Queueing Theory

Frank Y. S. Lin
Information Management Dept. National Taiwan University yslin@im.ntu.edu.tw
1

References
Leonard Kleinrock, Queueing Systems Volume I: Theory, New York: Wiley, 1975-1976. D. Gross and C. M. Harris, Fundamentals of Queueing Theory, New York: Wiley, 1998.

Agenda
Introduction Stochastic Process General Concepts M/M/1 Model M/M/1/K Model Discouraged Arrivals M/M/ and M/M/m Models M/M/m/m Model
3

Introduction

Queueing System
A queueing system can be described as customers arriving for service, waiting for service if it is not immediate, and if having waited for service, leaving the system after being served.

Why Queueing Theory

Performance Measurement
Average waiting time of customer / distribution of waiting time. Average number of customers in the system / distribution of queue length / current work backlog. Measurement of the idle time of server / length of an idle period. Measurement of the busy time of server / length of a busy period. System utilization.
6

Why Queueing Theory (contd)

Delay Analysis Network Delay = Queueing Delay + Propagation Delay (depends on the distance) + Node Delay Processing Delay
(independent of packet length, e.g. header CRC check) Adapter Delay (constant)

Characteristics of Queueing Process

Arrival Pattern of Customers
Probability distribution Patient / impatient (balked) arrival Stationary / nonstationary

Service Patterns
Probability distribution State dependent / independent service Stationary / nonstationary

Characteristics of Queueing Process (contd)

Queueing Disciplines
First come, first served (FCFS) Last come, first served (LCFS) Random selection for service (RSS) Priority queue Preemptive / nonpreemptive

System Capacity
Finite / infinite waiting room.

Characteristics of Queueing Process (contd)

Number of Service Channels
Single channel / multiple channels Single queue / multiple queues

Stages of Service
Single stage (e.g. hair-styling salon) Multiple stages (e.g. manufacturing process) Process recycling or feedback

10

Notation
A queueing process is described by A/B/X/Y/Z

11

Notation (contd)
For example, M/D/2//FCFS indicates a queueing process with exponential inter-arrival time, deterministic service times, two parallel servers, infinite capacity, and first-come, firstserved queueing discipline. Y and Z can be omitted if Y = and Z = FCFS.

12

Stochastic Process

13

Stochastic Process
Stochastic process: any collection of random variables (t), t T, on a common probability space where t is a subset of time.
Continuous / discrete time stochastic process Example: (t) denotes the temperature in the class on t = 7:00, 8:00, 9:00, 10:00, (discrete time)

We can regard a stochastic process as a family of random variables which are indexed by time. For a random process X(t), the PDF is denoted by FX(x;t) = P[X(t) <= x]
14

Some Classifications of Stochastic Process

Stationary Processes: independent of time
FX (x; t + ) = FX (x; t)

Independent Processes: independent variables

FX (x; t) = FX1,, Xn(x1,, xn ; t1,,tn) = FX1(x1; t1) FXn(xn; tn)

Markov Processes: the probability of the next state

depends only upon the current state and not upon any previous states. P[X(tn+1) = xn+1 | X(tn) = xn, ., X(t1) = x1] = P[X(tn+1) = xn+1 | X(tn) = xn]
15

Some Classifications of Stochastic Process (contd)

Birth-death Processes: state transitions take place
between neighboring states only.

Random Walks: the next position the process occupies

is equal to the previous position plus a random variable whose value is drawn independently from an arbitrary distribution.

16

General Concepts

17

Continuous-time Memoryless Property

If X ~Exp(), for any a,b > 0, P[X > a + b | X > a] = P[X > b] Proof: P[X > a + b | X > a]
= P[( X > a + b ) I ( X > a )] P ( X > a)

( X > a + b) ( X > a)
e ( a +b ) = a = e b = P ( X > b) e
18

P ( X > a + b) P ( X > a)

1 Fx ( a + b ) 1 Fx ( a )

Global Balance Equation

Define Pi = P[system is in state i] Pij = P[get into state j right after leaving state i]

Pj Pij =
i =0 (i j )

i =0 (i j )

P
i

Pij

rate out of state j = rate into state j

19

General Balance Equation

Define S = a subset of the state space

j =0 ( js )

P P = P P
j i =0 ( is ) ij i =0 ( is ) i j =0 ( js )

ij

rate in = rate out

20

General Equilibrium Solution

Notation:
Pk = the probability that the system contains k customers (in state k)

P
k =0

=1

k= the arrival rate of customers when the system is in state k. k= the service rate when the system is in state k.

