SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

MODULE-3

TELECOMMUNICATIONS TRAFFIC: Introduction, Unit of traffic,

Congestion, Traffic measurement, Mathematical model, lost call systems,
Queuing systems. SWITCHING SYSTEMS: Introduction, Single stage
networks, Gradings, Link Systems, GOS of Linked systems. [Text-1]
8Hours

Chapter 1: TELECOMMUNICATIONS TRAFFIC:

1.1 Introduction:
 When any industrial plant is to be designed, an initial decision must be made as to its size, in
order to obtain the desired throughput.
 In telecommunication system, it is the traffic to be handled. This determines the number of
trunks to be provided.

Trunk:

 In teletraffic engineering, trunk is used to describe any entity that will carry one call.
 It may be an international circuit with a length of thousands of kilometers or a few meters of
wire between switches in the same telephone exchange.

Trunking: the arrangement of trunks and switches within a telephone exchange is called trunking.

Traffic variation in minutes

 The figure 1.1 shows the average traffic in minutes. It shows the number of calls, in progress
on a large telecommunication system (example: telephone exchange) made over a few
minutes.
 Here the number of calls varies in a random manner as individual calls begin and end.

Traffic variation during a day

 Figure 1.2 shows the way in which the average traffic varies during the working day at a typical
medium size telephone exchange or a transmission route.
 There are very few calls during the night.
 The number of calls rises as people go to work and reaches a maximum by the middle of the
morning.
 It falls at the mid-day, as people go to home from working place and it has a further peak in the
evening as people make social calls.

1
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

 Busy hour: It is a period of one hour, which corresponds to the peak traffic load. In figure 1.2
Busy hour is from 10 am to 11 am.

Figure 1.1 Short term traffic variation

Figure 1.2 Traffic variations during a day

1.2 The unit of traffic

 Traffic intensity or simple traffic is defined as the average number of calls in progress.
 The unit of traffic is Erlang [E], named after A.K Erlang the Danish pioneer of traffic.
 On a group of trunks the average number of calls in progress depends on both the number of
calls which arrive and their duration.

2
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Fig 1.3 Examples of 1 Erlang of traffic carried on three trunks

Holding time:

Duration of call is often called its holding time, because its holds a trunk for that time.

The example in figure 1.3 shows how one Erlang of traffic can result from one trunk being busy all of
the time, for each of two trunks being busy for half of time or from each of three trunks being busy for
one third of the time as in figure a, b and c.

In North America, traffic expressed in terms of hundreds of call seconds per hour (CCS).

1 hour= 3600 seconds

1 Erlang=36CCS

From the definition of the Erlang that the traffic carried by the group of trunk is given by

𝐶ℎ
A= 𝑇 …………………..(1.1)

A= traffic in Erlangs

C= average number of call arrivals during time T

h= average call holding time.

If T=h, then A=C. Thus number of calls arriving during a period equal to the mean duration of the calls.

3
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Since a single trunk cannot carry more than one call, A<=1, the traffic is a fraction of an Erlang equal to
the average propagation of time for which the trunk is busy. This is called the Accupancy of the trunk.

P1.1. At certain exchange a total of 5000 calls are originated during the busy hour. If the average
𝟏
holding time of a call is 2 minutes, calculate the flow of traffic during this period.
𝟐

Solution:

A= C * t
5000∗2.5
=
60

= 208.3 E

P1.2. On an average during the busy hour, a company makes 120 outgoing calls of average
duration 2 minutes. It receives 200 incoming calls of average duration 3 minutes. Find the

a). Outgoing traffic

b) Incoming traffic

c). Total traffic

𝑜𝑢𝑡 𝑔𝑜𝑖𝑛𝑔 𝐶𝑎𝑙𝑙𝑠 ∗ ℎ 120∗2
AOG= = = 4E
𝑡 60

𝐼𝑛𝑐𝑜𝑚𝑖𝑛𝑔 𝐶𝑎𝑙𝑙𝑠 ∗ ℎ 200∗3

AIC= = = 10 E
𝑡 60

Total traffic =outgoing traffic + incoming traffic

=4+10

=14E

P1.3: During the busy hour, on average, a customer with a single telephone line makes three calls
and receives three calls. The average call duration is 2 minutes. What is the probability that a
caller will find the line engaged?

Occupancy of line: (3+3)*2/60 =0.1E = Probability of finding the line engaged.

1.3 Congestion
The situation, where all the trunks in a group of trunks are busy, and so it can accept no further calls,
this state is known as congestion.

In message switching system, calls that arrive during congestion wait in a queue until an outgoing
trunk becomes free. Thus they are delayed but not lost, such systems are queuing or delay system.

4
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

In circuit switching system (example: telephone exchange), all attempts to make a call over a congested
group of trunks are unsuccessful, such systems are called as lost-call systems.

In lost call system

Traffic Carried= trffic offered- traffic lost

The proportion of calls that is lost or delayed due to congestion is a measure of the service provided. It
is called as grade of service (B). For a lost call system, the grade of service, B can be defined as:
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑎𝑙𝑙𝑠 𝑙𝑜𝑠𝑡
B=
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑎𝑙𝑙𝑠 𝑜𝑓𝑓𝑒𝑟𝑒𝑑

Hence, also:
𝑇𝑟𝑓𝑓𝑖𝑐 𝑙𝑜𝑠𝑡
B=
𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑜𝑓𝑓𝑒𝑟𝑒𝑑

= Proportion of the time which congestion exists

= Probability of congestion

=Probability that a call will be lost due to congestion

Thus, if traffic A Erlangs is offered to a group of trunks having grade of service B, the traffic lost is AB
and the traffic carried= A-AB =A(1-B) Erlangs

The larger the grade of service, the worse is the service given. The grade of service is normally
specified for the traffic at the busy hour. At other times it is much better.

In practice, busy hour GOS can vary from, 1 in 1000 for cheap trunks inside an exchange to 1 in 100 for
interexchange connections and 1 in 10 for expensive international routes.

Dimensioning problem: It is the basic problem of determining the size of a telecommunication system.

P1.4: During the busy hour, 1200calls were offered to a group of trunks and six calls were lost.
The average call duration was 3 minutes. Find:

1. the traffic offered

2. the traffic carried
3. the traffic lost
4. the grade of services
5. the total duration of the periods of congestions

Solutions:
𝐶∗ ℎ 1200∗3
Traffic carried: A= = = 60E
𝑇 60

𝐶∗ ℎ 1194∗3
Traffic offered: A= = = 59.7E
𝑇 60

5
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

𝐶∗ ℎ 6∗3
Traffic lost: A= = = 0.3E
𝑇 60

B=GOS=traffic lost/ traffic carried =0.3/60=0.005

The total duration of periods of congestion=0.005*3600=18 Seconds

1.4 Traffic measurement

 It means that measuring of busy hour traffic is necessary & regularly operating company need
to measure and must keep records.
 By definition, measuring the traffic carried amounts to counting calls in progress at regular
intervals during the busy hour and averaging results
 In the past, engineers counted the plugs inserted in a manual switchboard or the number of
selector-off normal in an automatic exchange by peg count or switch count method. Later
automatic traffic records were used in automatic exchange.
 In modern stored program controlled system, the central processors generate records of the
calls that they set up.

P1.5: Observations were made of the number of busy lines in a group of junctions at intervals of 5
minutes during the busy hour. The results obtained were: 11,13,8,10,14,12,7,,15,17,16,12

It is therefore estimated that the traffic carried, in Erlangs was:

11+13+8+10+14+12+7+9+15+1+16+12
= 12E
12

1.5 A mathematical model

To obtain analytical solutions to teletraffic problems it is necessary to have a mathematical model of the
traffic offered to telecommunication system.

A simple model is based on the following assumptions

 Pure chance traffic

 Statistical equilibrium

Pure chance traffic: The assumption of pure chance traffic means that call arrivals and call
terminations are independent random events. Sometimes it is also called as Poissonian traffic.

If call arrivals are independent random events, their occurrence is not affected by previous calls.
Sometimes traffic is called as memoryless traffic.

This assumption of random call arrivals & termination leads to the following results.

1. The number of call arrivals in a given time has a Poisson distribution i.e.
𝜇𝑥
P(x) =
𝑥!
𝑒 −𝜇

6
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Where x is the number of call arrivals in a given time T and µ is the mean number of call
arrivals in time T. For this reason, pure-chance traffic is also called Possonian traffic.

2. The intervals, T, between call arrivals are the intervals between independent random event.
These intervals have a negative exponential distribution, i.e,:
P(T ≥ t)=𝑒 −𝑡/𝑇̅
Where 𝑇̅ is the mean interval between call arrivals.
3. Since the arrival of each call and also the intervals between two random events, call durations,
T, are also the intervals between two random events and have a negative exponential
distribution, i,e.,
P(T ≥ t)=𝑒 −𝑡/ℎ
Where h is the mean call duration (holding time)

The intervals, T, between call arrivals are the intervals between independent random events, have a
negative exponential distribution i.e.

Since the arrival of each call and its termination are independent random events, call duration, T are
also intervals between two random events have a negative exponential distribution i.e.,

Statistical equilibrium: The assumption of Statistical equilibrium means that the generation of traffic
is a stationary random process i.e., probabilities do not change during the period being considered.
Consequently the mean number of calls in progress remains constant.

Statistical equilibrium is not obtained immediately before the busy hour, when the calling rate is
increasing nor at the end of the busy hour, when calling rate is falling.

P1.7: On average one call arrivers every 5 seconds. During a period of 10 seconds, what is the
probability that:

1. No call arrives?
2. One call arrives?
3. Two call arrives?
4. More than two calls arrive?
𝜇𝑥
P(x) = 𝑒 −𝜇 , Where µ = 2
𝑥!

