Baltzer Journals Received 7 January 1997; revised 16 April 1997 An Interpolation Approximation for the GI/G/1 Queue Based on Multipoint Pade Approxima

Size: px
Start display at page:

Download "Baltzer Journals Received 7 January 1997; revised 16 April 1997 An Interpolation Approximation for the GI/G/1 Queue Based on Multipoint Pade Approxima"

Transcription

1 Baltzer Journals Received 7 January 1997; revised 16 April 1997 An Interpolation Approximation for the GI/G/1 Queue Based on Multipoint Pade Approximation Muckai K Girish 1 and Jian-Qiang Hu 1 Telesis Technologies Laboratory 5000 Executive Parkway, Suite 333 San Ramon, CA mgirish@ttl.pactel.com Dept. of Manufacturing Engineering Boston University 15 St. Mary's Street Boston, MA hu@enga.bu.edu The performance evaluation of many complex manufacturing, communication and computer systems has been made possible by modeling them as queueing systems. Many approximations used in queueing theory have been drawn from the behavior of queues in light and heavy trac conditions. In this paper, we propose a new approximation technique, which combines the light and heavy trac characteristics. This interpolation approximation is based on the theory of multipoint Pade approximation which is applied at two points: light and heavy trac. We show how this can be applied for estimating the waiting time moments of the GI/G/1 queue. The light trac derivatives of any order can be evaluated using the MacLaurin series analysis procedure. The heavy trac limits of the GI/G/1 queue are well known in the literature. Our technique generalizes the previously developed interpolation approximations and can be used to approximate any order of the waiting time moments. Through numerical examples, we show that the moments of the steady state waiting time can be estimated with extremely high accuracy under all ranges of trac intensities using low orders of the approximant. We also present a framework for the development of simple analytical approximation formulas. Keywords: GI/G/1 queue, heavy trac limits, MacLaurin series, multipoint Pade approximation. 1 Introduction Queueing systems are used extensively to model various manufacturing, communication and computer systems. The estimation of the congestion measures in these systems is one of the main issues in queueing theory and has been studied widely. The lack of Queueing Systems: Theory and Applications, Vol. 6, 69-84, 1997.

2 M. K. Girish and J-Q. Hu / An Interpolation Approximation analytical solutions for many systems have led to the development of approximations in the literature. These approximations can be very accurate, but, can also be inaccurate under certain conditions. Many approximations have been built upon the behavior of queues in light or heavy trac conditions which make them work well in and around the respective trac intensities. It has, thus, become clear that a robust approximation should takeinto account both the light trac and heavy trac characteristics. A few such approximations have appeared in the literature which use various interpolation methods. The contribution of this paper is the development of a novel approach ofinterpolation based on the light and heavy trac characteristics. In particular, we apply the approachto estimate the moments of the waiting time in the GI/G/1 queue. We use the heavy trac limits and our method provides the exibility tochoose as many light trac derivatives as needed. The performance measures that we are interested in estimating in this paper are the moments of the steady state waiting time. We consider a parameterized queue in which the service time is S, where is a scale parameter and S is the service time random variable in the original queue. Therefore, when = 0, the parameterized queue is in light trac and when =1= this queue is in heavy trac, where is the trac intensity of the original queue. Note that some authors prefer to set the arrival rate equal to zero in light trac and equal to the service rate in heavy trac. Then, we calculate the light trac derivatives and the heavy trac limits of all the waiting time moments. Using this information, we develop an approximation procedure based on the technique of multipoint Pade approximation to estimate the performance measures. The estimates of the waiting time moments are obtained by setting =1. Analysis of queues in light trac have been carried out by many authors. Methods are developed in [18] to calculate the derivatives of the performance measures of arbitrary order in light trac for single server queues with Poisson arrival processes. Likelihood ratio methods were used in [0] toevaluate the light trac derivatives. Extension of the light trac behavior to multi-server queues was initiated by []. A new way to calculate the light trac derivatives of the GI/G/1 queue was proposed in [7]. They showed that the moments of the waiting and system time can be expanded as MacLaurin series in terms of a scale paramenter in the service distribution. These MacLaurin series coecients can be calculated using simple recursive equations. We use this procedure to evaluate the light trac derivatives in this paper. A procedure was developed in [5] to analyze tandem queueing networks using the light trac derivatives. Their numerical results show that this MacLaurin series approximation technique works well in most cases. It was also realized that the heavy trac limits play a critical role in the accuracy of the performance measures especially when the trac intensity gets closer to one. The property of single server queues in heavy trac is well known. In heavy trac, the departure process approaches the service distribution and these heavy trac limits have been successfully used in many approximations. Simple approximations based on the heavy trac limits were prosed in [15] and [16]. This was extended to open queueing networks with general external arrival and service distributions by [1]. Diusion approximations and reected Brownian Motions are very popular heavy trac approximation schemes (see [11], [14] and the references therein). Atwo-moment approximation using this approach for open queueing networks was developed in [10]. An interpolation approximation method was studied by [19]. They consider a queueing system with Poisson arrival process and evaluate the light trac derivatives and the

3 M. K. Girish and J-Q. Hu / An Interpolation Approximation 3 heavy trac limit. The performance measures are then approximated by rational functions. But, the selection of the function is quite arbitrary and there is no systematic way of selecting the correct function to be used, except for some special cases. This interpolation approximation has been found to work well for those cases, for which the denominator of the rational functions can be chosen correctly. Our approach considers queues with general independent arrival and service distributions and we give avery easy procedure to calculate the coecients of the rational functions. Another interpolation approximation was proposed by [] for the average workload in the GI/G/1 queue. This method uses the rst three moments and the density function of the interarrival time and the mean and variance of the service distribution. This approximation also works well, but, is restricted in terms of the number of terms to use and the evaluation of the approximation parameters is not striaghtforward. An ecient technique to approximate the performance measures which have derivatives at any point is the so called Pade approximation (see [1] and [17] for a comprehensive mathematical treatment). [9] showed that Pade approximation can be applied to a wide class of these systems including the GI/G/1 queue. This method was used by [5] to approximate the moments of the waiting and the interdeparture time in tandem queueing networks. If a performance measure function has information (such as values or derivatives) at several points, then we can use multipoint Pade approximation. The mathematical theory of multipoint Pade approximation is described in detail in [1] which forms the basis of our approximation. Multipoint Pade approximation has been used by [8] for estimating the waiting time and is applied at many points where the derivatives could be found by derivative estimation techniques or simulation. But, they do not use the heavy trac limits. The performance measures of interest in this paper are the moments of the steady state waiting time. Our approximation scheme essentially considers the heavy trac limit and an arbitrary number of the light trac derivatives. We then apply the theory of multipoint Pade approximation to estimate the performance measures. We consider a number of numerical examples in this paper which show that the waiting time moments can be estimated with extremely high accuracy. A signicant feature of this method is that even low orders of the rational approximant produce excellent estimates. We also compare this method with an example of the interpolation approximation of [] for the mean workload in a GI/G/1 queue. The rest of this paper is organized as follows. In Section, we introduce some notation and review the results pertaining to light and heavy trac behavior of queues. We present the multipoint Pade approximation technique in Section 3. A framework for the development of simple approximations is presented in Section 4. In Section 5, we consider some numerical examples and this paper is concluded in Section 6. Light and Heavy Trac Analysis In this section, we review the results in queueing theory that treat a queue in light and heavy trac conditions in order to lay the groundwork for our approximation technique. The relevant theory is taken from [5], [7] and [14]. First we introduce the following

4 M. K. Girish and J-Q. Hu / An Interpolation Approximation 4 notation: W = steady-state waiting time of a job T = steady-state system time of a job A = interarrival time between two adjacent jobs S = service time of a job, where S is independent of = a scale parameter in the service time w nm = The coecient of m in the MacLaurin series of E[W n ]=n! t nm = The coecient of m in the MacLaurin series of E[T n ]=n! f() = probability density function (p.d.f.) of A k = f (k) (0 + )=k-th right hand derivative off()at0 k = E[S k ]=k! k = E[A k ]=k! = E[S]=E[A] The interarrival and service distributions that we consider in numerical examples in this paper are mostly mixed Erlang distributions. It has been established that mixtures of Erlang distributions are dense and can approximate any distribution. A mixed Erlang distribution is characterized by the following parameters: b = number of branches p i = probability of branch i, for i =1;:::;b n i = number of stages in branch i, for i =1;:::;b i = rate in branch i, for i =1;:::;b It is well known that E[W n ] is inversely proportional to (1, ) n and so as a rst step, we normalize E[W n ] by a factor of (1, ) n in our estimation. Let n = minimum(n 1 ;n ;:::;n b ). It was observed by [5] that for the GI/G/1 queue with mixed Erlang interarrival distribution, the MacLaurin series coecients of E[W n ], w nm = 0 for m =0;1;:::;n+n,1. Hence, we estimate (1, ) n E[W n ]= n+n in this paper. Note that we can set n = 1 without loss of generality, in case we use distributions other than mixed Erlang (for example, uniform or deterministic). The only reason for using n is that the approximation produces the same estimates using lower orders of the approximant..1 Light Trac Derivatives Queues in light trac have been studied in detail by many authors. It was shown by [7] that the light trac derivatives, w nm and t nm can be calculated based on the following recursive equations t nm = w nm = 8 < : P n ; m = n n i=1 n,iw i(m,n+i) ; m>n 0; m<n P m,n,1 i=0 i t (n+1+i)m ; m>n 0; mn (1) ()

