l
arge enough g and h\, our algorithm will outperform all cubic algorithms\,
a claim verified by numerical tests. In particular\, for realistic values
g and h\, algorithm L is more than 50 times faster than the different alg
orithms based on matrix analytic methods we found in literature.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/11/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Ruichao Jiang (UBC Okanagan)
DTSTART:20200822T160000Z
DTEND:20200822T163000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/12
DESCRIPTION:Title: An upper bound for the Galois group of weight walks with rational co
efficients in the quarter plane\nby Ruichao Jiang (UBC Okanagan) as pa
rt of BIRS workshop: Canadian Queueing Theorists and Practitioners Confere
nce\n\n\nAbstract\nUsing Mazur’s theorem on torsions of elliptic curves\
, an upper bound 24 for the order of the finite Galois group H associated
with weighted walks in the quarter plane Z2+ is obtained. The explicit cri
terion of H having order 4 or 6 is given by geometric argument. Using divi
sion polynomial\, a recursive criterion for H having order 4m or 4m + 2 is
also obtained and explicit criterion for H having order 8 is given.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/12/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vera Tilson (University of Rochester)
DTSTART:20200822T170000Z
DTEND:20200822T173000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/13
DESCRIPTION:Title: Models of the Impact of Triage Nurse Standing Orders on Emergency De
partment Length of Stay\nby Vera Tilson (University of Rochester) as p
art of BIRS workshop: Canadian Queueing Theorists and Practitioners Confer
ence\n\n\nAbstract\nStanding orders allow triage nurses in emergency depar
tments (EDs) to order tests for certain medical conditions before the pati
ent sees a physician\, which could reduce the patient’s ED length of sta
y (LOS). Several studies in the medical literature documented a decrease i
n average ED LOS for a target patient population\, resulting from the use
of standing orders. We formulate models of the operational impact of stand
ing orders and test several policies for whether to order tests at triage
for individual target patients\, as a function of ED congestion. We find t
hat a threshold policy\, with a threshold whose value can be estimated eas
ily from model primitives\, performs well across a wide range of parameter
values. We demonstrate potential unintended consequences of the use of st
anding orders\, including over testing and spillover effects on non-target
patients.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/13/
END:VEVENT
BEGIN:VEVENT
SUMMARY:George Zhang (Simon Fraser University)
DTSTART:20200822T173000Z
DTEND:20200822T180000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/14
DESCRIPTION:Title: Performance Analysis of a Markovian Queue with Service Rate and Cust
omers' Joining Decisions\nby George Zhang (Simon Fraser University) as
part of BIRS workshop: Canadian Queueing Theorists and Practitioners Conf
erence\n\n\nAbstract\nWe consider the customers' equilibrium strategy and
socially optimal strategy in a single server Markovian queueing system wit
h changeable service rates controlled by a threshold. When a customer arri
ves at an empty system\, he is served by the server at a lower service rat
e. When the queue length reaches the threshold\, customers are served at a
high service rate. The optimal joining strategies of customers are studie
d under two information scenarios. The first scenario\, where the server'
state and the queue length are observable\, is called a fully observable c
ase. The second scenario\, where the system state is not observable\, is c
alled an unobservable case. We analyze the steady-state distribution and p
erformance measures of the system\, and derive the equilibrium strategy. F
inally\, we compare the equilibrium strategy with socially optimal strateg
y via numerical examples.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/14/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Na Li (Mcmaster University)
DTSTART:20200822T180000Z
DTEND:20200822T183000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/15
DESCRIPTION:Title: A decision integration strategy for short-term demand forecasting an
d ordering for red blood cell components\nby Na Li (Mcmaster Universit
y) as part of BIRS workshop: Canadian Queueing Theorists and Practitioners
Conference\n\n\nAbstract\nBlood transfusion is one of the most crucial an
d commonly administered therapeutics worldwide. The need for more accurate
and efficient ways to manage blood demand and supply is an increasing con
cern in many healthcare systems. Building a technology-based\, robust bloo
