Loss Systems: Lect07.ppt S-38.145 - Introduction To Teletraffic Theory - Spring 2005

7.
Loss systems
lect07.ppt S-38.145 – Introduction to Teletraffic Theory – Spring 2005 1

7. Loss systems
Contents
• Refresher: Simple teletraffic model

• Poisson model (∞ customers, ∞ servers)
• Application to flow level modelling of streaming data traffic
• Erlang model (∞ customers, n < ∞ servers)
• Application to telephone traffic modelling in trunk network
• Binomial model (k < ∞ customers, n = k servers)
• Engset model (k < ∞ customers, n < k servers)
• Application to telephone traffic modelling in access network
2
7. Loss systems
Simple teletraffic model
• Customers arrive at rate λ (customers per time unit)

– 1/λ = average inter-arrival time
• Customers are served by n parallel servers
• When busy, a server serves at rate µ (customers per time unit)
– 1/µ = average service time of a customer
• There are n + m customer places in the system
– at least n service places and at most m waiting places
• It is assumed that blocked customers (arriving in a full system) are lost
µ1
λ µ
n+m µ
µ
n 3
7. Loss systems
Infinite system
• Infinite number of servers (n = ∞), no waiting places (m = 0)

– No customers are lost or even have to wait before getting served
• Sometimes,
– this hypothetical model can be used to get some approximate results for a
real system (with finite system capacity)
• Always,
– it gives bounds for the performance of a real system (with finite system
capacity)
– it is much easier to analyze than the corresponding finite capacity models
µ
1
λ µ
•
•
• ∞ 4
7. Loss systems
Pure loss system
• Finite number of servers (n < ∞), n service places, no waiting places

(m = 0 )
– If the system is full (with all n servers occupied) when a customer arrives,
it is not served at all but lost
– Some customers may be lost
• From the customer’s point of view, it is interesting to know e.g.
– What is the probability that the system is full when it arrives?
µ
1
λ µ
µ
µ
n 5
7. Loss systems
Contents

6
7. Loss systems
Poisson model (M/M/∞)
• Definition: Poisson model is the following simple teletraffic model:

– Infinite number of independent customers (k = ∞)
– Interarrival times are IID and exponentially distributed with mean 1/λ
• so, customers arrive according to a Poisson process with intensity λ
– Infinite number of servers (n = ∞)
– Service times are IID and exponentially distributed with mean 1/µ
– No waiting places (m = 0)
• Poisson model:
– Using Kendall’s notation, this is an M/M/∞ queue
– Infinite system, and, thus, lossless
• Notation:
– a = λ/µ = traffic intensity
7
7. Loss systems
State transition diagram
• Let X(t) denote the number of customers in the system at time t

– Assume that X(t) = i at some time t, and
consider what happens during a short time interval (t, t+h]:
• with prob. λh + o(h),
a new customer arrives (state transition i → i+1)
• if i > 0, then, with prob. iµh + o(h),
a customer leaves the system (state transition i → i−1)
• Process X(t) is clearly a Markov process with state transition diagram
λ λ λ
0 1 2
µ 2µ 3µ
• Note that process X(t) is an irreducible birth-death process

with an infinite state space S = {0,1,2,...}
8
7. Loss systems
Equilibrium distribution (1)
• Local balance equations (LBE):
π i λ = π i +1 (i + 1) µ (LBE)
⇒ π i +1 = (i +λ1) µ π i = i +a1π i
a i
⇒ π i = π 0 , i = 0,1,2, K
i!
• Normalizing condition (N):
∞ ∞ i
∑ π i = π 0 ∑ ai! = 1 (N)
i =0 i =0
−1
∞
 a  i
⇒ π 0 =  ∑ i!  = e a −1
( )
= e−a
 i =0  9
7. Loss systems
• Thus, the equilibrium distribution is a Poisson distribution:
X ∼ Poisson(a)
a i −a
P{ X = i} = π i = i! e , i = 0,1,2,K
2
E[ X ] = a, D [ X ] = a
• Remark: Insensitivity with respect to service time distribution
– The result is insensitive to the service time distribution, that is:
it is valid for any service time distribution with mean 1/µ
– So, instead of the M/M/∞ model,
we can consider, as well, the more general M/G/∞ model
10
7. Loss systems
Contents