5 M. K. Girish and J-Q. Hu / An Interpolation Approximation 5 The moments of the waiting time was shown to be analytic at = 0 under some mild conditions by [1]. By analyticity, we mean that the MacLaurin series exists and converges to the right value around =0. Since we are interested in estimating (1, ) n E[W n ]= n+n, n =1;;:::,we need to obtain the MacLaurin series for these which, derived with suitable modications is given below: (1, ) 1X n X n E[W n ]=n! w n+n n(m+n+n) (,) i n m+i (3) i m=0 i=0 The above MacLaurin series expansion is based on the derivatives at = 0 and the light trac derivatives for our performance function can be obtained from this.. Heavy Trac Limits It is very easy to determine the heavy trac limits of the GI/G/1 queue, which can be obtained by taking the limits as! 1. The heavy trac limit for the rst moment of the waiting time is well known in the literature and is given below: lim (1, )E[W a+c s) 1!1= ]=(c where c a and c s represent the squared coecients of variation (variance divided by the square of the mean) of the interarrival and service time distributions, respectively. For the GI/G/1 queue, as! 1, the moments of the interdeparture time approach the moments of the service time. i.e., E[D n ]! n! n n ; for n =1;;::: (5) where D is the interdeparture time random variable. Then we use the following relationship between the waiting and interdeparture time presented in [5]: where E[ W n X n n! ] = (,1) n+j n+1,j E[ W j,1 (j, 1)! ]n+1,j j=1 3 + (,1)n+1 4 D n+1 n+1 X n+1, E[ 1, 1 (n + 1)! ]+ (,1) j n+1,j n+1,j j 5 (6) j = " j, X i=0 ( j,i, 1 j,i,1 )E[ W i j=1 i! ]j,i,1 +(,1) j,i j,i ix l=0 i,l E[ W l l! ]i,l # Iteratively, starting with n =, we set = 1 = 1! 1 in Equation (6) and use Equations (4) and (5) to obtain the heavy trac limits of the moments of the waiting time in the GI/G/1 queue as: lim (1,!1= )n E[W n (c ]=n! a +c n s) 1 (7) In the next section, we will show how the light trac derivatives and the heavy trac limits can be combined using the idea of multipoint Pade approximation. (4)

6 M. K. Girish and J-Q. Hu / An Interpolation Approximation 6 3 The Interpolation Approximation In this section, we show how the Multipoint Pade approximation technique given in [1] can be applied to estimate the moments of the waiting time. Let us call the function to be approximated as g () =(1,) n E[W n ()]= n+n, for n =1;;:::. The derivatives of g () are known at 1 = 0, and its limit is known at = 1=. Let g n 1 () be the MacLaurin expansion of g n () up to order m 1, 1 and let g n () be the Taylor expansion of g n () expanded around = up to order 0. These can then be expressed as: g 1 () = mx 1,1 j=0 u 1j j + O[ m1 ] (8) g () = u 0 + O[, ]; (9) where, from Equations (3) and (7), we get for n =1;;:::, (c u 0 = n! n+n a + c s) 1 u 1m = n! n X i=0(,) i n i The rational approximant that we are interested in is g () P L () Q M n (10) w n(m+n+n,i): (11) () : (1) We call Equation (1) a [L/M] multipoint Pade approximant, where L + M = m 1. Note that we use m 1,1 light trac derivatives and the heavy trac limit. The approximating rational function has polynomials of order L and M in the numerator and denominator, respectively. Next we dene an appropriate auxiliary function and by substituting it in the general multipoint Pade approximation scheme derived by [1] and dierentiating and taking limits (we omit the details here since the procedure is quite straightforward), we get: mx 1,1 i = u 0 i,m1,1, q () = u 0 q,m1 m1 + k=0 X u 1(m 1,i,k) k+1 m 1,q,1 m 1,1,k,q k=0 X i=0 u 1(m 1,1,q,k,i) (, ) m1,k,1 It is now easy to evaluate the rational approximant which is given in Equation (15). i+1 (13) (14)

7 M. K. Girish and J-Q. Hu / An Interpolation Approximation 7 For its derivation, refer to [1]. P L () Q M () = det det M M,1 ::: M.... M+1 M ::: 1 M M M,1 ::: 0 M,1 ::: M... M+1 M ::: 1 M M,1 ::: 1 (15) In the next section, a framework for the development of simple formulas for estimating the waiting time moments is developed. Numerical results are presented in Section 4. 4 Simple Approximations The interpolation approximation developed in the previous section can be used to come up with simple approximations, by choosing a few coecients at a time. For example, let us start with a [/1] approximant for (1, )E[W ]= (note that we set n = 1). We omit the superscript for convenience since we deal only with the rst moment in this section. w 10 = w 11 =0 w 1 = 0 w 13 = ( ) w 14 = ( )+ 3 0( ) u 1m = w 1(m+), w 1(m+1) for m 0 1 = 3 u 0, u 1, u 11, 3 u 10 = u 0, u 11, u 10 1 () = u 0 3 +(1,)[u 10 +(u 11 + u 10 ) ] () = 3 u 0 3 +(1,)[u 10 +(u 11 + u 10 ) +(u 1 + u 11 + u 10 ) ] From the above, we have E[W()] = ( 0 (), 1 1 ()) (1, )(, 1 ) (16) Next, we estimate (1, )E[W ] with a [/1] approximant. Then after appropriate modications to the interpolation approximation formulas, we get the following leading to Equation (17): u 0 = (c a + c s) 1

8 M. K. Girish and J-Q. Hu / An Interpolation Approximation 8 1 = 3 u 0, u 1 = u 0 0 () = 3 u 0 3 +(1,)u 1 1 () = u 0 3 E[W ]= 1(c a+c s)( 0 1 (c s +1)+(1, )(c a + c s)) (1, )( 0 1 (c s +1)+(1,)(c a + c s)) The above can be extended to the higher moments as well as for higher orders of the approximant. This approach can serve as a framework for the development of simple approximations for the GI/G/1 queue. In the next section, we present several numerical examples of the sigle server queue. (17) 5 Numerical Examples In this section, we present several numerical examples to compare the accuracy of the multipoint Pade approximation with the exact (simulation) values and also with other approximation procedures. In the rst example, we compare three approximations. The MacLaurin Series Approximation (MSA) is based on the alternate method developed by [5] which uses rational approximation based on the MacLaurin series to estimate the moments of the interdeparture time and then calculates the moments of the waiting time. The Queueing Network Analyzer (QNA) is a popular two-moment based approximation developed by [1]. The interpolation approximation developed in this paper is compared with the above two techniques. In the second example, we compare the waiting time estimates of our interpolation approximation with that of simulation. In the last example, we compare our approximation with an interpolation approximation for the mean workload developed by []. Example 1. In this example, a number of arrival and service distributions are considered, which are either Erlang (E 3 ), hyperexponential (H ) or three-branched mixed Erlang (ME). The rst two moments of the waiting time are compared at various trac intensities ( =.1,.3,.5,.7,.9). The arrival rate is set at 1 for all the distributions. For H arrival distribution, the parameters are: (p 1 ;p ) = (.6,.4) and ( 1 ; ) = (3.0, 0.5). For H service distributions, the parameters are (p 1 ;p )=(:6;:4) and ( 1 ; ) = 8 >< For ME arrival distribution, the parameters are: >: (1:0; 8:0); when = 0:1 (3:0; 4:0); when =0:3 (:4; 1:6); when =0:5 (1:7143; 1:149); when = 0:7 (1:3333; 0:8889); when = 0:9. (p 1 ;p ;p 3 ) = (:;:3;:5) (n 1 ;n ;n 3 ) = (; 4; 3) ( 1 ; ; 3 ) = (:0; :4; 5:0);