20
1. P (0)= 𝑒 −2 = 0.135
0!
21
2. P(1)= 𝑒 −2 =0.270
1!
22 −2
3. P(2) = 𝑒 = 0.270
2!
4. P(>2)=1-P(0)-P(1)-P(2)
= 1-0.135-0.270-0.270
= 0.325

P1.8 In a telephone system the average call duration is 2 minutes. A call has already lasted 4
minutes. What is the probability that:

7
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

1. The call will last at least another 4 minutes?

2. The call will end within the next 4 minutes?

These probabilities can be assumed to be independent of the time which has already elapsed?
−𝑡
1. P(T)≥ t)=𝑒 ⁄ℎ = 𝑒 −2 0.135
2. P(T≤ t) =1- P(T ≥ t) = 1-0.135= 0.865

State transition diagram for N trunks

When we have N trunks the number of calls in progress varies randomly as shown in figure 1.1
It is an example of Birth and death process also called as renewal process. The number of calls in
progress is always between 0 and N. Thus it has N+1 states and its behavior depends on the probability
of change from each state to the one above and to the one below it. This is called as simple Markov
chain as shown in figure 1.4

Fig 1.4 state transition diagram for N trunks

In figure P(j) is the probability of state j and P(k) is the probability of the next state higher state
k, Pj,k is the probability of a state increase to k, given that the present state is j. P k,j is the
probability of a decrease to j, given that the present state is k. The probabilities P(0),
P(1),………..P(N) are called the state probabilities and the conditional probabilities Pj,k Pk,j,
are called the transition probabilities of the Markov chain. If there is statistical equilibrium,
these probabilities do not change and the process is said to be a regular Markov chain.

If there is statistical equilibrium, these probabilities do not change and process is said to be a
regular Markov chain.

Consider a very small interval of time ⸹t, starting at time t. Since ⸹t is very small, the
probability of something happening during it is small. The probability of two or more events
during ⸹t, is negligible. The events which can happen during ⸹t are thus as follows.

One call arriving, with probability P(a).

One call ending, with probability P(e).

No change, with probability 1-P(a)-P(e).

8
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

W. k. t the traffic carried by a group of trunk is given by

𝐶ℎ
A= 𝑇 -----(1.1),

Then the mean number of calls arriving during the average holding time, h is C=A. Thus the
mean number arriving during time One call arriving, with probability ⸹t is 𝐴𝛿𝑡⁄ℎ << and
represents the probability, P(a), of a call arriving during ⸹t.

Pj.k=P(a)=A⸹t/h…….(1.2)

If the mean holding time is h and the number of calls in progress is k, one expects an average
of k calls to end during a period h. The average number of calls ending during ⸹t is therefore
k⸹t/h. Since ⸹t is very small k⸹t/h<<1 and represents the probability, P(e), of a call ending
during ⸹t.

Pk,j=P(e)=k⸹t/h…….(1.3)

If the probability of j calls in progress at time t is P(j), then the probability of a transition from j
to k busy trunks during ⸹t is:

P(jk) = P(j) P(a) = P(j) A⸹t/h …………..(1.4)

If the probability of k calls in progress at time t is P(k), then the probability of a transition from
k to j busy trunks during ⸹t is:

P(kj) = P(k) P(e) = P(k) k⸹t/h………………..(1.5)

The assumption of statistical equilibrium requires that P(jk) =P(kj). Otherwise the number
of calls in progress would steadily increase or decrease. Thus, from equations (1.4) and (1.5)

K P(k) ⸹t/h = A P(j) ⸹t/h

𝐴
P(k) =𝑘 P(j)……………..(1.6)

𝐴
Hence: P(1) = P(0)
1

𝐴 𝐴2
P(2) = 2 P(1) =2.1 P(0)

𝐴 𝐴3
P(3) = 3 P(2) = 3∗2∗1 P(0)

And, in general:
𝐴𝑥
P(x) = 𝑃(0)…………………(1.7)
𝑥!

9
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

The assumption of pure chance traffic implies a very large number of sources. Thus, x can
have any value between zero and infinity and the sum of their probabilities must be unity. Then
𝐴𝑥
1= ∑∞ ∞ 𝐴
𝑥=0 𝑃 (𝑥 )= ∑𝑥=0 𝑥! 𝑃 (0) = 𝑒 𝑃 (0)

P(0) = 𝑒 −𝐴
𝐴𝑥
P(x) = 𝑒 −𝐴
𝑥!

Thus, if call arrivals have a Poisson distribution, so does the number of calls in progress. This
requires an infinite number of trunks to carry the calls. If the number of trunks available is
finite, then some calls can be lost or delayed and the distribution is no longer Poissonian.

3.6 Lost call systems

Theory:

Erlang determined the grade of service (i.e. the loss probability) of a lost call system having N trunks
when offered traffic A as shown in figure 1.5

The solution depends on the following assumptions

 Pure chance traffic

 Statistical equilibrium
 Full availability
 Calls encounter during congestion are lost.

Fig 1.5 Lost call system

The pure chance traffic implies that call arrivals and call terminations are independent random events.

The Statistical equilibrium implies that probabilities do not change.

Full availability means that every call that arrives can be connected to any outgoing trunk which is free.
If the incoming calls are connected to the outgoing trunks by switches, switches must have sufficient
outlets to provide access to every outgoing trunk.

The lost-call assumption implies that any attempted calls which encounter congestion are immediately
cleared from the system.

10
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Here we are assuming that the traffic offered is the total arising from all successful and unsuccessful
calls.

If there are x calls in progress, then equation (1.7) gives

𝐴𝑥
P(x) = 𝑃 (0)
𝑥!

However, there cannot be a negative number of calls and there cannot be more than N. Thus,
we know with certainty that 0 ≤ x ≤N.
𝐴𝑥
∑𝑁 𝑁
𝑥=0 𝑃 (𝑥 )=1= ∑𝑥=0 𝑃 (0)
𝑥!

Hence, P(0) = 1⁄ 𝐴𝑥
∑𝑁
𝑥=0 𝑥!

Substituting in equation (1.7) gives:

𝐴𝑥⁄
𝑥!
P(x) = ………………………….(1.8)
𝐴𝑘
∑𝑁
𝑘=0 𝑘!

This is called first Erlang distribution/Erlang’s lost call formula.

P(N) , Since this is the probability of congestion, i.e., the probability of a lost call, which is the grade of
service B, This is given the symbol E1,N(A) which denotes the loss probability for a full availability
group of N trunks offered traffic A Erlangs.

𝐴𝑁⁄
𝑁!
B= E1,N(A)= -------------(1.9)
𝐴 𝑘
∑𝑁
𝑘=0 𝑘!

Equation 1.9 is a special case of equation (1.8)

The grade of service of a loss system with N full availability trunks, offered A 𝑒𝑟𝑙𝑎𝑛𝑔𝑠 of traffic, is
given by E1,N(A).
𝐴𝑁−1⁄
(𝑁−1)!
B= E1,N-1=
𝐴𝑘
∑𝑁−1
𝑘=0
𝑘!

𝑘 𝐴𝑁−1⁄
𝐴 (𝑁−1)! 𝐴𝑁
∑𝑁
𝑘=0 𝑘! = +
𝐸1,𝑁−1(𝐴) 𝑁!

Substituting in equation (1.9)

𝐴𝐸1,𝑁−1(𝐴)
E1,N-1(A)=
𝑁+𝐴𝐸1,𝑁−1(𝐴)

11
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Since E1,0 =1, this iterative formula enables 𝐸1,𝑁(𝐴) to be computed for all values of N. Table
shows values 𝐸1,,𝑁(𝐴).

P1.8: A group of five trunks is offered 2 E of traffic. Find:

1. The grade of service

2. The probability that only one trunk is busy
3. The probability that only one trunk is free
4. The probability that at least one trunk is free.
𝐴𝑁⁄
𝑁!
B= E1,N(A)=
𝐴 𝑘
∑𝑁
𝑘=0 𝑘!

32⁄
120
B= E1,N(A)= 2 4 8 16 32 for N=5, A=2
1+1+2+6+24+120

0.2667
= =0.037
7.2667

From equation (1.8) P(1)=2/7.2667=0.275

16/24
P(4)= =0.0917
7.2667
P(x<5) =1-P(5)=1-B = 1-0.0037=0.963

P1.9 A group of 20 trunks provides a grade of service of 0.01 when offered 12E of traffic.

1. How much is the grade of service improved if one extra trunk is added to the group?
2. How much does the grade of service deteriorate if one trunk is out of service?

12∗𝐸1,20(12)
E1,21(12)=
21+12𝐸1,20(12)

12∗0.01
= = 0.0057
21+12∗0.01

12∗𝐸1,19(12)
E1,20(12)=0.01= =0.2+0.12E1,19(12)= 12E1,19(12)
20+12𝐸1,19(12)

E1,19(12)=0.017

Traffic performance:
If the offered traffic, A, increases, the number of trunks, N, must obviously be increased to provide a
given grade of service. However, for the same trunk occupancy the probability of finding all trunks
busy is less for a large group to trunks can have a higher occupancy than a small one, i.e. the large
group is more efficient. This is shown by figure 1.6. For a grade of grade of service of 0.002(i.e. one

12
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

lost call in 500). For example, 2E of traffic requires seven trunks and their occupancy is 0.27E.
However, 20E requires 32 trunks and their occupancy is 0.61E.

Fig 1.6 Trunk occupancies for full availability groups of various sizes (Grade of services
=0.002)

Fig 1.7 effect of overload on grade of service

The penalty paid for the high efficiency of large group of trunks is that the grade of service
(GOS) deteriorates more with traffic overloads than for small Groups of trunks. Figure 1.7
shows the variation of grade of service with respect to offered traffic for different sizes of
group, which were all dimensioned to provide a GOS of 0.002 at their normal traffic load. For
a group of five trunk, a 10% overload increases the GOS by 40%, However, for a group of 100
trunks, it causes the GOS to increase by 550%.

13
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

For the above reason the Digital switching system operating companies uses a two criteria.
Two grade of service are:

1) Normal criteria: for nominal traffic load.

2) Overload criteria: Larger GOS for normal load and overload

Sequential Selection:
In many switching systems, trunks in a group are selected by means of sequential search. A
call is not connected to trunk number2 unless number1 is busy. It is not connected to number 3
unless both numner1 and nunber2 are busy and so on.