11
7. Loss systems
Application to flow level modelling of streaming data traffic
• Poisson model may be applied to flow level modelling of streaming data

traffic
– customer = UDP flow with constant bit rate (CBR)
– λ = flow arrival rate (flows per time unit)
– h = 1/µ = average flow duration (time units)
– r = bit rate of a flow (data units per time unit)
– N = nr of active flows obeying Poisson(a) distribution
• When the total transmission rate Nr exceeds the link capacity C = nr,
bits are lost
– loss ratio ploss gives the ratio between the traffic lost and the traffic offered:
E[( Nr −C ) + ] E[( N − n) + ] 1 ∞ i −a
a
ploss = E[ Nr ]
= E[ N ] = a ∑ (i − n) i! e
i = n +1 12
7. Loss systems
Multiplexing gain
• We determine traffic intensity a so that loss ratio ploss < 1%

• Multiplexing gain is described by the traffic intensity per capacity unit,
a/n, as a function of capacity n
1
0.8
0.6
normalized traffic
a/n 0.4
0.2
20 40 60 80 100
capacity n
13
7. Loss systems
Contents

14
7. Loss systems
Erlang model (M/M/n/n)
• Definition: Erlang model is the following simple teletraffic model:

– Infinite number of independent customers (k = ∞)
– Interarrival times are IID and exponentially distributed with mean 1/λ
• so, customers arrive according to a Poisson process with intensity λ
– Finite number of servers (n < ∞)
• Erlang model:
– Using Kendall’s notation, this is an M/M/n/n queue
– Pure loss system, and, thus, lossy
• Notation:
15
7. Loss systems
• Let X(t) denote the number of customers in the system at time t

• with prob. λh + o(h),
a new customer arrives (state transition i → i+1)
• with prob. iµh + o(h),
a customer leaves the system (state transition i → i−1)
λ λ λ λ
0 1 n−1 n
µ 2µ (n−1)µ nµ

with a finite state space S = {0,1,2,…,n}
16
7. Loss systems
π i λ = π i +1 (i + 1) µ (LBE)
⇒ π i +1 = (i +λ1) µ π i = i +a1π i
a i
⇒ π i = π 0 , i = 0,1,K , n
i!
n n i
a
∑ π i = π 0 ∑ i! = 1 (N)
i =0 i =0
n −1
 a  i
⇒ π 0 =  ∑ 
i!
 i =0  17
7. Loss systems
• Thus, the equilibrium distribution is a truncated Poisson distribution:
ai
P{ X = i} = π i = i! , i = 0,1,K , n
n
aj
∑ j!
j =0
• Remark: Insensitivity with respect to the service time distribution

– The result is insensitive to the service time distribution, that is:
it is valid for any service time distribution with mean 1/µ
– So, instead of the M/M/n/n model,
we can consider, as well, the more general M/G/n/n model
18
7. Loss systems
Time blocking
• Time blocking Bt = probability that all n servers are occupied at an

arbitrary time = the fraction of time that all n servers are occupied
• For a stationary Markov process, this equals the probability πn of the
equilibrium distribution π. Thus,
an
Bt := P{ X = n} = π n = n!
n
aj
∑ j!
j =0
19
7. Loss systems
Call blocking
• Call blocking Bc = probability that an arriving customer finds all

n servers occupied = the fraction of arriving customers that are lost
• However, due to Poisson arrivals and PASTA property, the probability
that an arriving customer finds all n servers occupied equals the
probability that all n servers are occupied at an arbitrary time,
• In other words, call blocking Bc equals time blocking Bt:
an
Bc = Bt = n!
n
aj
∑ j!
j =0
• This is Erlang’s blocking formula

20
7. Loss systems
Contents

21
7. Loss systems
Application to telephone traffic modelling in trunk network
• Erlang model may be applied to modelling of telephone traffic in trunk

network where the number of potential users of a link is large
– customer = call
– λ = call arrival rate (calls per time unit)
– h = 1/µ = average call holding time (time units)
– n = link capacity (channels)
• A call is lost if all n channels are occupied when the call arrives
– call blocking Bc gives the probability of such an event
n
a
Bc = n!
n aj
∑ j =0 j!
22
7. Loss systems
Multiplexing gain
• We determine traffic intensity a so that call blocking Bc < 1%

a/n, as a function of capacity n
1
0.8
0.6
normalized traffic
a/n 0.4
0.2
20 40 60 80 100
capacity n
23
7. Loss systems
Contents