9 M. K. Girish and J-Q. Hu / An Interpolation Approximation 9 and for ME service distributions, the parameters are: (p 1 ;p ;p 3 ) = (:3;:;:5) (n 1 ;n ;n 3 ) = (4; ; 3) ( 1 ; ; 3 ) = 8 >< >: (40:0; 10:0; 50:0); when = 0:1 (15:0; 4:0; 1:5); when = 0:3 (8:0; 4:0; 6:0); when =0:5 (5:0; :71; 4:8); when = 0:7 (4:0; :0; 3:75); when = 0:9. The approximations that we consider are the Multipoint Pade (M-Pade), Girish and Hu's MacLaurin Series Approximation (MSA) and Ward Whitt's Queueing Network Analyzer (QNA). The simulation runs consist of 40 replications of 50,000 customers with 10,000 warm-up customers. All the simulation results in this paper are presented as estimate standard deviation. Numerical results are given in Tables I-VI. We evaluate the average absolute percentage error for all the approximants and this serves as a summary comparison characteristic among the various approximation schemes. For the M E=G=1 queue (Tables I and II), the M-Pade results are clearly superior to MSA as well as QNA for both E[W ] and E[W ] estimates. Even the [/] M-Pade approximation is better than the others; and the [3/3] M-Pade approximant improves the accuracy even more. Note that the [6/5] MSA approximant leads to large errors and so we used a very high approximant, [16/16], which of course does well, but not as well as the interpolation approximation. For the E[W ] estimates of the H =G=1 queue (Tables III), the [/] M-Pade approximation again leads to the lowest average absolute % error. But note that the [16/16] MSA approximant's error is only 0.73%. It has been found in [5] that MSA works very well when c a and/or c s are greater than one. But, we need higher orders of MSA to get to this accuracy level. For the E[W ] estimates, the error of the [/] M-Pade approximation is only worse than the [16/16] MSA error. The reduction in the order of the approximant that gives estimates with reasonably high accuracy is quite prominent and supports the use of the interpolation approximation. For the E 3 =G=1 queue, for both E[W ] and E[W ](Tables V and VI), the [/] M- Pade approximation is better than QNA and [6/5] MSA. Similar to the E[W ] estimates of the H =G=1 queue, the [16/16] MSA is very accurate. But the extra computation required for this high order of the approximant doesnotoutweigh the reduction in the accuracy for the interpolation approximation which is only slightly worse and is well within the acceptable level. Example. The arrival process in this example is uniformly distributed between [0,] and the service distribution is Erlang with stages. We consider various trac intensities and compare the rst two moments of the waiting time evaluated using multipoint Pade approximation with that of simulation. We actually estimate (1,) n E[W n ], for n =1; in this example. It can be easily observed that i =0,fori1. Numerical results are given in Table VII. Our interpolation approximation's errors are.006% and 9.394% for the rst and second moment, respectively which are reasonable. It was proved by [1] that the moments of the waiting time are not analytic for this system since the uniform interarrival distribution function is not analytic. We believe that this might explain the relatively poorer performance of the interpolation approximation.

10 M. K. Girish and J-Q. Hu / An Interpolation Approximation 10 Example 3. In this example, we compare our approximation technique with the results in Table II of [], where he developed approximations for the normalized mean workload, c z. The mean waiting time is related exactly to the mean workload by the following expression (see [3]): c z E[W ]= (1, ), 1 1 (18) We use Equation (18) to convert the mean workload given in Table II of [] to the mean waiting time. The rate parameters for the E and exponential distributions are given in Tables VIII and IX. For the H service distribution, p 1 = and p = and the rates are given in Table X. We have omitted the cases in Table II of [] with = 0.0 and 1.0. Numerical results for E[W ] are shown in Table VIII-X. It can be seen that our method is clearly superior (the average absolute % errors for the three cases are 0.45%, 0.19% and 0.576% for M-Pade and 7.374%, 7.431% and 3.433% for Whitt, respectively). 6 Conclusions In this paper, we proposed a novel approach of interpolation for the GI/G/1 queue which eciently combined the light and heavy trac behavior of queues in arriving at very accurate estimates for the moments of the waiting time. The theory of multipoint Pade approximation formed the basis of the method. This technique requires the heavy traf- c limits which can be easily calculated, and any number of the light trac derivatives which can be evaluated using recursive formulas easily. This method generalizes some of the interpolation approximations proposed in the literature. A signicant feature of this method is that it is applicable to any order of the moments. We also show that very low orders of the approximant produces very accurate and robust estimates. We conducted several numerical experiments and the results show that the multipoint Pade approximation is superior to any known approximations under all trac intensities. We also developed a framework for simple approximations using this interpolation scheme. Though not mentioned before, the same approach developed in this paper for the waiting time moments can be extended to the moments of the system time as well as the interdeparture time. The light trac derivatives of system time and interdeparture time have been derived in [7] and [13], respectively. The heavy trac limits of the moments of the system time and the interdeparture time can also be easily evaluated. A direction for future research will be the study of the convergence properties of multipoint Pade approximation. Our numerical experience indicated that when the number of stages in the mixed Erlang arrival distribution is high (more than 8), this procedure does not work well in medium to heavy trac conditions. Another possible direction is to extend this methodology to estimate the performance measures of other queueing systems as well for which light trac derivatives and heavy trac limits can be evaluated. An example is the estimation of the waiting and interdeparture time moments of the G/G/1 queue with Markov-modulated arrivals which has been widely used to model queueing systems with autocorrelated arrivals. The light trac derivatives of the waiting time and the interdeparture time were derived by [3] and [6], respectively which together with the heavy trac limits (if they exist) facilitate the application of interpolation approximations.

11 M. K. Girish and J-Q. Hu / An Interpolation Approximation 11 Acknowledgments This research is supported in part by the National Science Foundation under grants DDM , and EID-911. The authors wish to thank the anonymous referee for the helpful comments. References [1] G. A. Baker, Essentials of Pade Approximants. Academic Press, [] D. Y. Burman and D. R. Smith, A Light-Trac Theorem for Multi-Server Queues, Math. Oper. Res., 8, 15-5, [3] K. W. Fendick and W. Whitt, Measurements and Approximations to Describe the Oered Trac and Predict the Average Workload in a Single-Server Queue, Proceedings of the IEEE, Vol. 77, No. 1, , [4] K. W. Fendick, V. R. Saksena AND W. Whitt, Dependence in Packet Queues, IEEE Transactions on Communications, 37, , [5] M. K. Girish and J. Q. Hu, Higher Order Approximations for Tandem Queueing Networks, Queueing Systems: Theory and Applications, Vol., 49-76, 1996a. [6] M. K. Girish and J. Q. Hu, The Departure Process of the MAP/G/1 Queue, Journal of DEDS, submitted for publication, 1996b. [7] W. B. Gong and J. Q. Hu, The MacLaurin Series for the GI/G/1 Queue, J. Applied Prob., 9, , 199. [8] W. B. Gong, A. Yan and D. Tang, Rational Representations for Performance Functions of Queueing Systems, Princeton Conference on System & Information Sciences, 199. [9] W. B. Gong, S. Nananukul and A. Yan, Pade Approximation for Stochastic Discrete Event Systems, IEEE Trans. on Automatic Control, 40, No. 8, , [10] J. M. Harrison, and V. Nguyen, The QNET Method for Two-Moment Analysis of Open Queueing Networks, Queueing Systems: Theory and Applications, 6, 1-3, [11] J. M. Harrison and M. I. Reiman, Reected Brownian Motion on an Orthant, Ann. Prob., 9, , [1] J. Q. Hu, Analyticity of Single-Server Queues in Light Trac, Queueing Systems: Theory and Applications, 19, 63-80, [13] J. Q. Hu, The Departure Process of the GI/G/1 Queue, Operations Research, 44, No. 5, , [14] L. Kleinrock, Queueing Systems, Vol., Addison-Wesley, [15] P. J. Kuehn, Approximate Analysis of General Queueing Networks by Decomposition, IEEE Transactions on Communications, COM-7, , [16] K. T. Marshall, Some Inequalities in Queueing, Operations Research, 16, , [17] P. P. Petrushev, and V. A. Popov, Rational Approximation of Real Functions. Cambridge University Press, [18] M. I. Reiman and B. Simon, Open Queueing Systems in Light Trac, Math. Operations Research, 1988a. [19] M. I. Reiman and B. Simon, An Interpolation Approximation for Queueing Systems with Poisson Input, Operations Research, Vol. 36, No. 3, , 1988b. [0] M. I. Reiman and A. Weiss, Light Trac Derivatives via Likelihood Ratios, IEEE Trans. Inf. Technol., Vol. 35, , [1] W. Whitt, The Queueing Network Analyzer, The Bell System Technical Journal, 6, No.9, , [] W. Whitt, An Interpolation Approximation for the Mean Workload in the GI/G/1 Queue,

12 M. K. Girish and J-Q. Hu / An Interpolation Approximation 1 Operations Research, Vol. 37, No. 6, , [3] Y. Zhu and H. Li, The MacLaurin Expansion for a G/G/1 Queue with Markov-Modulated Arrivals and Services, Queueing Systems: Theory and Applications, , 1993.

