EP2200 Queueing theory and teletraffic systems

Similar documents
To describe a queuing system, an input process and an output process has to be specified.

Control chart for waiting time in system of (M / M / S) :( / FCFS) Queuing model

Module 4: Time Response of discrete time systems Lecture Note 1

OPTIMIZATION OF INTERDEPENDENT QUEUEING MODEL WITH BULK SERVICE HAVING VARIABLE CAPACITY

Correction for Simple System Example and Notes on Laplace Transforms / Deviation Variables ECHE 550 Fall 2002

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

Lecture 2: The z-transform

Control Systems Engineering ( Chapter 7. Steady-State Errors ) Prof. Kwang-Chun Ho Tel: Fax:

IEOR 3106: Fall 2013, Professor Whitt Topics for Discussion: Tuesday, November 19 Alternating Renewal Processes and The Renewal Equation

Mathematical modeling of control systems. Laith Batarseh. Mathematical modeling of control systems

Given the following circuit with unknown initial capacitor voltage v(0): X(s) Immediately, we know that the transfer function H(s) is

Reliability Analysis of Embedded System with Different Modes of Failure Emphasizing Reboot Delay

e st t u(t 2) dt = lim t dt = T 2 2 e st = T e st lim + e st

page 1 This question-paper you hand in, together with your solutions. ================================================================

ME 375 FINAL EXAM Wednesday, May 6, 2009

Chapter 2 Sampling and Quantization. In order to investigate sampling and quantization, the difference between analog

LAPLACE TRANSFORM REVIEW SOLUTIONS

ME 375 FINAL EXAM SOLUTIONS Friday December 17, 2004

Stochastic Perishable Inventory Control in a Service Facility System Maintaining Inventory for Service: Semi Markov Decision Problem

The continuous time random walk (CTRW) was introduced by Montroll and Weiss 1.

Laplace Transformation

RELIABILITY OF REPAIRABLE k out of n: F SYSTEM HAVING DISCRETE REPAIR AND FAILURE TIMES DISTRIBUTIONS

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get

A BATCH-ARRIVAL QUEUE WITH MULTIPLE SERVERS AND FUZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH

Call Centers with a Postponed Callback Offer

Performance Evaluation

Digital Control System

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL =

Chapter #4 EEE8013. Linear Controller Design and State Space Analysis. Design of control system in state space using Matlab

Proactive Serving Decreases User Delay Exponentially: The Light-tailed Service Time Case

Uniform Distribution. Uniform Distribution. Uniform Distribution. Graphs of Gamma Distributions. Gamma Distribution. Continuous Distributions CD - 1

Chapter 13. Root Locus Introduction

Root Locus Diagram. Root loci: The portion of root locus when k assume positive values: that is 0

CHAPTER 4 DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL

March 18, 2014 Academic Year 2013/14

ME 375 EXAM #1 Tuesday February 21, 2006

Lecture 8: Period Finding: Simon s Problem over Z N

On the Robustness of the Characteristics Related to (M\M\1) ( \FCFS) Queue System Model

Homework #7 Solution. Solutions: ΔP L Δω. Fig. 1

Properties of Z-transform Transform 1 Linearity a

Digital Control System

4.6 M/M/s/s Loss Queueing Model

Nonlinear Single-Particle Dynamics in High Energy Accelerators

Calculation of the temperature of boundary layer beside wall with time-dependent heat transfer coefficient

Design of Digital Filters

Markov Processes. Fundamentals (1) State graph. Fundamentals (2) IEA Automation 2018

Chapter 2: Problem Solutions

These are practice problems for the final exam. You should attempt all of them, but turn in only the even-numbered problems!

Bogoliubov Transformation in Classical Mechanics

What lies between Δx E, which represents the steam valve, and ΔP M, which is the mechanical power into the synchronous machine?

