See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/316974036

SIMULATING A SINGLE-SERVER QUEUE USING THE Q - SIMULATOR

Article · January 2012

CITATIONS READS

0 2,188

1 author:

Irene Kafui Vorsah Amponsah

University of Cape Coast
24 PUBLICATIONS   6 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

contribution to Society View project

Ghana Association of University Women sponsored Science Clinic for Girls View project

All content following this page was uploaded by Irene Kafui Vorsah Amponsah on 16 May 2017.

The user has requested enhancement of the downloaded file.

 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:6, No:6, 2012

Simulating a Single-Server Queue using the Q –

Simulator
Irene K. Amponsah, Bennony K. Gordor, Francis Dogbey

Abstract—This paper introduces a technique for simulating a Their analysis and simulation work gave insight into
single-server exponential queuing system. The technique called the predicting the system performance and behavior when
Q-Simulator is a computer program which can simulate the effect of employing a certain interrupt-handling scheme. Similarly,
traffic intensity on all system average quantities given the arrival Shaul K. Bar-Lev et al (2007), [6] conducted a research on the
and/or service rates. The Q-Simulator has three phases namely: the
applications of bulk queues to group testing models with
formula based method, the uncontrolled simulation, and the
controlled simulation. The Q-Simulator generates graphs (crystal incomplete identification. They discussed some assumptions
solutions) for all results of the simulation or calculation and can be that gave rise to the bulk queuing model M/G ( m, M ) / 1,
used to estimate desirable average quantities such as waiting times, where m and M > m are the decision variables where each
queue lengths, etc.
batch size can be between m and M. They computed the
International Science Index, Computer and Information Engineering Vol:6, No:6, 2012 waset.org/Publication/5290

Keywords—Automation system-Simulator, Simulation, Single- optimal values of the decision variables ( m, M ) that
server exponential system maximize the expected profit. Abolnikov et al (1993), [1] also
in a related publication, researched on a multilevel control
I. INTRODUCTION bulk queuing system with vacationing server. This article
studied a Markov chain describing the evolution of the queue
E STIMATION of waiting times and other steady-state
average quantities at service points such as banks, airports, in a general single-server bulk queuing system with
continuously operating (or, equivalently vacationing) server,
restaurants, theatres, post offices, and traffic lights has always
semi-Markov modulated compound Poisson input, queue
been an objective in the study and analysis of queues. Based
length dependent service time, and multilevel control server
on models developed, statistical simulation can be performed.
capacity. The studies established a necessary and sufficient
Statistical simulation is driven by a stream of randomly criterion for ergodicity of this Markov chain and found its
generated instructions, based on statistics collected during a stationary distribution. As an example a single-server queuing
single detailed simulation. This method can give accurate system was considered, and some explicit results were
performance estimates within minutes, allowing a large design obtained in this case. Kumar et al (2002),[3] carried out a
space to be simulated quickly. And of course, real data is research concerned with the analysis of a single-server queue
always used to validate the results of the simulation. with Bernoulli vacation schedules and general retrial times. It
A common approach to estimating these quantities is to was assumed that the customers, who found the server busy,
develop statistical models to estimate and write computer were queued in the orbit in accordance with an FCFS (first-
programs to stimulate queues in order to find optimal come-first-served) discipline and only the customer at the head
solutions. S. Nussbaum and J. E. Smith (2002),[4] did prior of the queue was allowed access to the server. They first
work that applied this technique to superscalar processors. presented the necessary and sufficient condition for the system
They evaluated the extension of statistical simulation to to be stable and derived analytical results for the queue length
Symmetric Multiprocessing (SMP) systems. Key program distribution, as well as some performance measures of the
parameters are identified, and program statistics are collected system under steady-state condition. They showed that the
during detailed simulations for both multi-programmed general stochastic decomposition law for M/G/1 vacation
workloads (SpecInt) and parallel scientific workload (Splash- models holds for the present system also. Some special cases
were also studied. These and other related studies can be found
2). The accuracy of statistical simulation is evaluated at
on stochastic queues and their applications.
different levels of model detail, and it is shown that for multi-
However, there are a number of few direct queue simulators.
programmed workloads a 10% average error can be achieved,
For example, an applet (University of Malta, 2010), [7]
and for parallel benchmark programs 15% average error can
simulates a queuing situation where it starts with a newly
be achieved. K. Salah et al (2007), [5] also developed a
opened service with no one being served or waiting. The
performance analysis and comparison of interrupt –handling
arrival rate and service rates are fixed and once you start the
schemes in gigabit net.
simulation, an indicator shows how many customers are
Irene K. Amponsah(nee Vorsah) is with the Department of Mathematics
currently in the system, updated once each simulated second.
and Statistics, School of Physical Science, University of Cape Coast-Ghana ( Another example of a software named Simulator #2 (Baker,
e-mail: Iamponsah@ucc.edu.gh). 2000), [2] requires that you provide the distribution of the
Prof Bennony K. Gordor (Rtd) was with the Department of Mathematics Arrival Intervals, the discipline, the distribution of the service
and Statistics, School of Physical Science, University of Cape Coast-Ghana,
(e-mail: benkgordor@yahoo.co.uk). duration and the simulation controls.
Mr. Francis Dogbey is with the Ghana-India Kofi Annan Centre for
Excellence on ICT- Osu Accra, Ghana (e-mail: fdoggy2002@yahoo.co.uk).