13 M. K. Girish and J-Q. Hu / An Interpolation Approximation 13 Table I E[W ] estimates for the ME=G=1 queue in Example 1 Service M-Pade M-Pade MSA MSA QNA Simulation Dist. [/] [3/3] [6/5] [16/16] ME ME ME ME ME H H H H H E E E E E Avg abs % err Table II E[W ] estimates for the ME=G=1 queue in Example 1 Service M-Pade M-Pade MSA MSA QNA Simulation Dist. [/] [3/3] [6/5] [16/16] ME ME ME ME ME H H H H H E E E E E Avg abs % err

14 M. K. Girish and J-Q. Hu / An Interpolation Approximation 14 Table III E[W ] estimates for the H =G=1 queue in Example 1 Service M-Pade MSA MSA QNA Simulation Dist. [/] [6/5] [16/16] ME ME ME ME ME H H H H H E E E E E Avg abs % err Table IV E[W ] estimates for the H =G=1 queue in Example 1 Service M-Pade MSA MSA QNA Simulation Dist. [/] [6/5] [16/16] ME ME ME ME ME H H H H H E E E E E Avg abs % err

15 M. K. Girish and J-Q. Hu / An Interpolation Approximation 15 Table V E[W ] estimates for the E 3=G=1 queue in Example 1 Service M-Pade MSA MSA QNA Simulation Dist. [/] [6/5] [16/16] ME ME ME ME ME H H H H H E E E E E Avg abs % err Table VI E[W ] estimates for the E 3=G=1 queue in Example 1 Service M-Pade MSA MSA QNA Simulation Dist. [/] [6/5] [16/16] ME ME ME ME ME H H H H H E E E E E Avg abs % err

16 M. K. Girish and J-Q. Hu / An Interpolation Approximation 16 Table VII E[W ] and E[W ] estimates for the U=E =1 queue in Example [4/] M-Pade Simulation [4/] M-Pade Simulation E[W ] E[W ] E[W ] E[W ] Avg abs % err Table VIII E[W ] estimates for the E =E =1 Queue in Example 3 [/] M-Pade Whitt Exact Avg abs % err Table IX E[W ] estimates for the E =M=1 Queue in Example 3 [/] M-Pade Whitt Exact Avg abs % err

17 M. K. Girish and J-Q. Hu / An Interpolation Approximation 17 Table X E[W ] estimates for the E =H =1 Queue in Example 3 1 [/] M-Pade Whitt Exact Avg abs % err

One important issue in the study of queueing systems is to characterize departure processes. Study on departure processes was rst initiated by Burke (

One important issue in the study of queueing systems is to characterize departure processes. Study on departure processes was rst initiated by Burke ( The Departure Process of the GI/G/ Queue and Its MacLaurin Series Jian-Qiang Hu Department of Manufacturing Engineering Boston University 5 St. Mary's Street Brookline, MA 2446 Email: hqiang@bu.edu June

More information

2 light traffic derivatives for the GI/G/ queue. We shall see that our proof of analyticity is mainly based on some recursive formulas very similar to

2 light traffic derivatives for the GI/G/ queue. We shall see that our proof of analyticity is mainly based on some recursive formulas very similar to Analyticity of Single-Server Queues in Light Traffic Jian-Qiang Hu Manufacturing Engineering Department Boston University Cummington Street Boston, MA 0225 September 993; Revised May 99 Abstract Recently,

More information

Higher order approximations for the single server queue with splitting, merging and feedback

Higher order approximations for the single server queue with splitting, merging and feedback European Journal of Operational Research 4 () 447±467 www.elsevier.com/locate/dsw Theory and Methodology Higher order approximations for the single server queue with splitting, merging and feedback Muckai

More information

A Robust Queueing Network Analyzer Based on Indices of Dispersion

A Robust Queueing Network Analyzer Based on Indices of Dispersion A Robust Queueing Network Analyzer Based on Indices of Dispersion Wei You (joint work with Ward Whitt) Columbia University INFORMS 2018, Phoenix November 6, 2018 1/20 Motivation Many complex service systems

More information

TOWARDS BETTER MULTI-CLASS PARAMETRIC-DECOMPOSITION APPROXIMATIONS FOR OPEN QUEUEING NETWORKS

TOWARDS BETTER MULTI-CLASS PARAMETRIC-DECOMPOSITION APPROXIMATIONS FOR OPEN QUEUEING NETWORKS TOWARDS BETTER MULTI-CLASS PARAMETRIC-DECOMPOSITION APPROXIMATIONS FOR OPEN QUEUEING NETWORKS by Ward Whitt AT&T Bell Laboratories Murray Hill, NJ 07974-0636 March 31, 199 Revision: November 9, 199 ABSTRACT

More information

THE HEAVY-TRAFFIC BOTTLENECK PHENOMENON IN OPEN QUEUEING NETWORKS. S. Suresh and W. Whitt AT&T Bell Laboratories Murray Hill, New Jersey 07974

THE HEAVY-TRAFFIC BOTTLENECK PHENOMENON IN OPEN QUEUEING NETWORKS. S. Suresh and W. Whitt AT&T Bell Laboratories Murray Hill, New Jersey 07974 THE HEAVY-TRAFFIC BOTTLENECK PHENOMENON IN OPEN QUEUEING NETWORKS by S. Suresh and W. Whitt AT&T Bell Laboratories Murray Hill, New Jersey 07974 ABSTRACT This note describes a simulation experiment involving

More information

Queueing Theory II. Summary. ! M/M/1 Output process. ! Networks of Queue! Method of Stages. ! General Distributions

Queueing Theory II. Summary. ! M/M/1 Output process. ! Networks of Queue! Method of Stages. ! General Distributions Queueing Theory II Summary! M/M/1 Output process! Networks of Queue! Method of Stages " Erlang Distribution " Hyperexponential Distribution! General Distributions " Embedded Markov Chains M/M/1 Output

More information

A Queueing System with Queue Length Dependent Service Times, with Applications to Cell Discarding in ATM Networks

A Queueing System with Queue Length Dependent Service Times, with Applications to Cell Discarding in ATM Networks A Queueing System with Queue Length Dependent Service Times, with Applications to Cell Discarding in ATM Networks by Doo Il Choi, Charles Knessl and Charles Tier University of Illinois at Chicago 85 South

More information

On the Partitioning of Servers in Queueing Systems during Rush Hour

On the Partitioning of Servers in Queueing Systems during Rush Hour On the Partitioning of Servers in Queueing Systems during Rush Hour Bin Hu Saif Benjaafar Department of Operations and Management Science, Ross School of Business, University of Michigan at Ann Arbor,

More information

Waiting time characteristics in cyclic queues

Waiting time characteristics in cyclic queues Waiting time characteristics in cyclic queues Sanne R. Smits, Ivo Adan and Ton G. de Kok April 16, 2003 Abstract In this paper we study a single-server queue with FIFO service and cyclic interarrival and

More information

HEAVY-TRAFFIC EXTREME-VALUE LIMITS FOR QUEUES

HEAVY-TRAFFIC EXTREME-VALUE LIMITS FOR QUEUES HEAVY-TRAFFIC EXTREME-VALUE LIMITS FOR QUEUES by Peter W. Glynn Department of Operations Research Stanford University Stanford, CA 94305-4022 and Ward Whitt AT&T Bell Laboratories Murray Hill, NJ 07974-0636

More information

On the Partitioning of Servers in Queueing Systems during Rush Hour

On the Partitioning of Servers in Queueing Systems during Rush Hour On the Partitioning of Servers in Queueing Systems during Rush Hour This paper is motivated by two phenomena observed in many queueing systems in practice. The first is the partitioning of server capacity

More information

On Approximations for the GI/GI/1 Queue and Generalized Jackson Open Queueing Networks Using Indices of Dispersion

On Approximations for the GI/GI/1 Queue and Generalized Jackson Open Queueing Networks Using Indices of Dispersion On Approximations for the GI/GI/1 Queue and Generalized Jackson Open Queueing Networks Using Indices of Dispersion Ward Whitt and Wei You Department of Industrial Engineering and Operations Research, Columbia

More information

1 Basic concepts from probability theory

1 Basic concepts from probability theory Basic concepts from probability theory This chapter is devoted to some basic concepts from probability theory.. Random variable Random variables are denoted by capitals, X, Y, etc. The expected value or

More information

1. Introduction. Consider a single cell in a mobile phone system. A \call setup" is a request for achannel by an idle customer presently in the cell t