24
7. Loss systems
Binomial model (M/M/k/k/k)
• Definition: Binomial model is the following (simple) teletraffic model:

– Finite number of independent customers (k < ∞)
• on-off type customers (alternating between idleness and activity)
– Idle times are IID and exponentially distributed with mean 1/ν
– As many servers as customers (n = k)
• Binomial model:
– Using Kendall’s notation, this is an M/M/k/k/k queue
– Although a finite system, this is clearly lossless
• On-off type customer:
service
1
idleness
0
25
7. Loss systems
On-off type customer (1)
• Let Xj(t) denote the state of customer j ( j = 1,2,…,k ) at time t

– State 0 = idle, state 1 = active = in service
– Consider what happens during a short time interval (t, t+h]:
• if Xj(t) = 0, then, with prob. νh + o(h),
the customer becomes active (state transition 0 → 1)
• if Xj(t) = 1, then, with prob. µh + o(h),
the customer becomes idle (state transition 1 → 0)
• Process Xj(t) is clearly a Markov process with state transition diagram
ν
0 1
µ
• Note that process Xj(t) is an irreducible birth-death process

with a finite state space S = {0,1}
26
7. Loss systems
On-off type customer (2)
π 0( j )ν = π 1( j ) µ ⇒ π 1( j ) = νµ π 0( j )
µ
π 0( j ) + π 1( j ) = π 0( j ) (1 + νµ ) = 1 ⇒ π 0( j ) = ν + µ , π 1( j ) = ν ν+ µ
• So, the equilibrium distribution of a single customer is the Bernoulli
distribution with success probability ν/(ν+µ)
– offered traffic is ν/(ν+µ)
• From this, we could deduce that the equilibrium distribution of the state
of the whole system (that is: the number of active customers) is the
binomial distribution Bin(k, ν/(ν+µ))
27
7. Loss systems
• Let X(t) denote the number of active customers

• if i < k, then, with prob. (k−i)νh + o(h),
an idle customer becomes active (state transition i → i+1)
an active customer becomes idle (state transition i → i−1)
kν (k−1)ν 2ν ν
0 1 k−1 k
µ 2µ (k−1)µ kµ

with a finite state space S = {0,1,…,k}
28
7. Loss systems

π i (k − i )ν = π i +1 (i + 1) µ (LBE)
( k −i )ν
⇒ π i +1 = (i +1) µ π i
⇒ π i = i!( kk−! i )! (νµ )i π 0 = (ik )(νµ )i π 0 , i = 0,1,K , k

k k
∑ π i = π 0 ∑ (ik )(νµ )i = 1 (N)
i =0 i =0
k −1
  µ k
⇒ π 0 =  ∑ (i )( )  = (1 + ν ) − k = (
k ν i
)
µ µ ν + µ 29
 i =0 
7. Loss systems
• Thus, the equilibrium distribution is a binomial distribution:
X ∼ Bin(k , ν ν+ µ )
µ
P{ X = i} = π i = (ik )(ν ν+ µ )i (ν + µ ) k −i , i = 0,1, K , k
2 µ kνµ
E[ X ] = kν , D [X ] = k ⋅ ν ⋅ =
ν +µ ν +µ ν +µ (ν + µ ) 2
• Remark: Insensitivity w.r.t. service time and idle time distribution
– The result is insensitive both to the service and the idle time distribution,
that is: it is valid for any service time distribution with mean 1/µ and any idle
time distribution with mean 1/ν
– So, instead of the M/M/k/k/k model,
we can consider, as well, the more general G/G/k/k/k model
30
7. Loss systems
Contents

31
7. Loss systems
Engset model (M/M/n/n/k)
• Definition: Engset model is the following (simple) teletraffic model:

– Finite number of independent customers (k < ∞)
• on-off type customers (alternating between idleness and activity)
– Idle times are IID and exponentially distributed with mean 1/ν
– Less servers than customers (n < k)
– No waiting places (m = 0) Note: If the system is
• Engset model: full when an idle cust.
tries to become an
– Using Kendall’s notation, this is an M/M/n/n/k queue active cust., a new idle
– This is a pure loss system, and, thus, lossy period starts.
• On-off type customer:
service blocking!
1
idleness idle idle
0
32
7. Loss systems
• Let X(t) denote the number of active customers