Call finding the last choice trunk is busy or lost. As a result, the first trunk has a very high
occupancy and the traffic carried by subsequent trunk is less. The last choice trunk is very
highly loaded indeed.

The behavior is shown in figure 1.8. The performance of such an arrangement can be analyzed
as follows. Let traffic A Erlangs be offered to the group of trunks. From equation (3.9) the
GOS of a single- trunk is

E1,1(A) =A/1+A

Traffic overflowing from the first trunk to the second is

A E1,1(A) =A2/1+A

Therefore traffic carried by the first trunk is

Traffic offered-traffic lost=A- A2/1+A= A/1+A

In general traffic carried by kth trunk = Traffic lost from group of first k-1 trunks – Traffic lost
from group of first k trunk

=A [E1,K-1(A) - E1,K(A)]

The traffic offered to first trunk is Poissonian. Traffic overflowing to the second trunk is more
peaky.

14
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Fig 1.8 Distribution of traffic over trunks of a group with sequential search

P1.11: If sequential is used for the group of trunks five trunks is offered 2E of traffic. Then how
much is the traffic carried by
1. The first choice trunk?
2. The last choice trunk?

1. The traffic carried by the first choice trunk is:

A/1+A =2/(1+2) = 0.67E
32⁄
120
E1,5 (2)= B= E1,N(A)= 2 4 8 16 32 for N=5, A=2
1+1+2+6+24+120
0.2667
=7.2667=0.037
16⁄
24
E1,4(2)= 2 4 8 16 for N=4, A=2
1+1+2+6+24
Traffic carried b the last choice trunk =2 (0.095-0.037) =0.12E.

Loss systems in tandem

GOS of several links in tandem is explained below. For connections consisting of two links,
having grade of service B1, B2, is offered traffic A Erlangs, then:
Traffic offered to second link = A(1-B1)
therefore reaching destination =A(1-B1)(1-B2)
= A (1+B1B2-B1-B2)
And the overall grade of service is B=B1+B2-B1B2

15
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

If B1,B2 <<1, as they should be, then B1,B2 is negligible and overall grade of service is
simply B=B1+B2
N general for n-trunk connection, we may write

B=∑𝑛𝑘=1 𝐵𝑘

Use of traffic tables:

Table 1.1: Traffic capacity table for full- availability groups

1.8 Queuing system:

The second Erlang distribution: The queuing system is shown in figure 1.9

16
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Fig 1.9 Queuing system

Here traffic A is offered to a queuing system with N trunks. Here Erlang determined the probability of
encountering delay. Trunks are often called as servers

Erlang solution depends on the following assumptions

 Pure chance traffic

 Statistical equilibrium
 Full availability
 Calls which encounter congestion enter a queue and are stored there until a server becomes
free.

It is some time called as M/M/N system

Note: This notation introduced by Kendall

It describes the operating system as X/Y/N Where X is the input process Y is the service time
distribution and N is the numbers of servers. The following symbols are used M stands for Markov
Process (i.e. random arrivals and terminations).

The pure chance traffic implies that call arrivals and call terminations are independent random events.

The Statistical equilibrium implies that probabilities do not change during the period being considered
i.e. A < N. If A ≥ N, calls are entering the system at a greater rate than they leave. As a result, the
length of the queue must increase towards infinity. This is not statistical equilibrium.

Let x be the total number of calls in the system. Thus, when x<N, then x calls are being served and
there is no delay. When x>, all the servers are busy and incoming calls encounter delay; there are N
calls being served and x-N calls in the queue.

If x ≤ N:

There is no queue and the behavior of the system is the same as that of a lost-call system in the absence
of congestion. Thus, from equations (1.7)
𝐴𝑥
P(x) = 𝑃(0)
𝑥!

If x≥N: The probability of a call arrival in a very short period of time, ⸹t, from equations (1.2)
is given by

P(a)=A⸹t/h

17
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Where h is the mean service time

Thus the probability of a transition from x-1 to x calls in the system during ⸹t, from equations
(1.4) is given by:

P(x-1x) = P(x-1) P(a) = P(x-1) A⸹t/h

Since all servers are busy, only the N calls being served can terminate (instead of x calls in a
lost- call system). Therefore, modified equation 1.3 is given by

P(e)= N ⸹t/h

And the probability of a transition from x to x-1 calls is given by:

P(xx-1) = P(x) P(e) = P(k) N ⸹t/h

For statistical equilibrium

P(xx-1) =P(x-1x).

P(x) N ⸹t/h = P(x-1) A ⸹t/h

𝐴
P(x) =𝑁 P(x-1)………..1.10

𝐴𝑁
But P(N) = 𝑁! P(0)

𝐴 𝐴𝑁+1
P(N+1) =𝑁 P(N) = 𝑁.𝑁! P(0)

𝐴 𝐴𝑁+2
P(N+2) =𝑁 P(N+1) = 𝑁2 𝑁! P(0) and so on

In general, for x ≥ N :

𝐴𝑥 𝑁𝑁 𝐴 𝑥
P(x)=𝑁𝑥−𝑁𝑁! 𝑃(0) = 𝑁! (𝑁) 𝑃(0) ----------(1.11)

If there is no limit to the possible length of Queue, the x can have any value between zero and
infinity,
∞

∑ P(x) = 1
x=0

Thus from equations 1.7 and 1.11

𝑁−1 ∞
1 𝐴𝑥 𝑁𝑁 𝐴 𝑥 𝐴 𝑘
= ∑ + ( ) ∑ ( ) − − − − − − − (1.12)
𝑃(0) 𝑥! 𝑁! 𝑁 𝑁
𝑥=0 𝑘=0

18
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

𝐴
Where k = x-N. Since ≤ 1 then
𝑁
∞
𝐴 𝑘 𝐴 −1
∑ ( ) = [1 − ]
𝑁 𝑁
𝑘=0

𝑁−1
1 𝐴𝑥 𝐴𝑁 𝐴 −1
= ∑ + [1 − ]
𝑃(0) 𝑥! 𝑁! 𝑁
𝑥=0

1 𝑁 𝐴𝑁 𝐴𝑥
= [𝑁!(𝑁−𝐴) + ∑𝑁−1
𝑥=0 𝑥! ]……………………..(1.13)
𝑃(0)

Thus, P(x) is given by equations 1.7 or 1.11 depending on whether x ≤ N or x ≥ N

N is given by equation (1.13). This is called as second Erlang distribution

Probability of delay:
Delay occurs if all servers are busy, i.e. x ≥ N. Now from equation 1.11 the probability that are at
least z calls in the system (where z ≥N) is given by

P(x ≥ z)= ∑∞
𝑥=𝑧 𝑃(𝑥)
∞
𝑁𝑁 𝐴 𝑥
= 𝑃(0) ∑ ( )
𝑁! 𝑁
𝑥=𝑧
∞
𝑁𝑁 𝐴 𝑧 𝐴 𝑘
= 𝑃(0) ( ) ∑ ( )
𝑁! 𝑁 𝑁
𝑘=0

Where k = x-N

𝑁𝑁 𝐴 𝑧 𝐴 −1
P(x ≥ z)= 𝑃(0) ( ) (1 − ) ------- (1.14)
𝑁! 𝑁 𝑁

𝑁𝑁 𝐴 𝑧 𝑁
P(x ≥ z)= 𝑃(0) ( )
𝑁! 𝑁 𝑁−𝐴

The probability of delay is PD= P(x≥N)

𝐴𝑁 𝑁
PD= 𝑃(0) -------- (1.15)
𝑁! 𝑁−𝐴

PD=E2,N (A)

The probability of delay, for a system with N servers offered traffic A Erlangs, is thus given by
equation 1.15 where P(0) is given by equation 1.1. This formula is called as Erlang delay formula.

The probability of delay increases towards 1.0 as an increase towards N. When A>N, the length of the
queue grows indefinitely, and is shown in figure 1.10.

19
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Fig 1.10 Delay probabilities for queuing systems (A= traffic in Erlangs, N=number of servers)

Finite queue capacity:

A practical system cannot contain an infinite queue. Thus, when the queue has become full,
calls that arrive subsequently are lost. If the queue can only hold up to Q calls, then x ≤ Q+N
and equation (1.12) becomes:
𝑁−1 𝑄
1 𝐴𝑥 𝑁𝑁 𝐴 𝑁 𝐴 𝑘
= ∑ + ( ) ∑ ( )
𝑃(0) 𝑥! 𝑁! 𝑁 𝑁
𝑥=0 𝑘=0

𝑁−1 𝐴
1 𝐴𝑥 𝐴𝑁 1 − ( )𝑄+1
= ∑ + 𝑁 − − − − − − − (1.16)
𝑃(0) 𝑥! 𝑁! 𝐴
𝑥=0 1−( )
𝑁
However, if the loss probability is small, there is negligible error in using equation (1.13).