d demand and supply chain that can achieve the goals of reducing ordering
frequency\, inventory level\, wastage and shortage\, while maintaining the
safety of blood usage\, is essential in modern healthcare systems. In thi
s study\, we summarize the key challenges in current demand and supply man
agement for red blood cells (RBCs). We combine ideas from statistical time
series modeling\, machine learning\, and operations research in developin
g an ordering decision strategy for RBCs integrating a hybrid demand forec
asting model using clinical predictors\, and a data-driven multi-period in
ventory problem considering inventory and reorder constraints. We have app
lied the integrated ordering strategy to the blood inventory management sy
stem in Hamilton\, Ontario using a large clinical database from 2008 to 20
18. The proposed hybrid demand forecasting model provides robust and accur
ate predictions\, and identifies important clinical predictors for short-t
erm RBC demand forecasting. Compared with the actual historical inventory
levels\, ordering decisions\, and wastage due to expiration\, our integrat
ed ordering strategy reduces the inventory level by approximately 40% and
decreases the ordering frequency by approximately 60%\, with low incidence
of shortages and wastage due to expiration. If implemented successfully\,
our proposed strategy can achieve significant cost savings for healthcare
systems and blood suppliers. The proposed ordering strategy is generaliza
ble to other blood products or even other perishable products.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/15/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Katsunobu Sasanuma (Stony Brook University)
DTSTART:20200822T183000Z
DTEND:20200822T190000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/16
DESCRIPTION:Title: Queueing and Markov chain decomposition method to analyze Markov-mod
ulated Markov chains\nby Katsunobu Sasanuma (Stony Brook University) a
s part of BIRS workshop: Canadian Queueing Theorists and Practitioners Con
ference\n\n\nAbstract\nWe present a Queueing and Markov chain decompositio
n method based on the total expectation theorem. Our decomposition method
requires partial flow to be conserved\, which we call a termination scheme
. This scheme is useful when deriving analytical formulas for complex queu
eing systems. As an example\, we apply our method to derive an exact set o
f stationary equations for the probability generating functions of decompo
sed chains of Markov-modulated continuous-time Markov chains.\n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/16/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Ahmed Sid Ali (Carleton University)
DTSTART:20200822T190000Z
DTEND:20200822T193000Z
DTSTAMP:20260422T185714Z
UID:BIRS_20w2253/17
DESCRIPTION:Title: Fluid model for multiple TCP and UDP connections through a network o
f queues in a random environment\nby Ahmed Sid Ali (Carleton Universit
y) as part of BIRS workshop: Canadian Queueing Theorists and Practitioners
Conference\n\n\nAbstract\nThe Transmission Control Protocol (TCP) is one
of the main protocols of the Internet protocol suite and major internet ap
plications rely on it. The TCP protocol provides reliability\, flow contro
l and congestion control. Alongside the TCP\, the User Datagram Protocol (
UDP) is another transport protocol which\, in contrast to TCP\, is a simpl
ified request-response protocol that does not have any connection setup ti
me and does not provide any flow\, congestion or error controls. We consid
er in this presentation a fluid model for multiple TCP and UDP connections
interacting through a network of queues. We suppose that the connections
are randomly routed according to a dynamical routing table protocol which
takes into account the topology of the network and adapts the routing dyna
mically. Our model extends the multi-class model studied in Graham et al (
2009). The dynamic of the TCP flows follows the additive increase/multipli
cative-decrease (AIMD) protocol and is represented by a stochastic differe
ntial equation w.r.t. a Poisson random measure and the UDP flows are repre
sented by simple point processes. Using an adequate scaling\, a mean-field
result is proved where\, as the number of connections goes to infinity\,
the behaviour of the different connections can be represented by the solut
ion of an original nonlinear stochastic differential equation. The existen
ce and uniqueness of the solution of this equation are derived. Moreover\,
we discuss some open problems and possible extensions. This talk is based
on a current ongoing joint work with Donald A.Dawson and Yiqiang Q.Zhao.\
n
LOCATION:https://researchseminars.org/talk/BIRS_20w2253/17/
END:VEVENT
END:VCALENDAR