1. Introduction. Consider a single cell in a mobile phone system. A \call setup is a request for achannel by an idle customer presently in the cell t Heavy Trac Limit for a Mobile Phone System Loss Model Philip J. Fleming and Alexander Stolyar Motorola, Inc. Arlington Heights, IL Burton Simon Department of Mathematics University of Colorado at Denver

More information

Lecturer: Olga Galinina

Lecturer: Olga Galinina Renewal models Lecturer: Olga Galinina E-mail: olga.galinina@tut.fi Outline Reminder. Exponential models definition of renewal processes exponential interval distribution Erlang distribution hyperexponential

More information

J. MEDHI STOCHASTIC MODELS IN QUEUEING THEORY

J. MEDHI STOCHASTIC MODELS IN QUEUEING THEORY J. MEDHI STOCHASTIC MODELS IN QUEUEING THEORY SECOND EDITION ACADEMIC PRESS An imprint of Elsevier Science Amsterdam Boston London New York Oxford Paris San Diego San Francisco Singapore Sydney Tokyo Contents

More information

Queueing Theory I Summary! Little s Law! Queueing System Notation! Stationary Analysis of Elementary Queueing Systems " M/M/1 " M/M/m " M/M/1/K "

Queueing Theory I Summary! Little s Law! Queueing System Notation! Stationary Analysis of Elementary Queueing Systems  M/M/1  M/M/m  M/M/1/K Queueing Theory I Summary Little s Law Queueing System Notation Stationary Analysis of Elementary Queueing Systems " M/M/1 " M/M/m " M/M/1/K " Little s Law a(t): the process that counts the number of arrivals

More information

2 optimal prices the link is either underloaded or critically loaded; it is never overloaded. For the social welfare maximization problem we show that

2 optimal prices the link is either underloaded or critically loaded; it is never overloaded. For the social welfare maximization problem we show that 1 Pricing in a Large Single Link Loss System Costas A. Courcoubetis a and Martin I. Reiman b a ICS-FORTH and University of Crete Heraklion, Crete, Greece courcou@csi.forth.gr b Bell Labs, Lucent Technologies

More information

Queuing Networks: Burke s Theorem, Kleinrock s Approximation, and Jackson s Theorem. Wade Trappe

Queuing Networks: Burke s Theorem, Kleinrock s Approximation, and Jackson s Theorem. Wade Trappe Queuing Networks: Burke s Theorem, Kleinrock s Approximation, and Jackson s Theorem Wade Trappe Lecture Overview Network of Queues Introduction Queues in Tandem roduct Form Solutions Burke s Theorem What

More information

Class 11 Non-Parametric Models of a Service System; GI/GI/1, GI/GI/n: Exact & Approximate Analysis.

Class 11 Non-Parametric Models of a Service System; GI/GI/1, GI/GI/n: Exact & Approximate Analysis. Service Engineering Class 11 Non-Parametric Models of a Service System; GI/GI/1, GI/GI/n: Exact & Approximate Analysis. G/G/1 Queue: Virtual Waiting Time (Unfinished Work). GI/GI/1: Lindley s Equations

More information

1. Introduction To understand the motivation for this paper, it will be useful to consider the simple queueing network pictured in Figure 1. The netwo

1. Introduction To understand the motivation for this paper, it will be useful to consider the simple queueing network pictured in Figure 1. The netwo Steady-State Analysis of RBM in a Rectangle: Numerical Methods and a Queueing Application J. G. Dai School of Mathematics and ISyE Georgia Institute of Technology, Atlanta, GA 30332, USA J. M. Harrison

More information

P (L d k = n). P (L(t) = n),

P (L d k = n). P (L(t) = n), 4 M/G/1 queue In the M/G/1 queue customers arrive according to a Poisson process with rate λ and they are treated in order of arrival The service times are independent and identically distributed with

More information

Tales of Time Scales. Ward Whitt AT&T Labs Research Florham Park, NJ

Tales of Time Scales. Ward Whitt AT&T Labs Research Florham Park, NJ Tales of Time Scales Ward Whitt AT&T Labs Research Florham Park, NJ New Book Stochastic-Process Limits An Introduction to Stochastic-Process Limits and Their Application to Queues Springer 2001 I won t

More information

Economy of Scale in Multiserver Service Systems: A Retrospective. Ward Whitt. IEOR Department. Columbia University

Economy of Scale in Multiserver Service Systems: A Retrospective. Ward Whitt. IEOR Department. Columbia University Economy of Scale in Multiserver Service Systems: A Retrospective Ward Whitt IEOR Department Columbia University Ancient Relics A. K. Erlang (1924) On the rational determination of the number of circuits.

More information

Link Models for Circuit Switching

Link Models for Circuit Switching Link Models for Circuit Switching The basis of traffic engineering for telecommunication networks is the Erlang loss function. It basically allows us to determine the amount of telephone traffic that can

More information

7 Variance Reduction Techniques

7 Variance Reduction Techniques 7 Variance Reduction Techniques In a simulation study, we are interested in one or more performance measures for some stochastic model. For example, we want to determine the long-run average waiting time,

More information

A PARAMETRIC DECOMPOSITION BASED APPROACH FOR MULTI-CLASS CLOSED QUEUING NETWORKS WITH SYNCHRONIZATION STATIONS

A PARAMETRIC DECOMPOSITION BASED APPROACH FOR MULTI-CLASS CLOSED QUEUING NETWORKS WITH SYNCHRONIZATION STATIONS A PARAMETRIC DECOMPOSITION BASED APPROACH FOR MULTI-CLASS CLOSED QUEUING NETWORKS WITH SYNCHRONIZATION STATIONS Kumar Satyam and Ananth Krishnamurthy Department of Decision Sciences and Engineering Systems,

More information

Probability Models in Electrical and Computer Engineering Mathematical models as tools in analysis and design Deterministic models Probability models

Probability Models in Electrical and Computer Engineering Mathematical models as tools in analysis and design Deterministic models Probability models Probability Models in Electrical and Computer Engineering Mathematical models as tools in analysis and design Deterministic models Probability models Statistical regularity Properties of relative frequency

More information

Multiplicative Multifractal Modeling of. Long-Range-Dependent (LRD) Trac in. Computer Communications Networks. Jianbo Gao and Izhak Rubin

Multiplicative Multifractal Modeling of. Long-Range-Dependent (LRD) Trac in. Computer Communications Networks. Jianbo Gao and Izhak Rubin Multiplicative Multifractal Modeling of Long-Range-Dependent (LRD) Trac in Computer Communications Networks Jianbo Gao and Izhak Rubin Electrical Engineering Department, University of California, Los Angeles

More information

QUEUING SYSTEM. Yetunde Folajimi, PhD

QUEUING SYSTEM. Yetunde Folajimi, PhD QUEUING SYSTEM Yetunde Folajimi, PhD Part 2 Queuing Models Queueing models are constructed so that queue lengths and waiting times can be predicted They help us to understand and quantify the effect of

More information

Part I Stochastic variables and Markov chains

Part I Stochastic variables and Markov chains Part I Stochastic variables and Markov chains Random variables describe the behaviour of a phenomenon independent of any specific sample space Distribution function (cdf, cumulative distribution function)

More information

[4] T. I. Seidman, \\First Come First Serve" is Unstable!," tech. rep., University of Maryland Baltimore County, 1993.

[4] T. I. Seidman, \\First Come First Serve is Unstable!, tech. rep., University of Maryland Baltimore County, 1993. [2] C. J. Chase and P. J. Ramadge, \On real-time scheduling policies for exible manufacturing systems," IEEE Trans. Automat. Control, vol. AC-37, pp. 491{496, April 1992. [3] S. H. Lu and P. R. Kumar,

More information

THE INTERCHANGEABILITY OF./M/1 QUEUES IN SERIES. 1. Introduction

THE INTERCHANGEABILITY OF./M/1 QUEUES IN SERIES. 1. Introduction THE INTERCHANGEABILITY OF./M/1 QUEUES IN SERIES J. Appl. Prob. 16, 690-695 (1979) Printed in Israel? Applied Probability Trust 1979 RICHARD R. WEBER,* University of Cambridge Abstract A series of queues

More information

Sensitivity Analysis for Discrete-Time Randomized Service Priority Queues

Sensitivity Analysis for Discrete-Time Randomized Service Priority Queues Sensitivity Analysis for Discrete-Time Randomized Service Priority Queues George Kesidis 1, Takis Konstantopoulos 2, Michael Zazanis 3 1. Elec. & Comp. Eng. Dept, University of Waterloo, Waterloo, ON,

More information

Classical Queueing Models.