International Scholarly and Scientific Research & Innovation 6(6) 2012 826 scholar.waset.org/1999.4/5290
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:6, No:6, 2012

This paper presents a simulator that simulates the simplest

of queues that is found around us, i.e. the M/1/1. This
simulator is named the Q-Simulator. What sets this simulator
and the others apart is its ability to find the characteristics that
is optimal for setting up an effective single-server queuing
system. This is achieved by the simulator’s ability to examine
the effects of traffic intensity on steady-state average
quantities, given the arrival rate/or the service rate.

II. STRUCTURE OF THE Q-SIMULATOR

The Q-Simulator, software developed for this research is to
specifically automate single-server queuing process. The
software allows the user to specify the arrival rate and the
service rate either in seconds, minutes, or hours. However,
these quantities are recalculated in minutes; this is because it is
more practicable where exponential and poison distributions
International Science Index, Computer and Information Engineering Vol:6, No:6, 2012 waset.org/Publication/5290

are concerned. The Q-Simulator has three options; namely: Fig. 2 Interface Two
formula based method, uncontrolled simulation and the
controlled simulation. A. Using Formula Based Modeling/ Analytical Method
• Select Modeling from Figure 2.
III. OPERATING THE Q-SIMULATOR This will display the screen shown in Figure 3.
• The Q-Simulator Installation (optional)
Follow the instruction in the readme file on the storage
media containing the installation file.
• On the task bar, Click on Start then Program Files then Q-
Simulator (This will display a screen shown in Figure 1).
• From this screen, specify the time unit of the
computation; which may be seconds, minutes, or hours.
• Select the Task menu to display the screen shown in
Figure 2.
• Select appropriate options (e.g. Simulation, Controlled).

Fig. 1 Interface One

Fig. 3 Analytical Method Interface

International Scholarly and Scientific Research & Innovation 6(6) 2012 827 scholar.waset.org/1999.4/5290
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:6, No:6, 2012

• Enter values for Arrival Rate and Service Rate • Click on Chart button to display the chart corresponding
• Click on Compute button to calculate the average to the set of values currently in the database. This will display
quantities or <<Back to Cancel the effect. the graph in Figure 5.
The formula based method calculates the average quantities
based on a known inter-arrival rate, λ and service rate µ . The
process assumes that the researcher computes λ and the µ
from the data. Then, by inputting these values the average
quantities are obtained.
International Science Index, Computer and Information Engineering Vol:6, No:6, 2012 waset.org/Publication/5290