The loss probability can be estimated by first assuming that the queue capacity is infinite and then
calculating P(x ≥ Q+N)

𝑁 𝐴 𝑄+𝑁 𝑁
P(x ≥ Q + N) = ( ) 𝑃(0)
𝑁! 𝑁 𝑁−𝐴
Now from equation 1.14

P(x ≥ Q+N) =

𝐴 𝑄
= ( ) 𝑃𝐷 …………………………(1.17)
𝑁

Hence, the queue capacity, Q, needed to obtain an adequately low loss probability can be found

20
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Some other useful results

Equations 1.1 to 1.15 lead to further results,

Follows:

1. Mean number of calls in the system:

i) When there is delay, the mean number of calls is
𝐴
̅̅̅
𝑥1 = +𝑁
𝑁−𝐴
ii) Averaged over all time, the mean number of calls is
𝐴
𝑥̅ = 𝐸 (𝐴) + 𝐴
𝑁 − 𝐴 2,𝑁
2. Mean length of Queue:
a. When there is delay, the mean queue length is

𝐴
̅̅̅
𝑞1 = ̅̅̅
𝑥1 − 𝑁 =
𝑁−𝐴
b. Mean length of queue averaged over all time is
𝐴
𝑞̅ = ̅̅̅
𝑞1𝑃𝐷 = 𝐸 (𝐴)
𝑁 − 𝐴 2,𝑁
3. Mean delay time when the queue discipline is first in first out (FIFO):
i) When there is a delay, mean delay is ̅̅̅̅ 𝑇1
ℎ
̅̅̅̅
𝑇1 =
𝑁−𝐴
Where h is the mean holding time
ii) Averaged over all time, the mean delay, 𝑇1 ̅̅̅̅, is
𝑇̅ = 𝐸2,𝑁 (𝐴)𝑇1̅̅̅̅
ℎ
= 𝐸2,𝑁 (𝐴) 𝑁−𝐴
4. Distribution of delays (FIFO queue discipline):

Since the holding times have a negative exponential probability distribution, so since the
holding times have a negative exponential probability distribution, so do the delays, TD. Hence:
−𝑡⁄
i). When there is delay, P(𝑇𝐷 ≥ t) = 𝑒 ̅̅̅
𝑇′

ii) Averaged over all time,

−𝑡⁄
P(𝑇𝐷 ≥ t) = 𝐸2,𝑁 (𝐴)𝑒 ̅̅̅
𝑇′

21
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Fig 1.11 Mean delays for queuing systems with FIFO queue discipline (T=mean delay,
h=mean service time, N= number of server)

System with single server:

When there is only single server, the probability of it being is simply its occupancy, A and this is the
probability of delay i.e. 𝐸2,𝑁 (𝐴)=A. As a result, the expression from previous section yields
below results

N=1, PD= 𝐸2,𝑁 (𝐴) = 𝐴

𝐴𝑁 𝑁
PD= 𝑃(0)
𝑁! 𝑁−𝐴

𝐴𝑁 𝑁 𝐴 1
Then P(0)= PD / ⌈ ⌉ P(0)= A / ⌈ ⌉
𝑁! 𝑁−𝐴 1! 1−𝐴

P(0)= 1-A
𝐴 1
̅̅̅
𝑥1 = 1−𝐴 + 1 =1−𝐴
𝐴 𝐴
𝑥̅ = ∗ 𝐴 + 𝐴=
1−𝐴 1−𝐴
𝐴
̅̅̅
𝑞1 = ̅̅̅
𝑥1 − 𝑁 =
1−𝐴
𝐴2
𝑞̅ = ̅̅̅
𝑞1𝑃𝐷 =
1−𝐴
ℎ
̅̅̅̅
𝑇1 =
1−𝐴
𝑇̅ = 𝐸2,𝑁 (𝐴)𝑇1
̅̅̅̅

𝐴ℎ
= 1−𝐴

𝑁𝑁 𝐴 𝑥
P(x)= = 𝑁! (𝑁) 𝑃(0) 𝐴𝑥 (1 − 𝐴)

P(x≥z)= 𝐴 𝑧

22
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Problem 1.12:

A PBX has three operators on duty and receives 400 calls during the busy hour. Incoming calls
enter a queue and are dealt with in order of arrival. The average time taken by an operator to
handle a call is 18 seconds. Calls arrivals are Poissonian and operator service times have a
negative exponential distribution.

1. What percentage of calls have to wait for an operator to answer them?

2. What is the average delay, for calls and for those which encounter delay?
3. What percentage of calls are delayed for more than 30 seconds?
1. A= 400 * 18/3600 =2.0E
From equation (4.13)

1 𝑁 𝐴𝑁 𝐴𝑥
=[ + ∑𝑁−1
𝑥=0 ]
𝑃(0) 𝑁!(𝑁−𝐴) 𝑥!

1 3 23 2 4
= [3!(3−2) + 1 + + 2] = 4 + 1 + 2 + 2 = 9
𝑃(0) 1

P0= 1/9
From equation 1.15
𝐴𝑁 𝑁
PD= 𝑃(0)
𝑁! 𝑁−𝐴

8 3 3 1 4
PD= ( + )=
3! 3−2 1 9 9

i.e. 44% of calls have delay on answer

2. When there is delay, the mean delay is:

ℎ 18
̅̅̅̅
𝑇1 = 𝑁−𝐴 = (3−2) =18 seconds
Where h is the mean holding time

The delay averaged over all time, the mean delay, ̅̅̅̅
𝑇1, is

𝑇̅ = 𝐸2,𝑁 (𝐴)𝑇1
̅̅̅̅= 18* 4/9 =8 seconds

−𝑡⁄ −30⁄
When there is delay, P(𝑇𝐷 ≥ t) = 𝑒 ̅̅̅
𝑇′ = 𝑒 18 = 18.9%

Averaged over all time,

−𝑡⁄
P(𝑇𝐷 ≥ t) = 𝐸2,𝑁 (𝐴)𝑒 ̅̅̅
𝑇′ = 18.9 *0.44 =8.3%

23
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Problem 1.14:

A message- switching center sends message on an circuit at the rate of 480 characters per
second. The average number of characters per message is 24 and the message lengths
have a negative exponential distribution. The input of messages is a Poisson process and
they are served in order of arrival.

How many messages can be handled per second if the mean delay averaged over al
messages) is not exceed 0.5 second?

For a single server the mean delay is

𝐴ℎ
𝑇̅ =
𝑁−𝐴
𝑇̅
A=ℎ+𝑇̅

Now h=24/480= 0.05 second

A=0.5/(0.05+0.5)=0.909=Ch

Where C= number of messages per second

C= 0.909/0.05=18.2

Delay tables:

𝐴 𝑘 𝐴𝑁⁄
∑𝑁
𝑘=0 𝑘! = 𝑁!
E1,N(A)

𝐴 𝑘 𝐴𝑁 𝐴𝑁
∑𝑁
𝑘=0 = 𝑁! 𝐸 −
𝑘! 1,𝑁 (𝐴) 𝑁!

Substitute the above equation in P(0)

1 𝑁 𝐴𝑁 𝐴𝑥
= [ + ∑𝑁−1
𝑥=0 𝑥! ]
𝑃(0) 𝑁!(𝑁−𝐴)

1 𝑁 𝐴𝑁 𝐴𝑁 𝐴𝑁
= [𝑁!(𝑁−𝐴) + 𝑁! 𝐸 − ]
𝑃(0) 1,𝑁 (𝐴) 𝑁!

𝑁 𝑁 𝐸1,𝑁 (𝐴)+(𝑁−𝐴)
= 𝐴𝑁! (𝑁−𝐴)𝐸1,𝑁 (𝐴)

Substituting in equation 1.15

𝐴𝑁 𝑁
PD= 𝑃(0) -------- (1.15)
𝑁! 𝑁−𝐴

24
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

𝐴𝑁 𝑁 𝑁! (𝑁−𝐴)𝐸1,𝑁 (𝐴)
𝐸2,𝑁 (𝐴)= 𝑁! 𝑁−𝐴 𝐴𝑁 𝑁 𝐸1,𝑁 (𝐴)+(𝑁−𝐴)

𝑁𝐸1,𝑁 (𝐴)
=
𝐴 𝐸1,𝑁 (𝐴)+(𝑁−𝐴)

(𝑁 − 𝐴)𝐸1,𝑁 (𝐴)
=
𝐴 𝐸1,𝑁 (𝐴) + (𝑁 − 𝐴)
𝑁
𝐸2,𝑁 (𝐴) = 𝐸1,𝑁 (𝐴)
(𝑁−𝐴)

We can calculate the, 𝐸2,𝑁 (𝐴)from 𝐸1,𝑁 (𝐴) tables.

Queues in tandem:

Queuing system are connected in tandem the delays are cumulative. If first stage has a Poissonian input
and a negative exponential distribution of holding times, The second and subsequent stages are also
Poissonian. Thus the operation can be considered as independent for calculating their delays.

The delay probability and the mean delay for the complete system are the sum of these for the
individual stages.

However, the probability distribution of such several random variables is obtained by convolution of
the separate distributions.

This computation is difficult, so it is usual to specify for each stage the probability of delay exceeding
a given value and add these probabilities to obtain a measure of the overall GOS.

This will be pessimistic estimate, because the probability of a long delay at more than one stage should
be small.

Applications of delay formulae

Delay formulas are useful in two application areas

In telephone exchange and its switching network  is a circuit switching system (or a lost call system)
In an exchange with registers when all registers are busy, incoming calls are lost

In queuing system Message (switching system or packet switching system). Here if outgoing trunks are
busy, messages or packets enter a queue until an outgoing trunk becomes free.

25
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Chapter2:
SWITCHING SYSTEMS: Introduction, Single stage networks, Gradings, Link
Systems, GOS of Linked systems.
2.1 Introduction:
In Module2 we learnt the basic prerequisite of this switching system. The basic function of an
exchange is making (switching) a connection between calling and called subscriber.
2.2 Classification of Switching Networks
The classification of switching node is based on inlets and outlets,
Single Stage switching Networks
Two Stage switching Networks
Three Stage switching Networks
Four Stage switching Networks

2.3 Single Stage networks:

Single stage switching network having M inlets and N outlets, consisting of a matrix of
crosspoints. These may, for example, be separate relays or electronic devices or the contacts of
a crossbar switch. The network could also be constructed by multiplying the banks of M
uniselectors or one level of a group of M two motion selectors, each having N outlets. In
Future system may use photonic switches, in which optoelectronics devices are used as
crosspoints to make connections between optical-fiber trunks. The below figure shows
electromechanical switches, the circle indicates the side of the switch associated with the
control mechanism (e.g. the wipers of a strowger switch or the bridge magnet of a crossbar
switch).
The matrix of crosspoints switch gives full availability; no calls are lost unless all outgoing trunks are
congested. The number of simultaneous connections that can be made is either M ( if M < N) or N (if
N<M). The switch contains MN crosspoints.
For M=N, total number of crosspoints = C=N2 ------(2.1)
Thus cost (as indicated by the number of crosspoints) increases as the square of the size of the switch.
However, efficiency decreases inversely with N.
It is therefore uneconomic to use a single stage network for large numbers of inlets and
outlets.