• if i < n, then, with prob. (k−i)νh + o(h),
an idle customer becomes active (state transition i → i+1)
an active customer becomes idle (state transition i → i−1)
kν (k−1)ν (k−n+2) ν (k−n+1)ν
0 1 n−1 n
µ 2µ (n−1)µ nµ

with a finite state space S = {0,1,…,n}
33
7. Loss systems

π i (k − i )ν = π i +1 (i + 1) µ (LBE)
( k −i )ν
⇒ π i +1 = (i +1) µ π i
⇒ π i = i!( kk−! i )! (νµ )i π 0 = (ik )(νµ )i π 0 , i = 0,1, K, n

n n
∑ π i = π 0 ∑ (ik )(νµ )i = 1 (N)
i =0 i =0
n −1
 k ν i 
⇒ π 0 =  ∑ (i )( ) 
µ
 i =0  34
7. Loss systems
• Thus, the equilibrium distribution is a truncated binomial distribution:
µ
(ik )(νµ )i (ik )(ν ν+ µ )i (ν + µ ) k −i
P{ X = i} = π i = n = n , i = 0,K, n
k ν j k ν ) j ( µ )k − j
∑ j µ
( )( ) ∑ j ν +µ ν +µ
( )(
j =0 j =0
• Offered traffic is kν/(ν+µ)

• Remark: Insensitivity w.r.t. service time and idle time distribution
– The result is insensitive both to the service and the idle time distribution,
that is: it is valid for any service time distribution with mean 1/µ and any
idle time distribution with mean 1/ν
– So, instead of the M/M/n/n/k model,
we can consider, as well, the more general G/G/n/n/k model 35
7. Loss systems
Time blocking
• Time blocking Bt = probability that all n servers are occupied at an

arbitrary time = the fraction of time that all n servers are occupied
• For a stationary Markov process, this equals the probability πn of the
equilibrium distribution π. Thus,
( kn )(ν ) n
µ
Bt := P{ X = n} = π n = n
∑ ( kj )(νµ ) j
j =0
36
7. Loss systems
Call blocking (1)
• Call blocking Bc = probability that an arriving customer finds all

n servers occupied = the fraction of arriving customers that are lost
– In the Engset model, however, the “arrivals” do not follow a Poisson
process. Thus, we cannot utilize the PASTA property any more.
– In fact, the distribution of the state that an “arriving” customer sees differs
from the equilibrium distribution. Thus, call blocking Bc does not equal time
blocking Bt in the Engset model.
37
7. Loss systems
Call blocking (2)
• Let πi* denote the probability that there are i active customers when an
idle customer becomes active (which is called an “arrival”)
• Consider a long time interval (0,T):
– During this interval, the average time spent in state i is πiT
– During this time, the average number of “arriving” customers (who all see
the system to be in state i) is (k−i)ν⋅πiT
– During the whole interval, the average number of “arriving” customers is
Σj (k−j)ν⋅πjT
• Thus,
(k − i )ν ⋅ π iT (k − i ) ⋅ π i
πi* = n = n , i = 0,1,K, n
∑ (k − j )ν ⋅ π jT ∑ (k − j ) ⋅ π j
j =0 j =0
38
7. Loss systems
Call blocking (3)
• It can be shown (exercise!) that
( ki−1 )(νµ )i
πi* = n , i = 0,1, K, n
k −1 ν j
∑ j )( )
(
µ
j =0
• If we write explicitly the dependence of these probabilities on the total

number of customers, we get the following result:
π i * (k ) = π i (k − 1), i = 0,1, K, n
• In other words, an “arriving” customer sees such a system where there
is one customer less (itself!) in equilibrium
39
7. Loss systems
Call blocking (4)
• By choosing i = n, we get the following formula for the call blocking

probability:
Bc (k ) = π n * (k ) = π n (k − 1) = Bt ( k − 1)
• Thus, for the Engset model, the call blocking in a system with k
customers equals the time blocking in a system with k−1 customers:
( k n−1 )(νµ ) n
Bc (k ) = Bt (k − 1) = n
∑ ( k −j 1 )(νµ ) j
j =0
• This is Engset’s blocking formula