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

To provide useful practical insight into the performance of service-oriented (non-revenue-generating) call

Introduction to Laplace Transform Techniques in Circuit Analysis

Math 201 Lecture 17: Discontinuous and Periodic Functions

Link Models for Circuit Switching

Lecture 9: Shor s Algorithm

Connectivity in large mobile ad-hoc networks

Lecture 4. Chapter 11 Nise. Controller Design via Frequency Response. G. Hovland 2004

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

Online Appendix for Managerial Attention and Worker Performance by Marina Halac and Andrea Prat

EE Control Systems LECTURE 6

Lecture 17: Analytic Functions and Integrals (See Chapter 14 in Boas)

Math 334 Fall 2011 Homework 10 Solutions

Question 1 Equivalent Circuits

Optimal revenue management in two class pre-emptive delay dependent Markovian queues

Spring 2014 EE 445S Real-Time Digital Signal Processing Laboratory. Homework #0 Solutions on Review of Signals and Systems Material

Name: Solutions Exam 3

SIMPLE LINEAR REGRESSION

Queueing Theory (Part 3)

ECE 3510 Root Locus Design Examples. PI To eliminate steady-state error (for constant inputs) & perfect rejection of constant disturbances

Midterm Test Nov 10, 2010 Student Number:

Solutions to homework #10

The Power Series Expansion on a Bulge Heaviside Step Function

Reading assignment: In this chapter we will cover Sections Definition and the Laplace transform of simple functions

Reading assignment: In this chapter we will cover Sections Definition and the Laplace transform of simple functions

Feedback Control Systems (FCS)

Math 273 Solutions to Review Problems for Exam 1

A Note on the Sum of Correlated Gamma Random Variables

From stochastic processes to numerical methods: A new scheme for solving reaction subdiffusion fractional partial differential equations.

NAME (pinyin/italian)... MATRICULATION NUMBER... SIGNATURE

SIMON FRASER UNIVERSITY School of Engineering Science ENSC 320 Electric Circuits II. R 4 := 100 kohm

Roadmap for Discrete-Time Signal Processing

SECTION x2 x > 0, t > 0, (8.19a)

Chapter 7. Root Locus Analysis

ME2142/ME2142E Feedback Control Systems

Non Markovian Queues (contd.)

Acceptance sampling uses sampling procedure to determine whether to

c n b n 0. c k 0 x b n < 1 b k b n = 0. } of integers between 0 and b 1 such that x = b k. b k c k c k

Factor Analysis with Poisson Output

SOLVING THE KONDO PROBLEM FOR COMPLEX MESOSCOPIC SYSTEMS

OVERFLOW PROBABILITY IN AN ATM QUEUE WITH SELF-SIMILAR INPUT TRAFFIC

CONGESTION control is a key functionality in modern

Chapter 4. The Laplace Transform Method

Strong approximations for Markovian service networks

Linear System Fundamentals

DYNAMIC MODELS FOR CONTROLLER DESIGN

POWER SYSTEM SMALL SIGNAL STABILITY ANALYSIS BASED ON TEST SIGNAL

Lecture #9 Continuous time filter

On the Stability Region of Congestion Control

CHEAP CONTROL PERFORMANCE LIMITATIONS OF INPUT CONSTRAINED LINEAR SYSTEMS

Transcription:

P00 Queueing theory and teletraffic ytem Lecture 9 M/G/ ytem Vitoria Fodor KTH S/LCN

The M/G/ queue Arrival proce memoryle (Poion( Service time general, identical, independent, f(x Single erver M/ r / and M/H r / are pecific cae, reult for M/G/ can be ued ule we can ue from the Marovian ytem x < for tability (ingle erver, no blocing Little: N T PASTA KTH S/LCN

The M/G/ queue ecall: M/M/: 0 μ μ At the arrival of the econd cutomer the time remaining from the ervice of the firt cutomer i till xp( M/G/: If we conider the ytem when a new cutomer arrive, then the remaining (reidual ervice time of the cutomer under ervice depend on the pat of the proce (on the elaped ervice time Conequently: the number of cutomer in the ytem doe not give a Marov chain KTH S/LCN 3

The M/G/ queue Solution method Average meaure N, T, etc. Mean value analyi Ditribution of the number of cutomer, waiting time, etc. Study the ytem at time point t 0, t, t,... when a cutomer depart, and extend for all point of time mbedded Marov chain KTH S/LCN 4