Classical Queueing Models. Sergey Zeltyn January 2005 STAT 99. Service Engineering. The Wharton School. University of Pennsylvania. Based on: Classical Queueing Models. Mandelbaum A. Service Engineering course, Technion. http://iew3.technion.ac.il/serveng2005w

More information

IEOR 8100: Topics in OR: Asymptotic Methods in Queueing Theory. Fall 2009, Professor Whitt. Class Lecture Notes: Wednesday, September 9.

IEOR 8100: Topics in OR: Asymptotic Methods in Queueing Theory. Fall 2009, Professor Whitt. Class Lecture Notes: Wednesday, September 9. IEOR 8100: Topics in OR: Asymptotic Methods in Queueing Theory Fall 2009, Professor Whitt Class Lecture Notes: Wednesday, September 9. Heavy-Traffic Limits for the GI/G/1 Queue 1. The GI/G/1 Queue We will

More information

Advanced Queueing Theory

Advanced Queueing Theory Advanced Queueing Theory 1 Networks of queues (reversibility, output theorem, tandem networks, partial balance, product-form distribution, blocking, insensitivity, BCMP networks, mean-value analysis, Norton's

More information

Robust Queueing Theory

Robust Queueing Theory Submitted to Operations Research manuscript Please, provide the mansucript number! Robust Queueing Theory Chaithanya Bandi*, Dimitris Bertsimas, Nataly Youssef We propose an alternative approach for studying

More information

A STAFFING ALGORITHM FOR CALL CENTERS WITH SKILL-BASED ROUTING: SUPPLEMENTARY MATERIAL

A STAFFING ALGORITHM FOR CALL CENTERS WITH SKILL-BASED ROUTING: SUPPLEMENTARY MATERIAL A STAFFING ALGORITHM FOR CALL CENTERS WITH SKILL-BASED ROUTING: SUPPLEMENTARY MATERIAL by Rodney B. Wallace IBM and The George Washington University rodney.wallace@us.ibm.com Ward Whitt Columbia University

More information

Buzen s algorithm. Cyclic network Extension of Jackson networks

Buzen s algorithm. Cyclic network Extension of Jackson networks Outline Buzen s algorithm Mean value analysis for Jackson networks Cyclic network Extension of Jackson networks BCMP network 1 Marginal Distributions based on Buzen s algorithm With Buzen s algorithm,

More information

LIMITS FOR QUEUES AS THE WAITING ROOM GROWS. Bell Communications Research AT&T Bell Laboratories Red Bank, NJ Murray Hill, NJ 07974

LIMITS FOR QUEUES AS THE WAITING ROOM GROWS. Bell Communications Research AT&T Bell Laboratories Red Bank, NJ Murray Hill, NJ 07974 LIMITS FOR QUEUES AS THE WAITING ROOM GROWS by Daniel P. Heyman Ward Whitt Bell Communications Research AT&T Bell Laboratories Red Bank, NJ 07701 Murray Hill, NJ 07974 May 11, 1988 ABSTRACT We study the

More information

S. T. Enns Paul Rogers. Dept. of Mechanical and Manufacturing Engineering University of Calgary Calgary, AB., T2N-1N4, CANADA

S. T. Enns Paul Rogers. Dept. of Mechanical and Manufacturing Engineering University of Calgary Calgary, AB., T2N-1N4, CANADA Proceedings of the 2008 Winter Simulation Conference S. J. Mason, R. R. Hill, L. Mönch, O. Rose, T. Jefferson, J. W. Fowler eds. CLARIFYING CONWIP VERSUS PUSH SYSTEM BEHAVIOR USING SIMULATION S. T. Enns

More information

Glossary availability cellular manufacturing closed queueing network coefficient of variation (CV) conditional probability CONWIP

Glossary availability cellular manufacturing closed queueing network coefficient of variation (CV) conditional probability CONWIP Glossary availability The long-run average fraction of time that the processor is available for processing jobs, denoted by a (p. 113). cellular manufacturing The concept of organizing the factory into

More information

VARIANCE REDUCTION IN SIMULATIONS OF LOSS MODELS

VARIANCE REDUCTION IN SIMULATIONS OF LOSS MODELS VARIANCE REDUCTION IN SIMULATIONS OF LOSS MODELS by Rayadurgam Srikant 1 and Ward Whitt 2 October 20, 1995 Revision: September 26, 1996 1 Coordinated Science Laboratory, University of Illinois, 1308 W.

More information

ON THE LAW OF THE i TH WAITING TIME INABUSYPERIODOFG/M/c QUEUES

ON THE LAW OF THE i TH WAITING TIME INABUSYPERIODOFG/M/c QUEUES Probability in the Engineering and Informational Sciences, 22, 2008, 75 80. Printed in the U.S.A. DOI: 10.1017/S0269964808000053 ON THE LAW OF THE i TH WAITING TIME INABUSYPERIODOFG/M/c QUEUES OPHER BARON

More information

Queueing systems. Renato Lo Cigno. Simulation and Performance Evaluation Queueing systems - Renato Lo Cigno 1

Queueing systems. Renato Lo Cigno. Simulation and Performance Evaluation Queueing systems - Renato Lo Cigno 1 Queueing systems Renato Lo Cigno Simulation and Performance Evaluation 2014-15 Queueing systems - Renato Lo Cigno 1 Queues A Birth-Death process is well modeled by a queue Indeed queues can be used to

More information

CDA5530: Performance Models of Computers and Networks. Chapter 4: Elementary Queuing Theory

CDA5530: Performance Models of Computers and Networks. Chapter 4: Elementary Queuing Theory CDA5530: Performance Models of Computers and Networks Chapter 4: Elementary Queuing Theory Definition Queuing system: a buffer (waiting room), service facility (one or more servers) a scheduling policy

More information

Environment (E) IBP IBP IBP 2 N 2 N. server. System (S) Adapter (A) ACV

Environment (E) IBP IBP IBP 2 N 2 N. server. System (S) Adapter (A) ACV The Adaptive Cross Validation Method - applied to polling schemes Anders Svensson and Johan M Karlsson Department of Communication Systems Lund Institute of Technology P. O. Box 118, 22100 Lund, Sweden

More information

Submitted to IEEE Transactions on Computers, June Evaluating Dynamic Failure Probability for Streams with. (m; k)-firm Deadlines

Submitted to IEEE Transactions on Computers, June Evaluating Dynamic Failure Probability for Streams with. (m; k)-firm Deadlines Submitted to IEEE Transactions on Computers, June 1994 Evaluating Dynamic Failure Probability for Streams with (m; k)-firm Deadlines Moncef Hamdaoui and Parameswaran Ramanathan Department of Electrical

More information

INDEX. production, see Applications, manufacturing

INDEX. production, see Applications, manufacturing INDEX Absorbing barriers, 103 Ample service, see Service, ample Analyticity, of generating functions, 100, 127 Anderson Darling (AD) test, 411 Aperiodic state, 37 Applications, 2, 3 aircraft, 3 airline

More information

GI/M/1 and GI/M/m queuing systems

GI/M/1 and GI/M/m queuing systems GI/M/1 and GI/M/m queuing systems Dmitri A. Moltchanov moltchan@cs.tut.fi http://www.cs.tut.fi/kurssit/tlt-2716/ OUTLINE: GI/M/1 queuing system; Methods of analysis; Imbedded Markov chain approach; Waiting

More information

ENGINEERING SOLUTION OF A BASIC CALL-CENTER MODEL

ENGINEERING SOLUTION OF A BASIC CALL-CENTER MODEL ENGINEERING SOLUTION OF A BASIC CALL-CENTER MODEL by Ward Whitt Department of Industrial Engineering and Operations Research Columbia University, New York, NY 10027 Abstract An algorithm is developed to

More information

Stochastic Models of Manufacturing Systems

Stochastic Models of Manufacturing Systems Stochastic Models of Manufacturing Systems Ivo Adan Systems 2/49 Continuous systems State changes continuously in time (e.g., in chemical applications) Discrete systems State is observed at fixed regular

More information

Stochastic process. X, a series of random variables indexed by t

Stochastic process. X, a series of random variables indexed by t Stochastic process X, a series of random variables indexed by t X={X(t), t 0} is a continuous time stochastic process X={X(t), t=0,1, } is a discrete time stochastic process X(t) is the state at time t,

More information

Elementary queueing system

Elementary queueing system Elementary queueing system Kendall notation Little s Law PASTA theorem Basics of M/M/1 queue M/M/1 with preemptive-resume priority M/M/1 with non-preemptive priority 1 History of queueing theory An old

More information

There recently has been great interest in the performance of queues

There recently has been great interest in the performance of queues Queueing Systems 0 (2000)?{? 1 The Impact of a Heavy-Tailed Service-Time Distribution upon the M/GI/s Waiting-Time Distribution Ward Whitt a a AT&T Labs, Shannon Laboratory, 180 Park Avenue, Florham Park,