Fig 2.1 switch symbols

26
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

For example, a switch with 100 inlets and outlets requires 10,000 crosspoints and only 1% of these can
be in use at any time. Switches for making connection between large numbers of trunks are therefore
constructed as networks containing several stages of switches.

Matrix of cross point switches issued to make connection between N similar circuits, then each circuit
is connected to both an inlet and an outlet. Operation of the crosspoints at coordinates (j, k) to connect
inlet j to outlet k thus performs the same function as operating crosspoints (k, j) to outlet j.
Consequently, half the crosspoints are redundant and can be eliminated. This results in the triangular
crosspoints matrix shown in figure 2.2. The number of crosspoints of required is
1
C1= N (N-1) …… (2.2)
2

These are used in providing ringing tone and ringing current are sent over separate one way trunks
depending on the customer’s line is calling or being called.

Fig 2.2 triangular cross point matrix for connecting both way trunks

2.4: Gradings
The number of outgoing trunks connected to incoming trunks is known as availability.

2.4.1: Principle:

For a route switch or a concentrator it is not necessary for each incoming trunk to have access
to every outgoing trunk. It is adequate if each incoming trunk has access to a sufficient number
of trunks on each route to give the required grade of service. This is known as limited
availability. The number of outgoing trunks to which an incoming trunk can obtain connection
is called the availability and corresponds to the outlet capacity of the switches used.

Figure 2.3(a) shows 20 trunks on an outgoing route to which incoming trunks have
access by means of switches giving an availability of only ten.

27
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Figure 2.3 Twenty trunks connected in two separate groups to switches of availability 10 (a) Full
diagram (b) Grading diagram

In Figure 2.3(a), the outlets of the switches are multiplied together in two separate
groups and ten outgoing trunks are allocated to each group.

· Figure 2.4 Sixteen trunks interconnected to two groups of switches of availability

10. (a) Full diagram. (b) Grading diagram.

28
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Ifthe traffic offered to the two groups of incoming trunks is random, peak loads will seldom occur
simultaneously in the two groups. Efficiency can therefore be improved through mixing the traffic by
interconnecting the multiples of the two groups so that some of the outgoing trunks are available to
both groups of switches.

Ifthe switches search sequentially for free outlets, the later-choice outlets carry the least traffic. It is
therefore desirable to connect the later-choice trunks to both groups of selectors, as shown in Figure
2.4(a) and grading diagram shown in figure 2.4(b).

In this arrangement, the first six outlets are in two separate full-availability groups; the last four
outlets are common to both groups and carry the traffic that overflows when the first six outlets of either
group are busy.

Design of progressive gradings

In order to form a grading, the switches having access to the outgoing route are multiplied
into a number of separate groups, known as graded groups.

On early choices each group has access to individual trunks and on late choices trunks are
common, as shown in Figure 2 .4. This diagram shows a small grading for only two groups -of
switches. For larger numbers of outgoing trunks, gradings may contain four or more groups.

Figure 2.5 shows four-group gradings. Since the traffic decreases with later choices of outlet, the
number of groups connected together increases from individual connections on the early choices
through partial commons (doubles) to full commons on the late choices.

Switches hunt over the outlets sequentially from a home position.

In designing a grading to provide access to N outgoing trunks from switches having

availability k, the first step is to decide on the number of graded groups g.
If all the choices were individual trunks, we would have
N = gk .
If all the choices were full commons,
N = k.
Since the grading contains a mixture of individuals, partial commons and full commons,
then k < N < gk .
1
A reasonable choice for N is N = 2gk and traffic simulations have shown that the efficiency
of such gradings is near the optimum.
1
The number of groups is thus chosen to be: N=2gk
2N=gk
2𝑁
𝑔= -------(2.3)
𝑘

29
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Since the grading should be symmetrical, g must be an even number, so the value of g given
by equation (2.3) is rounded up to the next even integer.

It is now necessary to decide how the gk trunks entering the grading are to be interconnected
to N outgoing trunks.

For a two-group grading there is only one solution. If the number of columns of 'singles' is
s and the number of commons is c, then:
Availability = k = s + c

No. of trunks = N = 2s + c

s = N - k and c = 2k - N

If the grading has more than two groups, there is no unique solution. It is
necessary to choose from the possible solutions the best one, i.e. the grading
with the greatest traffic capacity.

 The traffic offered to adjacent outlets will not differ greatly, so they should not be
connected to very different sizes of common.
 There should thus be a smooth progression on the choices from individuals to partial
commons, from smaller partial commons to larger ones, and from partial commons to
full commons.
 The numbers of choices of each type in a group should therefore be as nearly equal as
possible.
 This is achieved by minimizing the sum of the successive differences between the
number of choices of one type and those of the type following it.

Let g have q factors: f 1 < f 2 < …. < fi ... < fq. Where

f1 = 1 and fq = g

Let ri be the number of choices having their incoming trunks connected as fi tuples
𝑞

∑ 𝑟𝑖 = 𝑘 … … … . (2.4)
𝑖=1
Now each fi tuple contains g/fi outgoing trunks.
𝑞
𝑟𝑖 𝑔
∑ = 𝑘 … … … … . (2.5)
𝑓𝑖
𝑖=1

30
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Since there are only two equations and more than two unknowns (if q > 2), there are a
number of different solutions for (r1, ………. r 4 ). These are ·round and, for each, the
sum of the successive differences, D, is given by :

D = | r 1 - r 2 | + | r 2 -r3 |+…………………+| rq-1-rq | ---(2.6)

The best grading is that having the smallest value of D.

Design of Progressive grading:
This is the most commonly used to grading methods

Step1: Find out the number of groups

g=2N/k

Step2: Find out the number of single, doubles and quadruples

Step3: Write simultaneous equations

Step4: single, doubles and quadruples tabulate the results

Sl. No Single doubles Quadruples Sum of

successive
difference

1.

2.

3.

Note: The Scheme which gives the maximum successive difference is the best
grading.

Problem 2.1

Design a grading for connecting 20 trunks to switches having ten outlets.

Solution:
The number of graded groups, given by equation (2.3) is
g = 2*N/k= 40/ 10 = 4, and the factors of g are 1, 2 and 4.
Let the number of choices having singles = s

31
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

the number of choices having doubles = d

the number of choices having quadruples = q
substituting in equations (2.4)
𝑞

∑ 𝑟𝑖 = 𝑘
𝑖=1

s + d + q = 10 -------(1)

4s + 2d + q = 20------------(2)

------------------------------------

From (2) - (1) = 3s + d = 10

Substituting in equations (2.5)

𝑞
𝑟𝑖 𝑔
∑ =𝑘
𝑓𝑖
𝑖=1

s = 1: d = 7 and q = 10 - 8 = 2

s = 2: d = 4 and q = 10 - 6 = 4

s = 3: d = 1 and q = 10 - 4 = 6

s= 4: d < 0, so this is not possible .

There are thus three possible gradings, which are shown in Figure 2.5. The sums of the successive
differences for these gradings are respectively given by :
D = | r 1 - r 2 | + | r 2 -r3 |+…………………+| rq-1-rq |

Case1) D 1 = |7-1|+|2-7| D1 = 6 + 5 = 1 1

D2 = |4 – 2|+|4-4| = 2 + 0 = 2

D 3 = |1-3| + |6 - 1| = 2 + 5 = 7

The second grading (shown in Figure 5.5(b)) is therefore the best.

32
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Figure 2.5 Four group grading for 20 trunks (availability 10)

Ifa growth in traffic makes it necessary to increase the number of trunks co nnected to a
grading, this can be done by reducing the number of commons and part ial commons and
increasing the number of individuals. Figure 2.6 shows the g rading of Figure 2.5(b)
rearranged to provide access to 25 trunks.

Figure 2.6 Grading of Figure 2.5(b) modified to accommodate 25 trunks.

Problem 2:

Find the best formation for a 6 group grading for 32 trunks with availability of 10.

The number of graded groups, given by equation (2.3) is

g = 2*N/ k = 2*32/ 10 =6.4 = 6, and the factors of g are 1, 2 3, 6.

33
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Let the number of choices having singles = s

the number of choices having doubles = d
the number of choices having triples = t
the number of choices having six lines (full commons) = fc
Note: s, d, t, or fc can never exceed 10
s + d + t + fc = 10 availability -------(1)

6s + 3d + 2 t + fc = 32 trunks ------------(2)
------------------------------------
From (2) - (1) = 5s +2 d + t = 22
Note: s, d, t, fc can never exceed 10.
S cannot be more than 4.

Successive difference is calculated by

*D = | r 1 - r 2 | + | r 2 -r3 |+…………………+| rq-1-rq |

5s +2 d + t = 22
When

S= 4, 2d+t=2, d cannot be >1 hence d=1 and t=15

S= 3, 2d+t=7, d cannot be >3
S= 2, 2d+t=10, d cannot be >5
S= 1, 2d+t=17, d cannot be >8
S= 0, 2d+t=22, d cannot be >10

SL .No No. of No. of No. of No. of 6 lines or full Successive difference

singles double triple commons (fc) is calculated *
lines d line (t)
S S+d+t+fc=10
2d+t=7 5s+2d+t
=22

Case 1 4 0 2 4 4+2+2=8

Case 2 4 1 0 5 3+1+5=9

Case 3 3 3 1 3 0+2+2=4

Case 4 3 1 3 2 2+2+1=5

Case 5 3 1 5 1 2+4+4=10

34
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Case 6 3 0 7 0 3+7+7=17

Case 7 S=2 5 2 1 3+3+1=7

Case 8 S=2 D=4 T=4 Fc=0 |2-4|+|4-4|+|0-4|=

2+0+4=6

Case 9 S=2 D=3 T=6 Fc= -1 (invalid)

Case 10 2 2 8 -2 (invalid)

Case 11 2 1 10 -3(invalid)

Case 12 2 0 12 -4(invalid)

Case 13 1 8 1 0 7+7+1=15

Case 14 S=1 d=7 T=15 Fc= -7 once invalid

don’t consider that.
(invalid)

Case 15 S=1 D=6 T=11 c= -4

(invalid)

Case 16 1 5 7 Fc= -4
(invalid)

Case 17 1 4 9 Fc= -4
(invalid)

Case 18 1 3 11 Fc= -4
(invalid)

Case 19 1 2 13 Fc= -4
(invalid)

Case 20 1 1 15 Fc= -4
(invalid)

Case 21 1 0 17 Fc= -4
(invalid)

There is one case where in successive difference is minimum. i.e., 4

35
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Figure 2.6a) Grading diagram for minimum successive difference

Where in s=3 d=3 and t=1 and fc=3

Other form of Grading

Skipped Grading

In an O'Dell grading, the partial commons are arranged as separate groups, so each is available
to only some of the incoming trunks.
For example, in Figure 2.5(b) the upper of pairs serves only the first two groups. However, the
principle of grading is based on the sharing of outgoing trunks between different sets of incoming
trunks. Efficiency can be improved if this principle can be applied to the whole of a grading instead of
only to parts of it.
This can be done by connecting non-adjacent groups, in addition to adjacent groups,
as shown in Figure 2.7. This is known as skipping.
In this grading in addition to communing adjacent groups, non-adjacent groups also are
commonly connected. This avoids upper half and lower half of the group to be separated.
Traffic is evenly distributed in both the halves.