Fig. 5 Effect of Traffic Intensity on Time Customer Spends in Queue

It must be noted that all the system average quantities have

their corresponding graphs plotted against traffic intensity. The
researcher has the prerogative to choose from this list as
regarding the interest at hand.
The controlled simulation observes most of the assumptions
of a single server queuing system. The assumptions include
0 ≤ ρ < 1 and µ > λ . In the controlled simulation, one is
expected to provide the scale of ρ , indicating how ρ values
are increased between [0, 1]. In addition, either µ or λ is
specified which is used together with a ρ value to determine
λ or µ respectively. The computed average system quantity
therefore translates into the desired graphs.
C. Using Uncontrolled Simulation
• Select Simulation and then Uncontrolled from Figure 2,
this will display Figure 6.
• Specify the range of values for Service Rate or Arrival
Rate
• Click on the Simulate button to generate the average
quantities or Back button to leave this screen
• Click on Chart button to display the chart corresponding
to the set of values currently in the database. This will display
Fig. 4 Controlled Simulation Interface a screen similar to Figure 7.
B. Using Controlled Simulation
• Select Simulation and then Controlled from Figure 2.
This will display the screen shown in Figure 4.
• Select the Variable to be controlled. i.e. Service Rate or
Arrival Rate
• Supply the parameters for the scale of the Traffic
Intensity and Controlled variable.
• Click on the Simulate button to generate the average
quantities or Back button to leave this screen.

International Scholarly and Scientific Research & Innovation 6(6) 2012 828 scholar.waset.org/1999.4/5290
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:6, No:6, 2012

[6] Shaul, K. Bar-Lev, M. Parlar, Perry, D, Stadje, W. Van Der Duyn, F. A.

Schouten, “Applications of bulk queues to group testing models with
incomplete identification ” European Journal of Operational Research,
vol.183.1, 2007, pp.226–237
[7] University of Malta / Department of Statistics and OR Tutorial. 2010.
(Assessed9/17/2010) http://staff.um.edu.mt/jsk11/simweb/sq2/sq2.htm1
International Science Index, Computer and Information Engineering Vol:6, No:6, 2012 waset.org/Publication/5290

Fig. 6 Uncontrolled Simulation Interface

The uncontrolled simulation randomly generates λ and µ

within a specified range for a specified number of iterations.
These randomly generated λ and µ values are then used to
compute the rest of the system average quantities.
This process is useful in simulating real life situations where
there is the need to study the effects of uncontrolled λ and µ
on other system average quantities. It is worth noting that, this
process randomly estimates λ and µ purely based on the
specified range and as a result, most of the basic assumptions
of a queuing system are not observed. The essence is to
provide an opportunity to verify the basic assumptions of
queuing theory for single server exponential queues. Although
many of the average quantity computations are reasonable, the
associated graph drawn based on these quantities are
somewhat unrealistic.
A review of this paper in future will highlight queues in
complex systems using the Q – Simulator.

REFERENCES
[1] L. M. Abolnikov, J. H. Dshalalow and A. M. Duhovny “A multilevel
control bulk queuing system with vacationing server,” in Operations
Research Letters, vol.13.3, 1993, pp. 183–188.
[2] S. L. Bakers, “Queue simulation demo”,2000,(Assessed 9/17/2010)
http://hspm.sph.sc.edu/COURSES/J716/demos/StudentLines/Queues.ht
ml1
[3] Kumar, B. Krishna, D. Arivudainambi, A. Vijayakumar. The M(x)/G/1
retrial queue with Bernoulli schedule and general retrial times, Pacific
Journal of Operational Research, 2002 ,
http://findarticles.com/p/articles/mi_qa5473/is_200211/ai_n21323383/
[4] S. Nussbaum, J. E. Smith “Statistical Simulation of Symmetric
Multiprocessor Systems” 35th Annual Simulation Symposium
http://www.ece.wise.edu/
[5] K. Salah, K. El-Badawi and F. Haidari , “Performance analysis and
comparison of interrupt-handling schemes in gigabit net. Computer
Simulation, vol. 30.17, 2007, pp. 3425–3441.

International Scholarly and Scientific Research & Innovation 6(6) 2012 829 scholar.waset.org/1999.4/5290