More information

Operations Research Letters. Instability of FIFO in a simple queueing system with arbitrarily low loads

Operations Research Letters. Instability of FIFO in a simple queueing system with arbitrarily low loads Operations Research Letters 37 (2009) 312 316 Contents lists available at ScienceDirect Operations Research Letters journal homepage: www.elsevier.com/locate/orl Instability of FIFO in a simple queueing

More information

Lecture 7: Simulation of Markov Processes. Pasi Lassila Department of Communications and Networking

Lecture 7: Simulation of Markov Processes. Pasi Lassila Department of Communications and Networking Lecture 7: Simulation of Markov Processes Pasi Lassila Department of Communications and Networking Contents Markov processes theory recap Elementary queuing models for data networks Simulation of Markov

More information

2WB05 Simulation Lecture 7: Output analysis

2WB05 Simulation Lecture 7: Output analysis 2WB05 Simulation Lecture 7: Output analysis Marko Boon http://www.win.tue.nl/courses/2wb05 December 17, 2012 Outline 2/33 Output analysis of a simulation Confidence intervals Warm-up interval Common random

More information

On Queues with Inter-Arrival Times Proportional to Service Times

On Queues with Inter-Arrival Times Proportional to Service Times On Queues with Inter-Arrival Times Proportional to Service Times Israel Cidon* Department of Electrical Engineering Technion, Haifa 32000, Israel Asad Khamisyt Department of Electrical Engineering Technion,

More information

Contents LIST OF TABLES... LIST OF FIGURES... xvii. LIST OF LISTINGS... xxi PREFACE. ...xxiii

Contents LIST OF TABLES... LIST OF FIGURES... xvii. LIST OF LISTINGS... xxi PREFACE. ...xxiii LIST OF TABLES... xv LIST OF FIGURES... xvii LIST OF LISTINGS... xxi PREFACE...xxiii CHAPTER 1. PERFORMANCE EVALUATION... 1 1.1. Performance evaluation... 1 1.2. Performance versus resources provisioning...

More information

Pooling in tandem queueing networks with non-collaborative servers

Pooling in tandem queueing networks with non-collaborative servers DOI 10.1007/s11134-017-9543-0 Pooling in tandem queueing networks with non-collaborative servers Nilay Tanık Argon 1 Sigrún Andradóttir 2 Received: 22 September 2016 / Revised: 26 June 2017 Springer Science+Business

More information

A FAST MATRIX-ANALYTIC APPROXIMATION FOR THE TWO CLASS GI/G/1 NON-PREEMPTIVE PRIORITY QUEUE

A FAST MATRIX-ANALYTIC APPROXIMATION FOR THE TWO CLASS GI/G/1 NON-PREEMPTIVE PRIORITY QUEUE A FAST MATRIX-ANAYTIC APPROXIMATION FOR TE TWO CASS GI/G/ NON-PREEMPTIVE PRIORITY QUEUE Gábor orváth Department of Telecommunication Budapest University of Technology and Economics. Budapest Pf. 9., ungary

More information

Contents Preface The Exponential Distribution and the Poisson Process Introduction to Renewal Theory

Contents Preface The Exponential Distribution and the Poisson Process Introduction to Renewal Theory Contents Preface... v 1 The Exponential Distribution and the Poisson Process... 1 1.1 Introduction... 1 1.2 The Density, the Distribution, the Tail, and the Hazard Functions... 2 1.2.1 The Hazard Function

More information

PERFORMANCE ANALYSIS OF MULTI-SERVER TANDEM QUEUES WITH FINITE BUFFERS AND BLOCKING

PERFORMANCE ANALYSIS OF MULTI-SERVER TANDEM QUEUES WITH FINITE BUFFERS AND BLOCKING PERFORMANCE ANALYSIS OF MULTI-SERVER TANDEM QUEUES WITH FINITE BUFFERS AND BLOCKING Marcel van Vuuren a, Ivo J.B.F. Adan a and Simone A. E. Resing-Sassen b a Eindhoven University of Technology, P.O. Box

More information

BIRTH DEATH PROCESSES AND QUEUEING SYSTEMS

BIRTH DEATH PROCESSES AND QUEUEING SYSTEMS BIRTH DEATH PROCESSES AND QUEUEING SYSTEMS Andrea Bobbio Anno Accademico 999-2000 Queueing Systems 2 Notation for Queueing Systems /λ mean time between arrivals S = /µ ρ = λ/µ N mean service time traffic

More information

Computer Systems Modelling

Computer Systems Modelling Computer Systems Modelling Computer Laboratory Computer Science Tripos, Part II Lent Term 2010/11 R. J. Gibbens Problem sheet William Gates Building 15 JJ Thomson Avenue Cambridge CB3 0FD http://www.cl.cam.ac.uk/

More information

Modelling and simulating non-stationary arrival processes to facilitate analysis

Modelling and simulating non-stationary arrival processes to facilitate analysis Journal of Simulation (211) 5, 3 8 r 211 Operational Research Society Ltd. All rights reserved. 1747-7778/11 www.palgrave-journals.com/jos/ Modelling and simulating non-stationary arrival processes to

More information

Networking = Plumbing. Queueing Analysis: I. Last Lecture. Lecture Outline. Jeremiah Deng. 29 July 2013

Networking = Plumbing. Queueing Analysis: I. Last Lecture. Lecture Outline. Jeremiah Deng. 29 July 2013 Networking = Plumbing TELE302 Lecture 7 Queueing Analysis: I Jeremiah Deng University of Otago 29 July 2013 Jeremiah Deng (University of Otago) TELE302 Lecture 7 29 July 2013 1 / 33 Lecture Outline Jeremiah

More information

Kendall notation. PASTA theorem Basics of M/M/1 queue

Kendall notation. PASTA theorem Basics of M/M/1 queue Elementary queueing system Kendall notation Little s Law PASTA theorem Basics of M/M/1 queue 1 History of queueing theory An old research area Started in 1909, by Agner Erlang (to model the Copenhagen

More information

Inequality Comparisons and Traffic Smoothing in Multi-Stage ATM Multiplexers

Inequality Comparisons and Traffic Smoothing in Multi-Stage ATM Multiplexers IEEE Proceedings of the International Conference on Communications, 2000 Inequality Comparisons and raffic Smoothing in Multi-Stage AM Multiplexers Michael J. Neely MI -- LIDS mjneely@mit.edu Abstract

More information

Performance Evaluation of Queuing Systems

Performance Evaluation of Queuing Systems Performance Evaluation of Queuing Systems Introduction to Queuing Systems System Performance Measures & Little s Law Equilibrium Solution of Birth-Death Processes Analysis of Single-Station Queuing Systems

More information

M/G/1 and M/G/1/K systems

M/G/1 and M/G/1/K systems M/G/1 and M/G/1/K systems Dmitri A. Moltchanov dmitri.moltchanov@tut.fi http://www.cs.tut.fi/kurssit/elt-53606/ OUTLINE: Description of M/G/1 system; Methods of analysis; Residual life approach; Imbedded

More information

Introduction to Queueing Theory with Applications to Air Transportation Systems

Introduction to Queueing Theory with Applications to Air Transportation Systems Introduction to Queueing Theory with Applications to Air Transportation Systems John Shortle George Mason University February 28, 2018 Outline Why stochastic models matter M/M/1 queue Little s law Priority

More information

Exercises Stochastic Performance Modelling. Hamilton Institute, Summer 2010

Exercises Stochastic Performance Modelling. Hamilton Institute, Summer 2010 Exercises Stochastic Performance Modelling Hamilton Institute, Summer Instruction Exercise Let X be a non-negative random variable with E[X ]

More information

Completion Time in Dynamic PERT Networks 57 job are nished, as well as that the associated service station has processed the same activity of the prev

Completion Time in Dynamic PERT Networks 57 job are nished, as well as that the associated service station has processed the same activity of the prev Scientia Iranica, Vol. 14, No. 1, pp 56{63 c Sharif University of Technology, February 2007 Project Completion Time in Dynamic PERT Networks with Generating Projects A. Azaron 1 and M. Modarres In this

More information

6 Solving Queueing Models

6 Solving Queueing Models 6 Solving Queueing Models 6.1 Introduction In this note we look at the solution of systems of queues, starting with simple isolated queues. The benefits of using predefined, easily classified queues will

More information

Periodic Load Balancing

Periodic Load Balancing Queueing Systems 0 (2000)?? 1 Periodic Load Balancing Gisli Hjálmtýsson a Ward Whitt a a AT&T Labs, 180 Park Avenue, Building 103, Florham Park, NJ 07932-0971 E-mail: {gisli,wow}@research.att.com Queueing