Figure 2.7 Skipped grading

Homogeneous Grading

Progressive gradings are intended to be used with switches that hunt sequentially from a
fixed home position. However if switches do not hunt from a single position, or they select
outlets at random, there is no advantage in connecting some outlets to singles and others to

36
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

partial or full commons. The grading should then be designed to share each trunk between an
equal numbers of groups, as shown in figure 2.8 this is known as Homogeneous Grading.

Figure 2.8 Homogeneous grading.

Traffic capacity of gradings

In an ideal grading, the interconnections would ensure that each outgoing trunk carried
an identical traffic load.
Thus, if total traffic A is carried by N trunks, the occupancy of each trunk is A/ N .
It is assumed that each trunk being busy is an independent random event.
Each call has access k to trunks (where k is the availability), and the probability of all k
trunks being busy is thus:
B = (A/N )k
The number of trunks required to carry A erlangs with a GOS of B is therefore given by:
N = (AB)1/k (2.7)

This is Erlang's ideal grading formula and gives a linear relationship between the traffic
md the number of trunks required.

Practical gradings do not satisfy the conditions for Erlang's ideal grading.
However, they satisfy a linear relationship between traffic capacity A and number of
trunks for a given grade of service B.
An approximate curve of A against N can therefore be derived from Erlang's full-
availability theory for N k and extended as a straight line for N ≥ k.
From equation (2.7) this line is given by:
A = Ak + (N - k ) B1/k
k

Where Ak is the traffic carried by a full-availability group of k trunks (with GOS = B).
Figure 2.9 shows a family of curves plotted from the above modified Erlang
formula.

37
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Problem 2.3: Find the traffic capacity of the two-group grading shown in Figure 2.4
if the required grade of service is 0.01.

For k = 10 and, from Table 4.1, Ak = 4.5 E.

From equation (5.8):

A = Ak + ( N - k) B1/k

= 4.5 + (16 - 10) x 0.010·1

= 4.5 + 6 x 0.631

= 8.3 E

(Given in table 4.1 that a full-availability group of 16 trunks can handle 8.9 E with 0.01
grade of service.)
Application of Grading:
Gradings have been widely employed in step-by-step systems.

In trunk distribution frames (TDF) between the ranks of selectors provide cross-
connections in the form of gradings. Another example of the use of a grading in a link
system is in the Bell No.1 ESS system.
The su bscribers' concentrator of this system is shown in Figure 2.10 (a). The number of
crosspoints required in the primary switches is reduced by omitting them in a systematic
manner.
Each primary switch is equivalent to four groups of four-outlet selectors having access
to eight trunk s through the homogeneous grading shown in Figure 2.10(b).

38
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Figure 2.20 Two stage concentrator used in Bell No.1 ESS system. a) Arrangement of trunks,

b). Four group homogeneous grading incorporated at stage [AT & T lab copy right document]

2.5: Link Systems

General
Two stage Switching network:
In the two-stage network ( of pr evious chapte r) there is only one link between each
primary switch and each secondary switch. Thus, it may be impossible to make a
connection from a given incoming trunk to a selected outgoing trunk because the link is
already being used for another connection between that primary switch and that
secondary switch. This situation is called blocking.
It is also known as a mismatch, because free links exist but none of them can be used for
the required connection.

39
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

If connection must be made to one particular outgoing trunk the probability of

blocking is unacceptably high . For this application, it is therefore necessary to use a
network with more stages e.g. the four-stage network in order to have a choice of paths
through the network.
The two-stage network of Figure 2.15 can be used as a route switch. If it serves ten outgoing
routes with ten trunk on each route, then trunk no.1 of each route is connected to
secondary switch no. 1, trunk no. 2 is connected to switch no.2, and soon.
Thus, an incoming trunk can obtain connection to the selected outgoing route via any of
the links outgoing from its primary switch. The call is only lost if all the paths to free
outgoing trunks are blocked.
The probability of this occurring simultaneously for links is obviously much smaller
than the probability of a single link being bus Similarly, if the incoming trunks are from
several different routes, one trunk from each route is normally terminated on each primary
switch.
Step-by-step selection is unsuitable. This is because calls are lost by internal blocking
in addition to congestion of the external trunks.
It has been seen that the grade of service of a link system depends on the way it is used. We may
classify these uses as follows:
Mode 1: Connection is required to one particular free outgoing trunk. (Since conditional
selection is used, an attempt will not be made to set up this connection unless the trunk is
free.)

Mode 2: Connection is required to a particular outgoing route, but any free trunk on that
route may be used.

M ode 3: Connection may be made to any free outgoing trunk.

It will be seen from Figure 2.20 that a concentrator operates in mode 3, a route switch
operates in mode 2 and an expander operates in mode 1.
Note: The grades of service obtained are considered in Section 2.6.
Two-stage networks
If the two-stage network shown in Figure 2.11 has N incoming and N outgoing trunks and
contains primary switches having n inlets and secondary switches having n outlets.
Then no. of primary switches (g) = no. of secondary switches = no. of outlets per primary
switch = no. of inlets per secondary switch,
where
g = N/ n

40
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

The no. of crosspoints per primary switch = no. of crosspoints per secondary switch = gn =
N.
The total no. of crosspoints (C2) in the network = (no. of switches) x (crosspoints per switch)
i.e.
C2=2g N =2N2/n ---(2.9)
Since there is one link from each primary switch to each secondary switch, the num ber of links
is equal to no. of primary switches x no. of secondary switches, i .e
No. of links = g 2 = (N/n)2 (2.10)

Figure 2.11 Two stage switching network

The number of crosspoints thus varies as l/n, but the numbers of link varies as 1/n 2. If n is
made very large to reduce the number of crosspoints, there will be too few links to carry the
traffic. Let the number of links be equal to the number of incoming and outgoing trunks, a
reasonable choice, since each set of trunks carries the same total traffic.

Then g2 = N

Substituting in equation (2.10) gives

n = √N ………..(2.11)

Then the total number of crosspoints (from equation (2.9)) is

C2 = 2N 3/2 - - - - - - - - - - - - - - - (2.12)

Equation (2.11 ) can be only a guide; one should select the nearest integer to n that is a factor of
N.
Also, in practice, designers are often constrained to use switch units of fixed sizes. For example,
crossbar switches may be of sizes 10 x 10or 10 x 20.
The Bell No. 1 ESS system uses switches constructed from modules of size 8 x 8 and the
British Telecom TXE2 system uses modules of size 5 x 5.
The number of crosspoints per incoming trunk (from equation (2.12)) is 2 N 2 '

41
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

The cost per trunk therefore increases fairly slowly with the number of trunks. For large
networks, however, it becomes more economic to use network s with more than two stages.

Consider a concentrator with M incoming trunks and N outgoing trunks

(M >N ).
Let each primary switch have m inlets and each secondary switch have n outlets. Then

No. of primary switches = M/ m

No. of secondary switches = N/n
No. of crosspoints per primary switch = mN/ n
No. of crosspoints per secondary switch = nM/ m.

The total number of crosspoints is:

The number of links = no. of primary switches x no. of secondary switches
𝑀 𝑚𝑁 𝑁 𝑛𝑀
𝐶2 = +
𝑚 𝑛 𝑛 𝑚

1 1
𝑀𝑁 ⌈ + ⌉
𝑛 𝑚
The number of links = no. of primary switches x no. of secondary switches
𝑀𝑁
= 𝑚𝑛
Since the traffic capacity is limited by the number of outgoing trunks, there is little point
in providing more than this number of links, so let the number of links be N.

𝑀𝑁
=N
𝑚𝑛

And

n=M/m

Substituting in equation (2.13) from equation (2. 14)

𝑚 1
𝐶2 = 𝑀𝑁 ⌈ + ⌉
𝑀 𝑚
In order to minimize C2 treat m as if it were a continuous variable and differentiate w ith
respect to it
𝑑𝐶2 1 1
= 𝑀𝑁 ⌈ − 2 ⌉
𝑑𝑚 𝑀 𝑚
= 0 when m=√𝑀

42
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Hence, from equation (2.14):

m=𝑛 = √𝑀
Thus, the number of crosspoints is a minimum when the number of inlets per primary switch
equals the number of outlets per secondary switch.
Substituting in equation (2.13)
1 1
𝐶2 = 𝑀𝑁 ⌈ + ⌉
√𝑀 √𝑀
1
= 2𝑀 2 𝑁
Problem 2.4

Design a two-stage switching network for connecting 200 incoming trunks to 200 outgoing
trunks.

Now , √200 = 14.14.

However , n must be a factor of 200, so the nearest practicable values are n = 10 and n = 20.

Two possible networks are shown in Figure 2.12.