5 KTH S/LCN To calculate average meaure e tart with the average waiting time: the ervice of the waiting cutomer the remaining (or reidual ervice of the cutomer in the ervice unit Average remaining ervice time: Firt conditional average waiting time, then uncondition The M/G/ mean value analyi Pollacze-Khinchin mean formula ( ( (average include 0 remaining ervice time at tate 0 0, 0 (average waiting timefor cutomer arriving at tate,, 0,0,, 0 0,, q i i N r p p p p p

The M/G/ mean value analyi Pollacze-Khinchin mean formula e have to derive the average remaining ervice time : n: number of ervice in a large T number of Poion arrival: nt (ince the ytem i table T and n Note, have to include the 0 remaining ervice time at empty ytem. (t x x x x x 3 x 4 x 5 x n n n ( t i T n i ( i i Pollacze-Khinchin mean formula for the waiting time T t KTH S/LCN 6

7 KTH S/LCN From you can derive T, N, N q with Little theorem Comment: depend on the firt and the econd moment of the ervice time only Mean value increae with variance (cot of randomne The M/G/ mean value analyi Pollacze-Khinchin mean formula ( ( ( ( ( ( ( C x V V ( 0, /: / (, /: / C D M C M M x x

M/G/ waiting time ( ( C x 8 7 6 5 4 3 Hyper-xp, (C x 4 rlang-4 (C x /4 M/M/ 0. 0.4 0.6 0.8 lambda KTH S/LCN 8

Group wor: The M/G/ mean value analyi Pollacze-Khinchin mean formula Conider the following ytem: Single erver, infinite buffer Poion arrival proce, 0. cutomer per minute Service proce: um of two exponential tep, with mean time minute and minute. Calculate the mean waiting time x x ( ( ( C x KTH S/LCN 9

Ditribution of number of cutomer in the ytem *** Comment: called a queue-length in the Virtamo note! *** The number of cutomer, N t i not a Marov proce the reidual ervice time i not memoryle e can model the ytem at departure time and extend the reult to all point of time: in the cae of Poion arrival the ditribution of N at departure time i the ame a at arbitrary point of time (PASTA if we are lucy then N t follow a dicrete time Marov proce at departure time ince thi dicrete time Marov chain i rather complex, we can expre the tranform form (z-tranform of the ditribution of the number of cutomer in the ytem. KTH S/LCN 0

Ditribution of number of cutomer in the ytem In the cae of Poion arrival the ditribution of N at departure time i the ame a at arbitrary point of time (PASTA PASTA i proved for arrival intant however, departure intant ee the ame queue length ditribution Let u follow N, N, N, that i, the number of cutomer in the ytem after departure N : number of cutomer after the departure of a cutomer V : number of arrival during the ervice time of cutomer, b(x i the ervice time ditribution, then: N α i N V N N V N 0 depend only on N and V, V i independent from i Dicrete time Marov Proce P( V i ( x i! e x b( x dx KTH S/LCN

M/G/ number of cutomer in the ytem α α α 3 i- i i i α 0 α i-j j xpreing the teady tate of the Marov-chain decribing N, we get the z- tranform of the ditribution of N Pollacze-Khinchin tranform form: Q ( ( ( ( z z B z * * B ( z z where: and B*( i the Laplace tranform of the ervice time ditribution. (S*( in the Virtamo note Ditribution of N with invere tranform, or moment of the ditribution..g., M/M/ KTH S/LCN

M/G/ ytem time ditribution ithout proof: Pollacze-Khinchin tranform form for the ytem time and waiting time: * ( ( * B ( T * * ( B ( ( B * ( where: x and B*( i the Laplace tranform of the ervice time ditribution. (S*( in the Virtamo note.g., M/M/ ytem time KTH S/LCN 3

Group wor again: Conider the ytem: The M/G/ mean value analyi Pollacze-Khinchin mean formula Single erver, infinite buffer Poion arrival proce, 0. cutomer per minute Service proce: um of two exponential tep, with mean time minute and minute. Give the Laplace tranform of the waiting time, calculate the mean waiting time * ( ( * B ( KTH S/LCN 4

M/G/ equirement for the exam: Derive and ue P-K mean value formula Ue P-K tranform form typically: for given ervice time ditribution give the tranform form, calculate moment Do not forget: M/M/, M/D/, M/ r / and M/H r / are pecific cae of M/G/ KTH S/LCN 5