More information

Computer Networks More general queuing systems

Computer Networks More general queuing systems Computer Networks More general queuing systems Saad Mneimneh Computer Science Hunter College of CUNY New York M/G/ Introduction We now consider a queuing system where the customer service times have a

More information

HITTING TIME IN AN ERLANG LOSS SYSTEM

HITTING TIME IN AN ERLANG LOSS SYSTEM Probability in the Engineering and Informational Sciences, 16, 2002, 167 184+ Printed in the U+S+A+ HITTING TIME IN AN ERLANG LOSS SYSTEM SHELDON M. ROSS Department of Industrial Engineering and Operations

More information

Introduction to Queueing Theory

Introduction to Queueing Theory Introduction to Queueing Theory Raj Jain Washington University in Saint Louis Saint Louis, MO 63130 Jain@cse.wustl.edu Audio/Video recordings of this lecture are available at: 30-1 Overview Queueing Notation

More information

Stability, Queue Length and Delay of Deterministic and Stochastic Queueing Networks Cheng-Shang Chang IBM Research Division T.J. Watson Research Cente

Stability, Queue Length and Delay of Deterministic and Stochastic Queueing Networks Cheng-Shang Chang IBM Research Division T.J. Watson Research Cente Stability, Queue Length and Delay of Deterministic and Stochastic Queueing Networks Cheng-Shang Chang IBM Research Division T.J. Watson Research Center P.O. Box 704 Yorktown Heights, NY 10598 cschang@watson.ibm.com

More information

Chapter 8 Queuing Theory Roanna Gee. W = average number of time a customer spends in the system.

Chapter 8 Queuing Theory Roanna Gee. W = average number of time a customer spends in the system. 8. Preliminaries L, L Q, W, W Q L = average number of customers in the system. L Q = average number of customers waiting in queue. W = average number of time a customer spends in the system. W Q = average

More information

Queueing Review. Christos Alexopoulos and Dave Goldsman 10/25/17. (mostly from BCNN) Georgia Institute of Technology, Atlanta, GA, USA

Queueing Review. Christos Alexopoulos and Dave Goldsman 10/25/17. (mostly from BCNN) Georgia Institute of Technology, Atlanta, GA, USA 1 / 26 Queueing Review (mostly from BCNN) Christos Alexopoulos and Dave Goldsman Georgia Institute of Technology, Atlanta, GA, USA 10/25/17 2 / 26 Outline 1 Introduction 2 Queueing Notation 3 Transient

More information

Lecture Notes of Communication Network I y: part 5. Lijun Qian. Rutgers The State University of New Jersey. 94 Brett Road. Piscataway, NJ

Lecture Notes of Communication Network I y: part 5. Lijun Qian. Rutgers The State University of New Jersey. 94 Brett Road. Piscataway, NJ Lecture Notes of Communication Network I y: part 5 Lijun Qian Department of Electrical Engineering Rutgers The State University of New Jersey 94 Brett Road Piscataway, NJ 08854-8058 October 6, 1998 Abstract

More information

OPEN MULTICLASS HL QUEUEING NETWORKS: PROGRESS AND SURPRISES OF THE PAST 15 YEARS. w 1. v 2. v 3. Ruth J. Williams University of California, San Diego

OPEN MULTICLASS HL QUEUEING NETWORKS: PROGRESS AND SURPRISES OF THE PAST 15 YEARS. w 1. v 2. v 3. Ruth J. Williams University of California, San Diego OPEN MULTICLASS HL QUEUEING NETWORKS: PROGRESS AND SURPRISES OF THE PAST 15 YEARS v 2 w3 v1 w 2 w 1 v 3 Ruth J. Williams University of California, San Diego 1 PERSPECTIVE MQN SPN Sufficient conditions

More information

Fast Evaluation of Ensemble Transients of Large IP Networks. University of Maryland, College Park CS-TR May 11, 1998.

Fast Evaluation of Ensemble Transients of Large IP Networks. University of Maryland, College Park CS-TR May 11, 1998. Fast Evaluation of Ensemble Transients of Large IP Networks Catalin T. Popescu cpopescu@cs.umd.edu A. Udaya Shankar shankar@cs.umd.edu Department of Computer Science University of Maryland, College Park

More information

Average Reward Parameters

Average Reward Parameters Simulation-Based Optimization of Markov Reward Processes: Implementation Issues Peter Marbach 2 John N. Tsitsiklis 3 Abstract We consider discrete time, nite state space Markov reward processes which depend

More information

A tutorial on some new methods for. performance evaluation of queueing networks. P. R. Kumar. Coordinated Science Laboratory. University of Illinois

A tutorial on some new methods for. performance evaluation of queueing networks. P. R. Kumar. Coordinated Science Laboratory. University of Illinois A tutorial on some new methods for performance evaluation of queueing networks P. R. Kumar Dept. of Electrical and Computer Engineering, and Coordinated Science Laboratory University of Illinois 1308 West

More information

Jitter Analysis of an MMPP 2 Tagged Stream in the presence of an MMPP 2 Background Stream

Jitter Analysis of an MMPP 2 Tagged Stream in the presence of an MMPP 2 Background Stream Jitter Analysis of an MMPP 2 Tagged Stream in the presence of an MMPP 2 Background Stream G Geleji IBM Corp Hursley Park, Hursley, UK H Perros Department of Computer Science North Carolina State University

More information

Queueing Review. Christos Alexopoulos and Dave Goldsman 10/6/16. (mostly from BCNN) Georgia Institute of Technology, Atlanta, GA, USA

Queueing Review. Christos Alexopoulos and Dave Goldsman 10/6/16. (mostly from BCNN) Georgia Institute of Technology, Atlanta, GA, USA 1 / 24 Queueing Review (mostly from BCNN) Christos Alexopoulos and Dave Goldsman Georgia Institute of Technology, Atlanta, GA, USA 10/6/16 2 / 24 Outline 1 Introduction 2 Queueing Notation 3 Transient

More information

Stochastic Models 3 (1998)

Stochastic Models 3 (1998) CALCULATING TRANSIENT CHARACTERISTICS OF THE ERLANG LOSS MODEL BY NUMERICAL TRANSFORM INVERSION J. ABATE W. WHITT 900 Hammond Road AT&T Labs Research Ridgewood, NJ 07450-2908 Room 2C-178 Murray Hill, NJ

More information

queue KTH, Royal Institute of Technology, Department of Microelectronics and Information Technology

queue KTH, Royal Institute of Technology, Department of Microelectronics and Information Technology Analysis of the Packet oss Process in an MMPP+M/M/1/K queue György Dán, Viktória Fodor KTH, Royal Institute of Technology, Department of Microelectronics and Information Technology {gyuri,viktoria}@imit.kth.se

More information

ECE 3511: Communications Networks Theory and Analysis. Fall Quarter Instructor: Prof. A. Bruce McDonald. Lecture Topic

ECE 3511: Communications Networks Theory and Analysis. Fall Quarter Instructor: Prof. A. Bruce McDonald. Lecture Topic ECE 3511: Communications Networks Theory and Analysis Fall Quarter 2002 Instructor: Prof. A. Bruce McDonald Lecture Topic Introductory Analysis of M/G/1 Queueing Systems Module Number One Steady-State

More information

Introduction to queuing theory

Introduction to queuing theory Introduction to queuing theory Queu(e)ing theory Queu(e)ing theory is the branch of mathematics devoted to how objects (packets in a network, people in a bank, processes in a CPU etc etc) join and leave

More information

Queuing Theory. Using the Math. Management Science

Queuing Theory. Using the Math. Management Science Queuing Theory Using the Math 1 Markov Processes (Chains) A process consisting of a countable sequence of stages, that can be judged at each stage to fall into future states independent of how the process

More information

Derivation of Formulas by Queueing Theory

Derivation of Formulas by Queueing Theory Appendices Spectrum Requirement Planning in Wireless Communications: Model and Methodology for IMT-Advanced E dite d by H. Takagi and B. H. Walke 2008 J ohn Wiley & Sons, L td. ISBN: 978-0-470-98647-9

More information

Solution: The process is a compound Poisson Process with E[N (t)] = λt/p by Wald's equation.

Solution: The process is a compound Poisson Process with E[N (t)] = λt/p by Wald's equation. Solutions Stochastic Processes and Simulation II, May 18, 217 Problem 1: Poisson Processes Let {N(t), t } be a homogeneous Poisson Process on (, ) with rate λ. Let {S i, i = 1, 2, } be the points of the

More information

Transient Analysis of a Series Configuration Queueing System

Transient Analysis of a Series Configuration Queueing System Transient Analysis of a Series Configuration Queueing System Yu-Li Tsai*, Daichi Yanagisawa, and Katsuhiro Nishinari Abstract In this paper, we consider the transient analysis of a popular series configuration

More information