21

General Equilibrium Solution (contd)

Consider state k:
rate in = rate out Pk k = Pk +1 k +1
k-1

k-1
k

k
k+1

k Pk +1 = Pk k +1 k 1 Pk = Pk 1 k
. . .

k+1

k 1 k 2 0 Pk = P0 = k k 1 1

i P0 i =0 i +1
k 1

22

General Equilibrium Solution (contd)

P
k =0 k

=1

i P0 = 1 k = 0 i = 0 i +1
k 1

P0 =

i 1 + k = 0 i = 0 i +1
k 1

= Pk k , T =
k =0

waiting time w = T

23

Littles Result
N = average number of customers in the system

T = system time (service time + queueing time) = arrival rate

N = T
Black box Service time

Queueing time

System time T

24

M/M/1 Model
Single Server, Single Queue (The Classical Queueing System)

25

M/M/1 Queue
Single server, single queue, infinite population: k = k = Interarrival time distribution: p (t ) = e t Service time distribution
p (t < t0 ) = e t dt = 1 e t0
0 t0

Stability condition <

26

M/M/1 Queue (contd)

System utilization
= = P[system is busy], 1- = P[system is idel]

Define state Sn = n customers in the system (n-1 in the queue and 1 in service) S0 = empty system
rate out

S
rate in
27

M/M/1 Queue (contd)

Define pn = P[n customers in the system]
pn = pn +1 (rate in = rate out) pn +1 = pn = pn
pn +1 = n +1 p0

Since

pi = 1
i =0

p0 = 1
n i =0

p0 n = 1
i =0

p0 = 1 , pn = n (1 )
#
28

M/M/1 Queue (contd)

Average number of customers in the system
N = k (1 ) k = (1 ) k k
= (1 ) d k / d d = (1 ) k d d 1 = (1 ) d 1 = 1 N =

1
#
29

M/M/1 Queue (contd)

Average system time
T= N

(Littles Result)

1/ 1 = 1

k = (1 ) = (1 ) = k 1 i =k
i

P[ k customers in the system]

30

M/M/1/K Model
Single Server, Finite Storage

31

M/M/1/K Model
The system can hold at most a total of K customers (including the customer in service)
k = 0 k =

if k < K if k K

32

M/M/1/K Model (contd)

Pk = P0 = P0 i =0
k 1 k

kK
k>K

Pk = 0

P0 =

1 / k 1 + ( / ) = 1 ( / ) K +1 k =1
K

0k K
otherwise
33

Discouraged Arrivals

34

Discouraged Arrivals
Arrivals tend to get discouraged when more and more people are present in the system. k = k +1 k =

35

Discouraged Arrivals (contd)

/(i + 1) k 1 Pk = P0 = ( / ) P0 k! i =0
k 1

P0 =

1 1 + ( / )
k =1 k

1 k!

=e

Pk =

( / )
k!

N =

36

Discouraged Arrivals (contd)

( / ) ( / ) e = k Pk = k! k =0 k =0 k + 1

= 1 e ( / )

(Q = , = 1 P0 )

T=

/ (1 e / )

37

M/M/ and M/M/m

M/M/ - Infinite Servers, Single Queue (Responsive Servers) M/M/m - Multiple Servers, Single Queue (The m-Server Case)
38

M/M/ Queue
There is always a new server available for each arriving customer. k = k = k

39

M/M/ Queue (contd)

( / ) k / Pk = P0 = e k! i = 0 (i + 1)
k 1

N=
T = 1

(Littles Result)

40

M/M/m Queue
The M/M/m queue
An M/M/m queue is shorthand for a single queue served by multiple servers. Suppose there are m servers waiting for a single line. For each server, the waiting time for a queue is a system with service rate and arrival rate /m. The M/M/1 analysis has been done, at risk conclusion: 1 delay = / n throughput =
/n = n
41

M/M/m Queue (contd)

k = k = k if k m m if k > m

For k m For k > m

pk = p0 pk = p0

k 1 = p0 ( ) 2 k k! 1 1 = p0 ( ) k ( ) k n 2 n n n! n
42

M/M/n Queue (contd)

p
i =0 i

=1

1 p0 = n 1 (np ) k (np ) n 1 pi k ! + n ! (1 ) k =0

where = n

P[queueing] =

k =m

(/ ) n + p0 Total system time = 2 (n 1)!(n )

1
43

Comparisons (contd)
M/M/1 v.s M/M/4
If we have 4 M/M/1 systems: 4 parallel communication links that can each handle 50 pps (), arrival rate = 25 pps per queue. average delay = 40 ms. Whereas for an M/M/4 system, average delay = 21.7 ms.

44

Comparisons (contd)
Fast Server v.s A Set of Slow Servers #1
If we have an M/M/4 system with service rate =50 pps for each server, and another M/M/1 system with service rate 4 = 200 pps. Both arrival rate is = 100 pps delay for M/M/4 = 21.7 ms delay for M/M/1 = 10 ms

45

Comparisons (contd)
Fast Server v.s A Set of Slow Servers #2
If we have n M/M/1 system with service rate pps for each server, and another M/M/1 system with service rate n pps. Both arrival rate is n pps 1/ T1 = S2 1

S1

1/ n 1/ n = T2 = n 1 1 n T1 T2 = n

46

M/M/m/m
Multiple Servers, No Storage (m-Server Loss Systems)

47

M/M/m/m
There are available m servers, each newly arriving customers is given a server, if a customers arrives when all servers are occupied, that customer is lost e.g. telephony system. if k < m k = 0 if k < m k = k

48

M/M/m/m (contd)
Pk =
1 P0 ( / ) k!
k

if k m

0
k 1 P0 = ( / ) k ! k =0

if k > m
1

49

M/M/m/m (contd)
Let pm describes the fraction of time that all m servers are busy. The name given to this probability expression is Erlangs loss formula and is given by
pm = ( / ) m / m ! ( / ) k / k !
k =0 m

This equation is also referred to as Erlangs B formula and is commonly denoted by B(m,/) http://www.erlang.com
50