No of Crosspoints =2 N3/2
= 2 x (200)3/2
= 2 x 2.828 x 103
= 5656 crosspoints
= almost it contains 6000 crosspoints.
The network of Figure 2.12(a) is suitable for 20 outgoing routes, each having 10 trunks, and
that of Figure 2.12(b) is suitable for 10 outgoing routes, each having 20 trunks.
The network in Figure 2.11 has the same number of outgoing trunks as incoming trunks.
However, a concentrator has more incoming than outgoing trunks and an expander has
more outgoing than incoming trunks

43
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

.
Figure 2.12 Examples of two-stage networks. (a) For 20 outgoing routes (10 trunks on each).
(b) For 10 outgoing routes (20 trunks on each).

Problem 2.5

Design a two stage network to connect 12 incoming trunks to 9 outgoing trunks and
arrive at the optimum solution

Solution:

Incoming trunks M =12

Outgoing trunks N =9

If M>N Hence this is a concentrator

For minimum crosspoints

Aa m= n= √𝑀=√12 =3.47

But m and n should be integer and factors of 12. Hence two cases are possible

Aa m=n= 3

44
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Aa n=m=4

When m=n= 3

𝑚 1
𝐶2 = 𝑀𝑁 ⌈ + ⌉
𝑀 𝑚
3 1
𝐶2 = 12 ∗ 9 ⌈ + ⌉
12 3
= 63
When m=n= 4

𝑚 1
𝐶2 = 𝑀𝑁 ⌈ + ⌉
𝑀 𝑚
4 1
𝐶2 = 12 ∗ 9 ⌈ + ⌉
12 4
= 63

Practical use

Aa n=√𝑁 = √9 = 3
𝑀 12 12
m= = = =4
𝑛
√ √9 3

𝑚 1
𝐶2 = 𝑀𝑁 ⌈+ ⌉
𝑀 𝑚
4 1
𝐶2 = 12 ∗ 9 ⌈ + ⌉
12 4
= 63
Any of the above design can be used.

Three Stage Networks:

In the three stage networks of figure 2.13 there are
N incoming and N outgoing trunks has primary switches with n inlets and tertiary switches
with n outlets,
then

No. of primary switches (g1) = No. of tertiary switches ( g ) = N/n.

3

The secondary switches have N /n inlets and outlets.

45
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Figure 2.13 Fully interconnected three stage switching network

Let there be N number of primary-secondary links (A links) and N Number of secondary-

tertiary links (B are each N)

then the number of secondary switches is

g2= N ÷ (N/n)

= no. of outlets per primary switch = no. of inlets per tertiary switch

No. of crosspoints in primary stage = n2(N/n) = nN

No. of crosspoints in secondary stage = n (N/n)2 = N2/n
No. of crosspoints in tertiary stage = n2(N/n) - nN
and the total number of crosspoints is
𝑁2
=nN+ + nN
𝑛

𝑁2
=2nN+ 𝑛

𝑁
𝐶3 = 𝑁 (2𝑛 + )
𝑛
By differentiating equation (2.17) with respect to n and equating to zero, it can be
shown that the number of crosspoints is a minimum when

𝑛 = √𝑁⁄2

46
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

3⁄
And then 𝐶3 = 2√2𝑁 2

= √2 𝐶2
3⁄ −1⁄
=2 2 𝑁 2𝐶1

Design for M>N

If a three-stage concentrator has M incoming trunks and N outgoing trunks (M
> N), its primary switches each have m inlets and its tertiary switches each
have n outlets. Then
No. of primary switches = M/ m
No. of tertiary switches = N/ n
If there are g2 secondary switches, then

Crosspoints per primary switch = m g2

𝑀𝑁
Crosspoints per secondary switches= 𝑚𝑛

Crosspoints per tertiary switch = g2 n

The total number of crosspoints is

𝑀 𝑀𝑁 𝑁
𝐶3 = 𝑋 𝑚𝑔2 + 𝑔2 𝑋 + 𝑋𝑔2 𝑛
𝑛 𝑚𝑛 𝑛
𝑀𝑁
= 𝑔2 ⌈𝑀 + 𝑁 + ⌉
𝑚𝑛
Since M > N, let no. of A links = no. of B links = N.

𝑴 𝑵
= 𝒈𝟐 = 𝒈𝟐
𝒎 𝒏

Hence, g 2 = n and m = n M / N .

Substituting in equation (2.20):

C3 = (M + N) n + N 2/ n

Differentiating with respect to n to find a minimum gives:

𝑀 𝑁
I,e. 𝑚 = 𝑛=
√𝑀+𝑁 √𝑀+𝑁

C3 = 2N √N + M

To obtain an expander, M is exchanged with N and m with n.

47
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Problem 2.5

Design a three-stage network for connecting 100 incoming trunks to 100

outgoing trunks:

100
√ = 7.07 use n = 5 or n = 10
2

1. If n= 5, there are:
20 primary switches of size 5 x 5
5 secondary switches of size 20 x 20
20 tertiary switches of size 5 x 5

2. If n=10, there are 10 primary switches , 10 secondary switches and 10 tertiary

switches, each of size 10 x 10
𝟑⁄
No. of crosspoints = 𝑪𝟑 = 𝟐√𝟐𝑵 𝟐

𝟑⁄
𝑪𝟑 = 𝟐√𝟐𝟏𝟎𝟎 𝟐

= 2828 cross points

= 3000 Cross points

Problem 2.6

Design a three-stage network for I 00 incoming trunks and 400 outgoing trunks.
100
√100 + 400
= 4.47

400
= 11.89
√100+400

1. m = 4 or 5 ; n = 16 or 20

If m = 5, n = 20, there are:

20 primary switches of size 5 x 5

5 secondary switches of size 20 x 20
20 tertiary switches of size 5 x 2 0

2. If m = 4, n = 16, there are:

25 primary switches of size 4 x 4

48
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

4 secondary switches of size 25 x 25

25 tertiary switches of size 4 x 16 ·
C3 = 2N √N + M

= 2 x √100 + 400
= 200 x 22.3
= 4460
Say =4500
Both networks contain 4500 crosspoints.
However, the first contains more secondary Switches and will therefore cause less blocking.
In a three-stage network, the number of the selected outgoing trunk is given by outlet numbers
used on the secondary and tertiary switches.

It is not related to the ou tlet used in the primary switch, since any secondary switch may
be used for connection to a given outgoing trunk.

For each connection, two sets of links must be interrogated for the busy/free condition and
matched to choose a pair connected to the same secondary switch.
The control of a three-stage network is thus more complex than that of a two-stage one.

For this reason, electromechanical systems usually use trunkings containing a number of
separate two-stage networks in tandem. However systems having electronic central control
often employ three-stage switching networks.

A fully interconnected three-stage network (as shown in Figure 2.13) requires a ge number of
crosspoints when N is large. A reduction can be made in the number of crosspoints (at the
expense of an increase in blocking) if the secondary switches have lin k s to only some of the
primary and tertiary switches, as shown in Figure 2.14.

The secondary and tertiary switches are arranged in separate groups (frames) and are fully
interconnected only within their groups. Each primary switch has one link to each of bese
secondary-tertiary groups. (Alternatively, the primary and secondary switches may be
arranged in separate groups, to produce the mirror image of Figure 2.14.)
The number of switches is 3n2 and each has n 2 crosspoints, so the total number of crosspoints
is:
4⁄
𝐶3 = 3𝑛4 = 3𝑁 3 ----------- (2.23)
1⁄
And the number of crosspoints per incoming trunk is 3𝑁 3

49
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Figure 2.14 Partially interconnected three-stage network.

In Figure 2.13, the third stage added to the two-stage network does not increase the number of
outgoing trunks; it increases the mixture of paths available to reach them in order to reduce
blocking. The additional stage may therefore be called a mixing stage.

In Figure 2.14 the added third stage does not increase the number of paths to an outgoing
trunk; it increases the number of outgoing trunks over which the incoming traffic can be
distributed. It is therefore called a distribution stage.

In Figure 2.13 a primary switch has a link to every secondary switch, so any secondary
switch can be used for a connection to a given outgoing trunk. In Figure 2.14,
however, there are many more primary switches and each has a link to only one
secondary switch of each two-stage frame.

50
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Four-stage networks
A four-stage network can be constructed by considering a complete two-stage network as a
single switch and then forming a larger two-stage array from such switches. Figure 2.15
shows a four-stage network for 1000 incoming and 1000 outgoing trunks constructed
from two-stage networks (frames) of 100 inlets and 100 outlets using 10 x 10 switches.

Figure2.15 Four stage switching network for 1000 incoming trunks and 1000 outgoing trunks using 10
X 10 switches
It is necessary that one trunk (B link) be connected from each secondary switch of an
incoming frame to a primary switch of an outgoing frame. These trunks are connected to
switches of corresponding numbers on the two frames, thus facilitating marking of the
network. Four-stage networks of this type are used in crossbar systems.
If a four-stage network with N incoming and N outgoing trunks is constructed with
switches of size n x ·n, then N = n 3 and the total number of switches is 4n 2
Thus, the total number of crosspoints is:

C4 = 4n2 .n2

= 4 N4/3 ------------------(2.24)
The number of crosspoints per incoming trunk is 4 N1/3

51
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

It should be noted that the partially interconnected three-stage network of Figure 2.14
corresponds to the four-stage network of Figure 2.17 truncated at the A links. Adding
the fourth stage has not increased the number of trunks, although it has i n creased the
number of crosspoints by one third.

Grade of Service of Linked systems

General
A Simple theory for calculating the probability of loss in link system, due to C Y Lee concepts
explained here
In this method, assumes that trunks and links being busy constitute independent random
events.
If two links are connected in tandem, and the probability of one being busy is “p” and if the
other being busy is “q” then probabilities of each being free are (1-p) and (1-q) respectively, so
the probability of both being free is (1-a)(1-b). Therefor the probability of the path being
blocked is 1- (1-a) (1-b).
The occupancy at each stage is the total traffic carried divided by the number links at that
stage. However, if the loss is small (as it should be), little error is introduced by using the
traffic offered instead of the traffic carried.

