A Case Study of Bank Queueing Model: Kasturi Nirmala, DR - Shahnaz Bathul
A Case Study of Bank Queueing Model: Kasturi Nirmala, DR - Shahnaz Bathul
A Case Study of Bank Queueing Model: Kasturi Nirmala, DR - Shahnaz Bathul
Flat No: 108, F Block, Sathyanarayana Enclave, Madinaguda, Miyapur, Hyderabad, 500049, (A.P), India Professor, Dept of Mathematics, JNTUH College of Engineering, Kukatpally, Hyderabad, 500085 (A.P), India.
Abstract:- This paper deals with the Queueing theory and the analysis of queueing system by using probability curves. Starting with the basis of the distributions and important concepts of queueing theory, probability curves of Gamma distribution are used to analyse the banking service. Keywords:- Arrival rate, Service rate, Poission process, Probability distribution, Gamma function, Gamma distributions, Probability graph, Random variable.
I.
INTRODUCTION
Important concepts Discrete Random variables [4] If is an experiment having a sample space S and X is a function that assigns a real number X(e) to every outcome e S, then X(e) is called Random variable. And when it is associated with counting then the Random variable is termed as Discrete Random variable. Continuous random variable If is an experiment having a sample space S and X is a function that associates with one or more intervals, then X is called continuous. Probability Distribution of a Random variable The family of probability distributions that has special importance in probability is the following. If X is a discrete random variable, we associate a number P x (xi) = P x (X = xi) with each outcome xi in R x for I = 1, 2 , n ... where the numbers P x (xi) satisfy the following: P x (xi) > 0 for all i. =1 The function P x is called the probability law of random variable and the pairs [x i, P x (xi)] is called the probability distribution of X. If X is continuous random variable, then for > 0, P X (X = x) = x+ )-F x (x)] = 0 We define probability density function f x (x) as f x (x) = F X (x) And it follows that F X (x) = f x (x) 0 for all x RX
Then P {e S: a< X (e) b} = P X (a X b) = Characteristics of some discrete distributions: The stem of the probability distributions is Bernolli trials. There are experiments such as each experiment consists of n trials. The possible outcome of each trial is success or failure.
Bernoullis Distribution: Perform an experiment I (i th ) whose sample space is I = {S, F}. Now the random variable Xi defined as X I (s) is mapped to 1 and X I (f) is mapped to 0 of Real numbers. The range space Rx is subset of real numbers is [0, 1]. The probability law defined in this Bernoullis distribution with respective to the random variable is
11
X is defined as no. of successes in n Bernoullis trials. Geometric distribution: The geometric distribution is also dependant on the Bernoullis trials. The difference between the Binomial Bernoullis trials and Geometric Bernoullis trials is the number of trials in Binomial is fixed and in that of geometric the number of trials is not fixed. The random variable is defined as the number of trials required to get first success. The probability distribution of random variable X is P (r) = p qr-1 when r = 1, 2, 3 =0 otherwise.
The Pascal distribution: The Pascal distribution also has its origin in Bernoullis trials. In this random variable X described as the trial on which the r th success occurs, where r is an integer. The probability of X is defined as P(r) = n-1cr-1 pr (1-p) n-r when n = r, r+1, r+2 . = 0 otherwise. The Poission distribution: This distribution is used in real world phenomenon. We are not doing any artificial experiments in which we get the results either success or failure. We consider only time oriented occurrences (arrivals). As in the previous distributions we do not make n trials but instead we take number of arrivals that occur in the particular time interval [0, t]. This distribution is discrete distribution and developed in two ways. The first development involves the Poission process. The second development involves that the Poission distribution to be a limiting case of the Binomial distribution. Development of Poission process: In this process the random variable has practical applicability. In defining Poission process, we initially consider a collection of arbitrary time oriented occurrences also called arrivals or births. The random variable X t is the number of arrivals that occur in the particular time interval [0, t]. The range space R Xt = {0, 1, 2, 3 .}. In developing the Poission probability distribution of random variable X t , it is necessary to take some assumptions. They are 1. The number of arrivals during the non overlapping time intervals is independent random variables.