40
7. Loss systems
Contents

41
7. Loss systems
Application to telephone traffic modelling in access network
• Engset model may be applied to modelling of telephone traffic in

access network where the nr of potential users of a link is moderate
– customer = call
– ν = call arrival rate per idle user (calls per time unit)
– 1/µ = average call holding time (time units)
– k = number of potential users
– n = link capacity (channels)
• A call is lost if all n channels are occupied when the call arrives
– call blocking Bc gives the probability of such an event
( k n−1 )(νµ ) n
Bc = n
∑ ( k −j 1 )(νµ ) j
j =0 42
7. Loss systems
Multiplexing gain
• We assume that an access link is loaded by k = 100 potential users

• We determine traffic intensity kν/(ν+µ) so that call blocking Bc < 1%
kν/(n(ν+µ)) , as a function of capacity n
1
0.8
0.6
normalized traffic
kν/(n(ν+µ)) 0.4
0.2
20 40 60 80 100
43
capacity n
7. Loss systems
THE END
44

Loss Systems: Lect07.ppt S-38.145 - Introduction To Teletraffic Theory - Spring 2005

Uploaded by

Copyright:

Available Formats

Loss Systems: Lect07.ppt S-38.145 - Introduction To Teletraffic Theory - Spring 2005

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Loss Systems: Lect07.ppt S-38.145 - Introduction To Teletraffic Theory - Spring 2005

Uploaded by

Copyright:

Available Formats

7.

lect07.ppt S-38.145 – Introduction to Teletraffic Theory – Spring 2005 1

• Refresher: Simple teletraffic model

Simple teletraffic model

• Customers arrive at rate λ (customers per time unit)

• Infinite number of servers (n = ∞), no waiting places (m = 0)

Pure loss system

• Finite number of servers (n < ∞), n service places, no waiting places

• Refresher: Simple teletraffic model

Poisson model (M/M/∞)

• Definition: Poisson model is the following simple teletraffic model:

State transition diagram

• Let X(t) denote the number of customers in the system at time t

• Note that process X(t) is an irreducible birth-death process

Equilibrium distribution (1)

• Local balance equations (LBE):

Equilibrium distribution (2)

• Thus, the equilibrium distribution is a Poisson distribution:

• Refresher: Simple teletraffic model

Application to flow level modelling of streaming data traffic

• Poisson model may be applied to flow level modelling of streaming data

• We determine traffic intensity a so that loss ratio ploss < 1%

• Refresher: Simple teletraffic model

Erlang model (M/M/n/n)

• Definition: Erlang model is the following simple teletraffic model:

State transition diagram

• Let X(t) denote the number of customers in the system at time t

• Note that process X(t) is an irreducible birth-death process

Equilibrium distribution (1)

• Local balance equations (LBE):

Equilibrium distribution (2)

• Thus, the equilibrium distribution is a truncated Poisson distribution:

• Remark: Insensitivity with respect to the service time distribution

• Time blocking Bt = probability that all n servers are occupied at an

• Call blocking Bc = probability that an arriving customer finds all

• This is Erlang’s blocking formula

• Refresher: Simple teletraffic model

Application to telephone traffic modelling in trunk network

• Erlang model may be applied to modelling of telephone traffic in trunk

• We determine traffic intensity a so that call blocking Bc < 1%

• Refresher: Simple teletraffic model

Binomial model (M/M/k/k/k)

• Definition: Binomial model is the following (simple) teletraffic model:

On-off type customer (1)

• Let Xj(t) denote the state of customer j ( j = 1,2,…,k ) at time t

• Note that process Xj(t) is an irreducible birth-death process

On-off type customer (2)

• Local balance equations (LBE):

State transition diagram

• Let X(t) denote the number of active customers

• Note that process X(t) is an irreducible birth-death process

Equilibrium distribution (1)

• Local balance equations (LBE):

⇒ π i = i!( kk−! i )! (νµ )i π 0 = (ik )(νµ )i π 0 , i = 0,1,K , k

Equilibrium distribution (2)

• Thus, the equilibrium distribution is a binomial distribution:

• Refresher: Simple teletraffic model

Engset model (M/M/n/n/k)