In a practical system the assumption of independence may not be valid; because there is
usually some degree of dependence between links. This reduces the probability of blocking,
because traffic peaks at different stages coincide more often than would happen if they
were independent random events.
This overlapping of peaks tends reduce the total time during which blocking occurs.
Consequently, Lee's methods over estimates the loss probability. Nevertheless, the method
gives reasonably accurate results in most cases. It also has the merit of simplicity. For these
reasons, it is widely used.
Two-stage networks

For a two-stage network as shown in Figure 2.11, let the occupancy of the links be a and the
occupancy of the outgoing trunks be b. (If the numbers of links and trunks are equal then a = b.)
For mode I (i .e. connection to a particular outgoing trunk) only one link can be used. The
probability of this being busy is a and this is the probability of loss.

For example, to provide a GOS of B 1 = 0.01. each link a nd outgoing trunk could only carry
0.01E.

52
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

For mode 2 (i.e. connection to an outgoing route with one trunk on each secondary switch)
any free link can be used. The probability of loss using a particular link is
= 1 - probability that both link and trunk are free
= 1 - ( 1 - a)( 1 - b)

Let there are g paths available. Assuming that each being blocked is an i ndependent r a ndom
event, the probability of simultaneous blocking for all g paths is:
B2 9= [1-(1- a)(1 – b )]g

= [a + ( l – a) b]g (2.25)

"Where g is the number of secondary switches

If connection may be made to any outgoing trunk that is free (i.e. mode 3) then it possible to make
the con nection unless all the outgoing trunks are busy .Thus, if the numbers of incoming trunks,
links and outgoing trunks are equal, no calls can be lost.

Ho wever, this mode of operation is normally used with a concentrator. The number of incoming
trunks is then much larger than the number of outgoing trunks. so the grade of service is given by
𝐵3 = 𝐸1,𝑁 (𝐴)
there A is the total traffic offered to the network .
Problem 2.7

1. Find the grade of service when a total of 30 E is offered to the two-stage switching
network of Figure 2.13 and 2.14. the traffic is evenly distributed over the 10
outgoing routes,

Here n=10 and g=10

Incoming trunk=outgoing trunk=100=N
Traffic offered=30E
Switch size= 10x10

53
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

The link and trunk occupancies are a = b =A/N= 30/ 100 = 0.3 E.

For the two stage network= B = [1 - ( 1 - a)(1 - b)]n

B = [1 - ( 1 - 0.3)(1 - 0.3)]10 = 0.51 10 = 0.0012

2. Find the traffic capacity of the network if the grade of service is not to exceed
0.01.
B = [1 - ( 1 - a)(1 - b)]n
If a=b, n=10 and B= 0.01
B ≤ 0.01 = [1 - (1 - a)2]10
0.1
1 - ( l - a)2 ≤ 0.01
1
= 0.631

a ≤ 0.39
Thus offered traffic is given by ,
A= ax number of links or outgoing links or outgoing trunks or incoming trunks
A= 0.3925x100
A ≤ 39.25 E
Problem 2.8
1. Find the grade of service when a total of 30 E is offered to the two-stage switching
network. The traffic is evenly distributed over the outgoing routes. Assume the
number of trunks to be 200.
Also find the traffic capacity of the network if the grade of service is not to
exceed 0.01.
Solution:
Here n=40 and g=5
Incoming trunk=outgoing trunk=200=N
Traffic offered=30E
Let Switch size be = 40x40
Number of switches =g=5
The link and trunk occupancies are a = b =A/N= 30/ 200 = 0.15 E.

For the two stage network= B = [1 - ( 1 - a)(1 - b)]g

B = [1 - ( 1 – 0.15)(1 - 0.15)]5 = 0.51 5 = 0.001646

B = [1 - ( 1 - a)(1 - b)]g

54
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

If a=b, g=5 and B= 0.01

B ≤ 0.01 = [1 - (1 - a)2]5
0.1
1 - ( l - a)2 ≤ 0.01
1

a ≤ 0.2242E
Thus offered traffic is given by ,
A= a x number of links or outgoing links or outgoing trunks or incoming trunks
A= 0.2242x200
A ≤ 44.84 E

Three-stage networks
For a fully interconnected three-stage network (as shown in Figure 2.13)
Let Occupancy of A links be a
Occupancy of B links be b
Occupancy of outgoing trunks be c.
For mode l (i.e. connection to a particular outgoing trunk), the choice of a second switch
determines the A and B links.
Probability that both links are free = (1-a) (l -b)
Probability of blocking = 1 - (1 - a) ( 1 - b)
However, there are g 2 secondary switches.
Probability that all g 2 independent paths are simultaneously blocked is
B 1 = [1 - ( 1 -a)( 1 - b)]g2
= [ a + (1 - a)b]g2

Thus, for similar occupancies, the three-stage network provides the same GOS
connection is to individual trunks as the two-stage network does for connections group of
trunks.

For mode 2 (i.e. a connection to any free trunk in a route having one to connected to
each tertiary switch):
Probability of blocking for a particular trunk

= 1 - (1 - B1) ( l - c)
= B 1 +. (l - B1 ) c

55
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Therefore Probability of simultaneous blocking for all g 3 independent paths

is B2 = [B 1 + c ( l - B1)] g 3

Where g3 is the number of tertiary switches

Problem 2.7

1a. Design three stage network interconnecting 100 incoming trunks to 100
outgoing trunks
1b. Compare the grades of service provided by the two networks of Example
when each operates in mode l and is offered 30 E of traffic:
1c. What is the traffic capacity of each network if the required grade of service is 0.01?
Solution: (a):

100
√ = 7.07
2

use n = 5 or n = 10
1. If n= 5, there are:
20 primary switches of size 5 x 5
5 secondary switches of size 20 x 20
20 tertiary switches of size 5 x 5

If n=10, there are

10 primary switches of size 10 x 10
10 secondary switches 10 x 10 and
10 tertiary switches of size 10 x 10

b). a = b = A/N= 30/ 100 = 0.3 E

(b): B = [l - (1 - a) (1 - b)]g

i.e. g=5=n

B = [l - (1 - 0.3) (1 - 0.3)]5 = 0.51 5 = 0.035 for n=5

For network (b): B = 0.51 10 = 0.0012

B = [l - (1 - 0.3) (1 - 0.3)]10 = 0.012 for n=10

56
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

c) Solution

B = [l - (1 - a) (1 - b)]n for a=b,

[1 - (1 - a)2)5 = 0.01

1 - (1 -a) 2 = O.Ol1/5 = 0.398

a = 0.224

Total traffic capacity = A= a x N

= 100 x 0.224 = 22.4 E
For network (b): N=100, n=5, B=0.01

B = [l - (1 - a) (1 - b)]n for a=b,

[1 - (1 - a)2]10 = 0.01

1 - (1 - a)2 = 0.011/10 = 0.631

a = 0.393

Total traffic capacity = 100 x 0.393 = 39.3 E

Partially interconnected Networks GOS
For a partially interconnected three-stage network, as shown in Figure 2.14, there is only
one path between an incoming trunk and an outgoing trunk.
The probability that this is free is (1 - a)(l - b)
and the probability of blocking is 1 - ( 1 - a)( l - b).
For a connection to a trunk on an outgoing route with n trunks, each connected to a different
frame, the probability of loss using a particular trunk is
1 - (l - a) (l - b)( l - c)

But there are n such trunks available. Assuming that each being busy is an independent
random event, the probability of simultaneous blocking for all path s is
B2 = [1- (1 - a)(l - b)(l - c)]n (2.28)

Four-stage networks
For a four-stage network, as shown in Figure 2.15 let Occupancy of A links be a

57
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

Occupancy of B links be b
Occupancy of C links be c
Occupancy of outgoing trunks be d
For a connection from a given inlet on an input frame to a particular outlet on an output
frame (i.e. mode 1 ), the call may use any primary switch in the output frame. This switch is
connected by a Blink to only one secondary switch in the particular input frame. From this
switch there is only one A link to the primary switch of the given incoming trunk .
Probability of this path being free is (1 - a)( l - b)(l - c)
:. Probability of this path being blocked is 1 - ( 1 - a)(l - b) (l - c)
Probability that all g 2 independent paths are simultaneously blocked is
B 1 = (1 - (1 - a) (1 - b) (1 - c)]g2
2

Where g2 is the number of secondary switches in input frame = number of primary

switches in output frame.
For a route of n outgoing trunks:
Probability of loss for a particular trunk
= 1 - ( 1 – B1)( 1 - d)

= B 1 + (1 – B1) d

:. Probability of simultaneous blocking for all n independent paths is

B 2 = [ B 1 + d(l - B 1)]n

Design a grading for connecting 25 trunks to switches having 12 outlets.

Solution:
The number of graded groups,
g = 2*N/k= 50/ 12= 4.1, and the factors of g are 1, 2 and 4.
Let the number of choices having singles = s
the number of choices having doubles = d
the number of choices having quadruples = q
substituting in equations

58
 SAI VIDYA INSTITUTE OF TECHNOLOGY Module 3

s + d + q = 12 -------(1)

4s + 2d + q = 25------------(2)

------------------------------------

From (2) - (1) = 3s + d = 13

Substituting in equations (2.5)

s = 1: d = 10 and q = 12 - 11= 2

s = 2: d = 7 and q = 12 - 9= 3

s = 3: d = 4 and q = 12 - 7= 5

s= 4: d= 1 , and q= 12-5= 7
s=5, d<1 so this is not possible .
There are thus three possible gradings, which are shown in Figure 2.5. The sums of the successive
differences for these gradings are respectively given by :
D = | r 1 - r 2 | + | r 2 -r3 |+…………………+| rq-1-rq |

Case1) D 1 = |1-10|+|10-2| D 1 = 9+ 8 = 1 7

D2 = |2– 7|+|7-3| = 5+ 4 = 9

D 3 = |3-4| + |4 - 5| = 1 + 1 = 2

D 3 = |4-1| + |1 - 7| = 3 + 6 = 9

Best grading

having Successive Difference is least value is good grading method or allocation.

S=3,d=4 and q=5 then SD=2 (least Successive Difference)

59