12
= 0 otherwise. The development of Poisson distribution from the Binomial: The Binomial distribution is P(r) = ncr pr (1-p) n-r when n = 0, 1, 2, 3, n. = 0 otherwise. If we let np = c so that p = and 1-p = 1- = and if we then replace terms involving p with corresponding terms involving c we obtain P(r) = [ ]r [ ]n-r =
-r
In letting n and p 0 in such a way that np = c remains fixed, the terms (1- ), (1- ), (1approach to 1, as does (1- ) . we know (1- ) e as n .
Thus limiting form of equation (1) is P n (t) = which is Poisson distribution. Some continuous distributions: The uniform distribution: A random variable X is said to follow continuous uniform distribution in an interval [a, b] if its density function is constant over the entire range of X. I.e. f(x) = K, a x b. We know that =1 =1 K[b-a] = 1 K= Therefore f(x) = a x b. Then p(c x d) = = . The exponential distribution: In this random variable defined as the time between the occurrences. It has the density function f(x) = e-x when x 0. 0 otherwise, where is real positive constant. Gamma distribution: Gamma density function is f (t) = (t)r-1e-r, t > 0 = 0 otherwise. The parameters are r > 0 and > 0. The parameter r is called shape parameter, And the parameter is called scale parameter. The Weibull distribution: The important usage of Weibull distribution is that it gives a fairly accurate approximation to the probability law of many random variables. Time to failure in electrical and mechanical components is one of the applications of this distribution. The probability density function of Weibull distribution is -1 f (x) = [ exp[-( )], x , = 0, otherwise. Introduction Constructive time expended constitutes a small percentage of the total time spent by us on various activities. Severe traffic congestions and bottlenecks eat away a major chunk of time while travelling. Due to a
13
The probability that an arriving has to wait (which is known as Erlang C-formula or deley formula). P n=
Erlangs paper On the rational determination of number of circuits deals with the calculation of the optimum number of channels so as to reduce the probability of loss in the system. Whole theory started with a congestion problem in tele-traffic. The application of queueing theory scattered many areas. It include not only tele-communications but also traffic control, hospitals, military, call-centers, supermarkets, computer science, engineering, management science and many other areas. Important concepts in Queueing theory Little law One of the feet of queueing theory is the formula Little law. This is N = T
This formula applies to any system in equilibrium (steady state).
Where is the arrival rate T is the average time a customer spends in the system N is the average number of customers in the system Little law can be applied to the queue itself. I.e. N q = T q
14
II.
Input process If the occurrence of arrivals and the offer of service strictly follow some schedule, a queue can be avoided. In practice this is not possible for all systems. Therefore the best way to describe the input process is by using random variables which we can define as Number of arrivals during the time interval or The time interval between successive arrivals. If the arrivals are in group or bulk, then we take size of the group as random variable. In most of the queueing models our aim is to find relevant probability distribution for number of customers in the queue or the number of customers in the system which is followed by assumed random variable. Service Process Random variables are used to describe the service process which we can define as service time or no of servers when necessary. Sometimes service also be in bulk. For instance, the passengers boarding a vehicle and students understanding the lesson etc.. Number of servers Queueing system may have Single server like hair styling salon or multiple servers like hospitals. For a multiple server system, the service may be with series of servers or with c number of parallel servers. In a bank withdrawing money is the example of former one where each customer must take a token and then move to the cashier counter. Railway reservation office with c independent single channels is the example of parallel server system which can serve customers simultaneously. System capacity Sometimes there is finite waiting space for the customers who enter the queueing system. This type of queueing systems are referred as finite queueing system. Queue discipline This is the rule followed by server in accepting the customers to give service. The rules are FCFS (First come first served). LCFS (Last come first served). Random selection (RS). Priority will be given to some customers. General discipline (GD). Kendalls notation Notation for describing all characteristics above of a queueing model was first suggested by David G Kendall in 1953. The notation is with alphabet separated by slashes. o A/B/X/Y/Z Where A indicates the distribution of inter arrival times B denotes the distribution of the service times X is the capacity of the system Y denotes number of sources Z refers to the service discipline Examples of queueing systems that can be defined with this convention are M/M/1 M/D/n G/G/n Where M stands for Markov D stands for deterministic G stands for general Definition- state dependent service: The situation in which service depends on the number of customers waiting is referred to as state dependent service. A survey was conducted in relation to the operations of Andhra bank, JNTUH, Hyderabad. The bank provides service to various types of customers on daily basis. The mean deviation of arrival rate of customers in each week is different. In the first week the bank has 6 customers/minute. For the second week it was 4.662 customers/minute. For third and fourth weeks 3.666 customers/minute, and 3 customers/minute respectively. Since the service is state dependent service, the time taken by employee to complete one service on average is different in different weeks. By observations and data from the management of the Bank the time to serve one customer on average is listed below: For first week ----- 2 minutes/customer
15
16
Let us take time t = 5 minutes, then time taken to complete the service within 5 minutes for different weeks is tabulated.
17
1st week
2nd week
3rd week
4th week
2 6 0.060361
3 3 0.087442
Though the arrival rate is more in first week, due to the high service rate the probability of completing the service in 5 minutes is 0.060361. Due to the little bit decrease in service rate, the probability in second week is 0.054344. With all most all equal service rate as second week and with less arrival rate, in third week the probability is 0.074055. In the last week of month though service rate is less, the probability of getting serviced in less time is 0.087442. So a customer enters the bank will completes his work quickly in 4 th week among all weeks. And he takes more time in 2nd week in comparative with other weeks.
III.
CONCLUSION
This paper studies the Queueing theory and by the use of probability curves the analysis of the queueing system. From the work surveyed above we have found arrival and service rates at the bank are different in different weeks. The conclusions are developed from the application of Gamma probability distribution. The arrival rate is higher in week 1. However, the higher arrival rate is offset by higher service productivity. Therefore, the probability of completing the service in 5 minutes in week 1 is 0.060361. The efficiency factor is not constant. The service productivity diminishes and as a result the probability in the second week drops to 0.054344. In the third week the rate of arrival decreases. Consequently, maintaining the same service rate as in the second week, the probability goes up to 0.074055. The service rate is less in the last week of the month as the arrival rates decrease considerably, the probability of completing the service with in less time peaks as compared to the remaining weeks.i.e0.087442.
REFERENCES
[1]. [2]. [3]. The effect of probabilities of arrivals with time in a bank. By Kasturi Nirmala and Dr.shahnaz Bathul; volume 3, issue 8, August 2012, International Journal of Science and Engineering Research. The Effect of probabilities of departures with time in a bank, by K. Nirmala, Dr. Shahnaz Bathul; volume 4, issue 4, October 2012, International Journal of Science and Engineering Research. The study of difference of efficiency of a bank in a rural area and a bank in town, by K. Nirmala, Dr. Shahnaz Bathul; volume 4, issue 4, October 2012, International Journal of Engineering Research and Development. Fundamentals of Queueing theory by James M Thomson, Donald Gross; Wiley series. Operations research by Hamdy Taha. An application of Queueing theory to the relationship between Insulin Level and Number of Insulin Receptors by Cagin Kandemin-Cavas and Levent Cavas. Turkish Journal of Bio Chemistry-2007; 32(1); 32-38 Queueing theory and its applications; A personal view by Janus Sztrik. Proceedings of the 8 th International conference on Applied Informatics, Eger, Hungary, Jan 27-30, 2010, Vol.1, PP.9-30. Case study for Restaurant Queueing model by Mathias, Erwin. 2011 International conference on management and Artificial Intelligence. IPDER Vol.6 (2011) (2011) IACSIT Press, Bali Indonasia. A survey of Queueing theory Applications in Healthcare by Samue Fomundam, Jeffrey Herrmann. ISR Technical Report 2007-24.
18